Description
This paper reports the results of an experiment designed to investigate how mandatory audit firm rotation affects auditor–
client negotiations. Drawing upon process theories of negotiation, we examine the strategies used by auditors and clients
as well as the outcomes of their negotiations in alternative settings in which mandatory rotation is imposed or is not
imposed. We posit that mandatory rotation changes (1) the dynamics of the audit market by increasing the number of
clients who are in the market for a new auditor, and (2) the political costs to a client who switches auditors.
The impact of auditor rotation on auditor–client negotiation
Karl J. Wang
a,
*
, Brad M. Tuttle
b,1
a
The Patterson School of Accountancy, University of Mississippi, Oxford, MS 38677, United States
b
School of Accounting, Moore School of Business, University of South Carolina, Columbia, SC 29208, United States
Abstract
This paper reports the results of an experiment designed to investigate how mandatory audit ?rm rotation a?ects audi-
tor–client negotiations. Drawing upon process theories of negotiation, we examine the strategies used by auditors and cli-
ents as well as the outcomes of their negotiations in alternative settings in which mandatory rotation is imposed or is not
imposed. We posit that mandatory rotation changes (1) the dynamics of the audit market by increasing the number of
clients who are in the market for a new auditor, and (2) the political costs to a client who switches auditors. These changes,
in turn, alter the willingness of the auditor and the client to cooperate during negotiation. The results suggest that with
mandatory rotation auditors adopt less cooperative negotiation strategies, producing asset values that are more in line with
the auditor’s preferences than with the client’s preferences and more negotiation impasses.
Ó 2008 Elsevier Ltd. All rights reserved.
‘‘All of the technical aspects [of auditing] are
important but they can’t substitute for the pri-
mary skill, which is the art of negotiation.”
–Michael Buxbaum
The CPA Journal, 72(5), p. 80, May 2002
Introduction
This paper reports the results of an experiment
designed to investigate process di?erences in audi-
tor–client negotiation under conditions with and
without mandatory audit ?rm rotation (hereafter,
mandatory rotation). Prior experimental research
shows that client-preferred reporting by auditors is
less likely under mandatory rotation (Dopuch,
King, & Schwartz, 2001). The present study builds
on previous ?ndings by adding negotiation to the
auditor–client mix and then, using the negotiation
protocols, explores the process that produces
audited asset values. This research methodology
enables us to look inside each negotiation script to
determine the negotiation strategies used by both
parties, and to relate these strategies to the negoti-
ated outcomes.
Negotiation applies to auditing (Gibbins,
Salterio, & Webb, 2001; McNair, 1991; Nelson,
Elliott, & Tarpley, 2002; Ng & Tan, 2003) because
for many ?nancial statement accounts (1) uncer-
tainty exists about the true value to report, and (2)
di?erent incentives exist for managers and auditors
(c.f., Murnighan & Bazerman, 1990). Uncertainty
0361-3682/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aos.2008.06.003
*
Corresponding author. Tel.: +1 662 915 3980; fax: +1 662 915
7483.
E-mail addresses: [email protected](K.J. Wang), tuttle@
sc.edu (B.M. Tuttle).
1
Tel.: +1 803 777 6639.
Available online at www.sciencedirect.com
Accounting, Organizations and Society 34 (2009) 222–243
www.elsevier.com/locate/aos
implies that a range of possible and reasonable val-
ues exist so that the selection of any one particular
value is not necessarily right or necessarily wrong
ex ante. In this context, di?ering incentives can lead
client managers and auditors to prefer di?erent val-
ues within the range of possibilities. Hence, negoti-
ation in an auditor–client context is a natural
process of reconciling di?erences in incentive-
induced preferences within a range of possibilities
and not necessarily about truthful versus deceitful
reporting as has been explored in much of the exist-
ing auditing literature.
We draw upon process theories of negotiation to
predict the nature and outcome of auditor–client
negotiations. Negotiation theories developed in a
social context suggest that concerns for self and
for the other party in the relationship dominate
the negotiation process (c.f., Bazerman, Curhan,
Moore, & Valley, 2000). Extending these theories
to an economic context, Savage, Blair, and Sorensen
(1989) link concern for self to the substantive out-
come of the immediate negotiation episode and con-
cern for the other party to the economic bene?ts one
expects from a continued relationship. Applying
this model to an audit setting, the current period’s
audit outcome equates to the substantive outcome
of the immediate negotiation while the ability to
retain a client over multiple audit periods provides
continued economic bene?ts and motivates concern
for the relationship.
Mandatory rotation a?ects the auditor’s con-
cern for the relationship. The possibility of retain-
ing the same client inde?nitely provides incentives
for the auditor to prefer the same ?nancial state-
ment values as does the client in order to maintain
a good relationship with the client even when these
values con?ict with those preferred by investors
(c.f., Fisher, Frederickson, & Pe?er, 2006; Zhang,
1999). Mandatory rotation diminishes the expected
future bene?ts to the individual auditor who is
negotiating and thus alters the auditor’s concern
for the relationship from a negotiation perspec-
tive.
2
In addition, the costs of ending the relation-
ship di?er to both the auditor and the client with
mandatory rotation than without mandatory rota-
tion. Based on these theoretical considerations, we
predict the negotiation strategies that each party is
motivated to adopt. Speci?cally, we predict, and
?nd, that mandatory rotation results in the auditor
being more likely to adopt non-cooperative negoti-
ation strategies and that the negotiation is more
likely to end in impasse.
This study contributes to the auditing literature
by combining experimental economics methods
with real-time negotiation/verbal protocol analysis
methods developed in the negotiation communica-
tion literature. This combination enables us to pro-
vide a more complete image of the process that
produces the experimental outcome, the process
that is both issue and incentive driven. The adoption
of this research methodology also contributes to the
negotiation literature as it attempts to capture the
unique aspects of negotiation activity. As Davis
and Holt (1993, p. 244) point out, the rich mes-
sage-space and the real-time nature of unstructured
negotiation create a larger set of strategies than is
typically considered in traditional economic
research. Our results may stimulate negotiation
research outside the context of auditing. In addi-
tion, our study contributes to the negotiation litera-
ture also in that we ?nd support for the dual
concern theory in a highly institutionalized setting
in which rotation of the negotiation parties occurs.
The remainder of this paper proceeds as follows:
The next section provides the theoretical back-
ground and develops the hypotheses. Section
‘‘Method” outlines our research method, Section
‘‘Results” presents the results, and ‘‘Discussion”
section concludes the paper with a discussion of
the results and limitations of the study.
Literature and hypotheses
Auditor–client negotiation
A number of studies investigate issues pertinent
to auditor–client negotiations (e.g., Antle &
Nalebu?, 1991; Beattie, Brandt, & Fearnley, 2000;
Demski & Frimor, 1999; Farmer, Rittenberg, &
Trompeter, 1987; Goodwin, 2002; Nelson et al.,
2002; Ng & Tan, 2003; Trotman, Wright, & Wright,
2005; Zhang, 1999). These studies generally suggest
that auditor–client negotiations can materially a?ect
?nancial statements while identifying a number of
accounting and other contextual issues that impact
auditor–client negotiations. Other studies, which
we will brie?y review below, have looked speci?cally
2
We acknowledge that with only a limited number of large
accounting ?rms, the loss of the audit engagement may result in
an increase in other non-audit revenues. While this situation
mitigates the audit incentives to the ?rm, the bene?ts of increased
non-audit revenues may not speci?cally accrue to the individual
auditor who is negotiating.
K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243 223
at the process of auditor–client negotiation with
implications for how mandatory rotation may a?ect
this process.
Gibbins et al. (2001) develop a process model of
auditor–client negotiation. They use this model to
analyze a survey of 93 partners in large audit ?rms
and draw several conclusions related to our study.
They ?nd that auditor–client negotiation is a nor-
mal part of auditing and that audit partners view
negotiation as a part of client service in which audi-
tors help produce appropriate ?nancial statements.
They also ?nd that at the start of the negotiation,
the partners believe that a range of mutually accept-
able outcomes exist and do not typically expect the
?nal outcome to be the same as their initial position.
Within this setting, both clients and auditors hold a
mutual desire to reach agreement. Interestingly,
their data suggest that retention risk a?ects the
whole process of auditor–client negotiation. Contin-
uing this line of research, Gibbins, McCracken, and
Salterio (2005) provide responses from both audi-
tors and clients and analyze the problem solving
process used by eight matched audit partner–CFO
dyads. These studies, however, re?ect a presump-
tion that the auditor–client relationship is on-going.
Most studies of auditor–client negotiation share this
assumption and so do not examine how incentives
imbedded in an on-going relationship drive audi-
tor–client negotiations or how changes in incentives
as a result of a mandatory rotation policy impact
such negotiations.
Brown and Johnstone (2005) examine the in?u-
ence of engagement risk and auditor negotiation
experience on negotiations. Using computer simu-
lated clients, they infer the negotiation strategies
of audit managers and partners based on their open-
ing and ?nal bids. Speci?cally, they infer a contend-
ing strategy when the ?nal bid is the same as the
opening bid, a concessionary strategy when the ?nal
bid is closer to the client’s initial position than to the
auditor’s opening bid, and an integrating strategy
when ?nal bids are closer to (but not the same as)
the auditor’s opening bid. They ?nd that auditors
with less client negotiation experience use more con-
cessionary negotiation strategies when engagement
risk is high. While Brown and Johnstone investigate
negotiation strategy, they only simulate the client
therefore holding its negotiation strategy constant.
Furthermore, they only infer the negotiation strat-
egy based on the negotiated outcomes. In dynamic
negotiations, i.e., those in which both sides actively
participate, it is possible for more than one strategy
to arrive at similar outcomes depending upon the
strategy of the opponent (Pruitt & Carnevale,
1993; Pruitt & Rubin, 1986). Hence, negotiation
strategy can be ascertained only by measuring the
negotiation strategy directly from the negotiation
itself.
Several studies examining certain negotiation tac-
tics of auditors and clients are also related to our
study. Using decision case methodology, Hat?eld,
Agoglia, and Sanchez (2005), for example, ?nd that
auditors use reciprocity tactics when faced with a
non-cooperative client. Sanchez, Agoglia, and
Hat?eld (2007) ?nd that reciprocity tactics in?uence
the client into accepting more proposed audit
adjustments and feeling better about the process.
Bame-Aldred and Kida (2007) investigate auditor
and client initial negotiation positions and tactics,
and ?nd that auditors, compared to clients, are less
willing to make concessions in negotiations. These
studies, however, do not consider the economic con-
ditions in which such negotiation tactics were used.
The results of these studies may not hold when
incentives that drive auditor–client negotiations
are changed.
Mandatory rotation
Several recent studies speci?cally examine the
possible e?ects of mandatory rotation. Using
decision case methods, Hat?eld, Jackson, and
Vandervelde (2007) ?nd that proposed audit adjust-
ments are more conservative under the conditions of
either audit partner rotation or audit ?rm rotation.
However, Hat?eld et al. only look at proposed
adjustments and not at the ?nal booked adjust-
ments. Proposed adjustments are ‘‘pre-negotiation”
whereas ?nal booked adjustments are likely to be
negotiated outcomes. Daniels and Booker (2006)
?nd that the imposition of mandatory rotation
alters loan o?cers’ perceptions about auditor inde-
pendence but not about audit quality. Kaplan and
Mauldin (2008) ?nd that non-professional investors
do not have the similar perception of the e?ect of
mandatory rotation on auditor independence, con-
sistent with the GAO study (2003). In contrast,
Kim, Min, and Yi (2004) ?nd that discretionary
accruals in the quasi-mandatory rotation environ-
ment actually implemented in Korea tended to be
low, suggesting a substantial impact of mandatory
rotation on ?nancial reporting. These studies, how-
ever, do not examine the e?ect of mandatory rotation
on auditor–client interactions.
224 K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243
In a seminal study, Dopuch et al. (2001) experi-
mentally test the e?ect of mandated auditor rotation
and mandated auditor retention (in non-rotating
periods) on auditors’ willingness to issue client-
preferred reports. They ?nd that mandatory rota-
tion changes the interaction between the auditor’s
reporting decision and the client’s investing deci-
sion. Speci?cally, they ?nd that mandatory rotation
reduces the willingness of the auditor-subjects to
issue client-preferred reports and the willingness of
the client-subjects to invest (which would reduce
the risk of auditor misreporting).
Unlike the present study, Dopuch et al. do not
examine the process of an auditor and a client
directly negotiating a reporting choice. This limita-
tion restricts the generalizability of Dopuch et al.’s
?ndings with respect to two processes that are criti-
cally important to auditing: client reporting decision
and negotiation. While our study helps to extend the
generalizability of Dopuch et al., its primary contri-
bution is that it directly examines the process by
which the auditor and the client reach a reporting
decision. It provides new understanding about audi-
tor–client negotiation strategies under the condition
of mandatory rotation.
Negotiation as a social process
A primary theory of negotiation as a social pro-
cess is dual concern theory (Pruitt & Carnevale,
1993). According to this theory, individuals mod-
ify their negotiating strategy by taking into
account their individual incentives and their rela-
tionship to the other party. That is, the theory
posits that concern for one’s self and concern for
the other party in?uence the negotiation strategy.
Savage et al. (1989) extend this theory to an eco-
nomic context by both recognizing that the pri-
mary motivators in a business setting are
economic outcomes and de?ning economic out-
comes in terms of relationship.
3
They characterize
two general types of negotiations. One type
includes negotiations in which ‘‘negotiators are
motivated to establish or maintain positive rela-
tionships and willingly share the pie through
mutually bene?cial collaboration.” Hence, concern
for the other party derives not so much out of
concern for the other’s well-being but out of a
desire to maintain a mutually bene?cial economic
relationship. The other type of negotiation
involves ‘‘substantive outcomes that can bene?t
one negotiator only at the expense of the other”
(p. 38). This latter concern for the substantive out-
come motivates negotiators to ‘‘discount the rela-
tionship and claim as much of the pie as
possible.” From an auditor–client engagement
point of view in which economic forces are domi-
nant, concern for the substantive outcome of the
negotiation equates to a focus on the current per-
iod’s ?nancial statements while concern for the
other party results in a focus on maintaining the
auditor–client relationship over continued engage-
ments. Because audits can be characterized as a
series of repeated negotiations from year to year,
concern for the relationship implies that auditors
and clients attempt to ensure the continuation of
engagement even if it comes at the expense of
the current year’s negotiation. Before we describe
how mandatory rotation in?uences these concerns,
?rst, we should note that both concerns for sub-
stantive outcome and for the relationship can vary
in degree, resulting in four basic combinations
leading to four basic negotiation strategies as illus-
trated in Fig. 1.
According to Savage et al. (1989), when negotia-
tors are less concerned about the relationship, they
are more likely to adopt a non-cooperative strategy.
For instance, negotiators who have low concerns for
both the substantive outcome and the relationship
3
See Bazerman et al. (2000) for a comprehensive review of
negotiation research and Gelfand, Major, Raver, Nichii, and
O’Brian (2006) for an outline of recent developments. Up until
recently, academic work in negotiations had been ‘‘the province
of mainly economists of game theory and its applications . . .”
(Lewicki, 1997, p. 589). One limitation with this body of
knowledge, according to Greenhalgh and Chapman (1995), is
its failure to address relationships – the most common element in
real-world negotiations and a particularly important element in
auditor–client negotiations.
Unilateral Negotiation Strategies
Concern for Substantive Outcome
Concern for Relationship High Low
High
A
Integrating
Strategy
B
Obliging
Strategy
Low
C
Contending
Strategy
D
Inaction
Strategy
Fig. 1. Unilateral negotiation strategies. (Adapted from Savage
et al. (1989).)
K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243 225
tend to adopt a strategy of inaction as shown in cell
D of Fig. 1. Inaction strategies are typically associ-
ated with the goal to avoid a loss. Such strategies
may exhibit wait-and-see behavior that results in
agreement only if something viewed as a gain is
o?ered. For example, the auditor may ignore the cli-
ent’s request for negotiating a mutually agreeable
asset value to report and stick to the most conserva-
tive approach when client punishment is unlikely
(Antle & Nalebu?, 1991). Likewise, negotiators
who are unconcerned about the relationship but
who are concerned about the substantive outcome
are more likely to adopt a contending strategy as
shown in cell C of Fig. 1. Negotiators who adopt
a contending strategy actively argue for their own
positions without considering the costs to the other
party or parties. For example, the client may press
the auditor to accept an aggressive accounting treat-
ment by using, explicitly or implicitly, a dismissal
threat. Inaction and contending strategies do not
rely on cooperation to achieve agreement.
On the other hand, when negotiators are more
concerned about the relationship, they will be more
likely to adopt a cooperative strategy. For instance,
as shown in cell A of Fig. 1, negotiators who have
high concerns for both the substantive outcome and
the relationship are more likely to adopt an integrat-
ing strategy. Negotiators who adopt an integrating
strategy may seek a mutually compromising, ‘‘fair”
agreement that is somewhere between the two oppos-
ing initial positions. Alternatively, they may propose
‘‘win–win” solutions to the negotiation by introduc-
ing a new issue and/or a new solution that neither
party initially considered. For example, the auditor
and the client may both agree to concessions within
the current period or over a series of periods as sug-
gested by Dopuch et al. (2001). In like manner, nego-
tiators who have low concerns for the substantive
outcome but high concern for the relationship tend
to adopt an obliging strategy. Negotiators who adopt
an obliging strategy seek agreement to maintain the
relationshipwith less regard toits cost totheir current
payo?s. Integrating and obliging strategies rely on
cooperation to achieve agreement.
To implement the dual concern model in an audi-
tor–client negotiation setting, we identify the eco-
nomic incentives associated with each party’s
actions concerning ?nancial reporting in the rota-
tion and non-rotation regimes. That is, we de?ne
how the auditor’s and the client’s concerns change
as a function of the economic conditions that varies
with whether mandatory rotation is imposed. In
absolute terms, auditor–client relationships and spe-
ci?c audit outcomes are very important to both cli-
ents and auditors. In relationship to the various
negotiation contexts manipulated in the present
study, however, the two dimensions di?er in relative
importance. Hence, we frame the discussion in
terms of one negotiation setting relative to other
negotiation settings. We ?rst outline the e?ect of
mandatory rotation on the negotiation strategies
of the auditor followed by a discussion of its e?ects
on the client’s negotiation strategies.
Auditor negotiation strategies
We ?rst assume that whether or not mandatory
rotation is imposed, the auditor is guaranteed the
audit fee for the current year and that the current
fee amount is una?ected by any negotiation regard-
ing the ?nancial statements. We further assume a
healthy client without risk of insolvency so that
the risk of litigation is relatively low. While these
assumptions do not describe every client, we believe
that they represent the general audit environment.
Together, these assumptions work to lower an audi-
tor’s concern for the immediate substantive
outcome.
On the other hand, the current audit environ-
ment contains incentives for auditors to be con-
cerned about their long-term relationship with the
client. The Government Accounting O?ce (GAO,
2003) reports that the average auditor tenure for
Fortune 1000 companies is 22 years. Likewise, a
study published by Fulcrum Financial Group
(2003) reports that 10% of the companies in their
sample have had the same auditor for 50 or more
years, with the average tenure of this group being
75 years. Although the opportunity costs to a large
audit ?rm from losing a single client may be mini-
mal, individual partners and managers (who are
doing the negotiation) face very high personal
opportunity costs related to their reputations and
to possible lost long-term pro?ts for their o?ces if
negotiations fail (Defond, Francis, & Carcello,
2005; Francis, Maydew, & Sparks, 1999; Reynolds
& Francis, 2000). Under these economic incentives,
auditors may be willing to concede some items in
the short term in order to preserve the long-term
relationship with their clients, as suggested by some
proponents of mandatory rotation (Benson, 2002;
Imho?, 2003; Wolf, Tackett, & Claypool, 1999).
Hence, we argue that the auditor’s concern for the
substantive outcome is relatively low compared to
226 K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243
their concern for the on-going relationship.
4
Given
that the concern for this relationship is higher with-
out mandatory rotation and lower with mandatory
rotation, as we will explain below, we propose
‘‘Obliging” as the dominating strategy for auditors
without mandatory rotation and ‘‘Inaction” as the
dominating strategy for auditors with mandatory
rotation, as displayed in Fig. 2.
Mandatory rotation potentially a?ects an audi-
tor’s incentives during a negotiation in three ways:
First, in the ?nal audit year before the rotation
occurs, it removes much of the possible reputation
e?ects to the auditor arising from an auditor
switch.
5
This freedom from reputation e?ects
accrues to the individual auditor both from inside
and outside of the ?rms, reducing the auditor’s con-
cern for the relationship with the client. Second, in
non-?nal years, mandatory rotation reduces the
prospect of long-term rents arising from subsequent
audits beyond the limit. As a result, the auditor’s
concern for the relationship is reduced, which is
the argument typically employed by mandatory
rotation proponents (e.g., Gietzmann & Sen, 2002;
Imho?, 2003; SEC, 1994; US Senate, 1976, 1977).
Third, mandatory rotation alters the market for cli-
ents by increasing the number of clients looking for
new auditors in any given year, thus dramatically
reducing the auditor’s concern for the relationship.
6
In summary, we argue that without mandatory
rotation, auditors have relatively low concern for
the substantive outcome and relatively high concern
for the relationship compared to auditors with man-
datory rotation. This in turn leads to the greater
adoption of obliging strategies during negotiation.
We also argue that, with mandatory rotation, audi-
tors have relatively low concerns for both substan-
tive outcome and the relationship, which leads to
a greater use of inaction strategies. These arguments
are formalized in the following hypotheses regard-
ing auditor strategies in auditor–client negotiations:
H1 : Mandatory rotation a?ects the negotiation
strategies used by auditors.
H1a : Auditors will use an obliging negotiation
strategy more frequently without mandatory
rotation than with mandatory rotation.
H1b : Auditors will use an inaction negotiation
strategy more frequently with mandatory
rotation than without mandatory rotation.
Client negotiation strategies
When negotiating with the auditor, the client
management (client) has strong reasons to be con-
cerned with the substantive outcome. Unlike the
auditor, who receives a guaranteed audit fee, cur-
rent period ?nancial statements can in?uence the
client’s immediate compensation. Hence, the client
has strong incentives to in?uence the outcome of
the current negotiation as set forth in the earnings
management literature (c.f., Defond & Jiambalvo,
1993). Imposing auditor term limits has no impact
on these incentives as they are endogenous to the
auditor–client relationship. We, therefore, assert
that client concern for substance in the negotiated
outcome is relatively high regardless of mandatory
rotation.
In terms of concern for the relationship, without
mandatory rotation the client generally bene?ts
from a good relationship by avoiding switching
costs associated with training a new auditor. These
switching costs are thought to be substantial
(GAO, 1996) thus leading to a high concern for
the relationship. In addition, clients that switch
auditors can incur political costs associated with
market perceptions of opinion shopping.
7
We assert
4
If one alternatively assumes that the auditor’s concern about
the substantive outcome of the current negotiation is high, the
auditors’ speci?c negotiation strategy changes from contending to
inaction. As pointed out by a reviewer, neither strategy involves
cooperation and the gist of our predictions holds.
5
We implement our experiment in an anonymous setting, thus
precluding reputation e?ects while allowing us to focus on the
other e?ects of mandatory rotation. Analysis of the negotiation
scripts revealed that no one ever made an e?ort to identify who
their opponent was.
6
Mandatory rotation increasing the negotiating auditor’s
ability to replace a lost client is a critical distinction between an
audit environment with mandatory rotation and the current audit
environment. Assuming that the average auditor tenure is 22
years or more, it follows that less than 5% of clients are in the
market for new auditors in any given year. Hence, at the
individual o?ce level, a partner involved in the negotiation
knows that the likelihood of replacing the client is very small.
However, should the auditor term be limited to just four years as
has been advocated by some (US Senate, 2002), one out of four
clients every year will be looking for a new auditor. As a result,
the prospects of replacing a lost client when mandatory rotation
is imposed are improved. The fact that audit ?rm rotation
dramatically alters the availability of replacement clients is a key
di?erence between audit ?rm rotation and audit partner rotation
that has not been considered in previous studies.
7
There may be, however, some limited circumstances, such as a
switch from a non-Big-Four auditor to a Big-Four auditor, in
which this may not be the case.
K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243 227
that market perceptions of opinion shopping by the
client are more likely in the early years of an audi-
tor–client relationship and are more likely for more
frequent switches. These political costs are consid-
ered to be substantial and to be present without
mandatory rotation and in the non-?nal years with
mandatory rotation. This suggests that in all these
circumstances, the client’s concern for the relation-
ship will be high.
In the ?nal year under mandatory rotation, a
change in auditor after the current period is auto-
matic so the client incurs the same switching costs
(i.e., those related to e?ciency) regardless of the
outcome of negotiations. However, the automatic
change in auditor prevents investors from attribut-
ing this event to opinion shopping without addi-
tional information. Hence, the client is unlikely to
experience political costs in this year regardless of
the reason for the change. This suggests that in
the ?nal year under mandatory rotation, the client’s
concern for maintaining a long-term relationship
with the auditor is lower than in non-?nal years,
as illustrated in Fig. 2.
As illustrated in Fig. 1, clients who have high
concerns for the substantive outcome and high con-
cerns for the relationship are more likely to adopt
an integrating strategy of negotiation. This occurs
when mandatory rotation is not imposed and in
non-?nal years when it is imposed. On the other
hand, in the ?nal year under mandatory rotation,
with high concern for the substantive outcome but
low concern for the relationship, clients are more
likely to adopt a contending strategy. These predic-
tions are summarized in the following hypotheses:
H2 : Mandatory rotation a?ects the negotiation
strategies used by clients.
H2a : Clients will use an integrating negotiation
strategy more frequently without mandatory
rotationthanwithmandatoryrotation, particu-
larly inthe ?nal year before rotationtakes place.
H2b : Clients will use a contending negotiation
strategy more frequently with mandatory
rotation, particularly in the ?nal year before
rotation takes place, than without mandatory
rotation.
Relative Concern for Relationship and Substantive Outcome
And the Tendency to Adopt Specific Negotiation Strategies
a
Mandatory Rotation
No Mandatory Rotation Non-Final Periods Final Periods
Auditor Client Auditor Client Auditor Client
Panel A: Unilateral Strategies
Concern for
Relationship
Higher Higher Lower Higher Lower Lower
Concern for
Substantive
Outcome
Lower Higher Lower Higher Lower Higher
Unilateral
Negotiation
Strategy
More
Obliging
More
Integrating
More
Inaction
More
Integrating
More
Inaction
More
Contending
Cooperation Higher Higher Lower Higher Lower Lower
Panel B: Interactive Strategies
Interactive
Negotiation
Strategy
Trusting
Collaboration
Trusting
Collaboration
Principled
Collaboration
& Soft
Competition
Focused
Subordination
Responsive
Avoidance
& Firm
Competition
Principled
Collaboration
& Soft
Competition
Fig. 2. Relative concern for relationship and substantive outcome and the tendency to adopt speci?c negotiation strategies. (In absolute
terms, auditor–client relationships and speci?c audit outcomes are very important to both clients and auditors. In relationship to the
various negotiation contexts manipulated in the present study, however, the two dimensions di?er in relative importance thus leading to
the prediction that the use of the various negotiation strategies will di?er across conditions.)
228 K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243
Negotiation outcomes
Particularly in the ?nal year under mandatory
rotation, the auditor appears to wield complete
power over the client. The economic prediction in
this case is that no negotiation will take place and
that auditors will simply dictate their positions to
their clients. However, the literature on ultimatum
and dictator games, in which one party has com-
plete power over the other (c.f., Davis & Holt,
1993; Kagel & Roth, 1995), suggests that people
do not strictly adhere to economic predictions but
often follow social norms related to fair play even
when given complete power over another individ-
ual. Hence, we make comparative predictions, rec-
ognizing the relative economic power rather than
the absolute economic power that dictates equilib-
rium predictions without considering the impact of
social factors. Notice from Fig. 2 that both auditors
and clients tend to use cooperative strategies (oblig-
ing and integrating) without mandatory rotation
but tend to use non-cooperative strategies (inaction
and contending) when mandatory rotation is
imposed. To the extent that cooperation leads to
successful negotiation, the following hypotheses
are proposed:
H3a : Mandatory rotation leads to less cooperation
by auditors in negotiation.
H3b : Mandatory rotation leads to less cooperation
by clients in negotiation.
H4 : More negotiations will end in impasse with
mandatory rotation than without mandatory
rotation.
When an agreement is reached, the party that
uses an obliging strategy is more likely to receive a
lower payo? and the party that uses an inaction
strategy is more likely to receive a higher payo? than
the party that uses other strategies (Pruitt &
Carnevale, 1993). This ?nding, when applied to
Fig. 2, leads us to predict
H5 : Negotiation outcomes will be closer to audi-
tors’ preferences with mandatory rotation and
closer to the clients’ preferences without man-
datory rotation.
Negotiation theory recognizes that individual
negotiators may tend towards certain unilateral
strategies but that these strategies can change over
time after interacting with the opponent as shown
in Panel B of Fig. 2. For instance, Savage et al.
(1989, Exhibit 3) predict that when one negotiator
unilaterally adopts an obliging strategy while the
opponent adopts an integrating strategy, they both
may possibly evolve into using a strategy of collab-
oration based on trust. The key point from Fig. 2 is
that the various interactive strategies that are perti-
nent to this study retain their cooperative or non-
cooperative nature from their root unilateral strate-
gies (Greenhalgh & Chapman, 1995). For instance,
all low cooperation unilateral strategies depicted in
Fig. 2 may evolve into interactive strategies that
emphasize competition. The high cooperation uni-
lateral strategies shown in Fig. 2 may evolve into
trusting collaboration or subordination. Because
people evolve into using integrative strategies that
are either cooperative or non-cooperative in a man-
ner that re?ects their cooperation under their initial
unilateral strategies, predictions about negotiated
outcomes are likely to be the same. Furthermore,
to design a study that investigates the use of integra-
tive strategy requires complex negotiations over an
extended time. For these reasons, we only examine
unilateral strategies.
Method
Participants and experimental task
Fifty-four graduate business students partici-
pated in a laboratory negotiation experiment con-
ducted over six sessions with nine participants in
each session. The typical participant was 26.2 years
old with 3.7 years of work experience. About 44% of
the participants were women.
Within each session, four participants were ran-
domly assigned the role of manager (i.e., client)
and ?ve participants were assigned the role of veri-
?er (i.e., auditor). Together in pairs, one manager
and one veri?er negotiated a value to be reported
for an asset. Hence, each session consisted of four
pairs of negotiators with one veri?er in each period
sitting out and thus earning zero fees for that
period.
In all negotiations, managers and veri?ers shared
the same set of information indicating the range and
associated probabilities of the actual value of the
asset, as shown in Fig. 3. The actual value of the
asset for each negotiation period was pre-deter-
mined using a computer-generated random process
consistent with the probabilities shown in Fig. 3.
The actual value of the asset was revealed to both
K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243 229
managers and veri?ers after each negotiation period
was over.
Experimental manipulations
Negotiation, including auditor–client negotia-
tion, is a process by which two or more parties hav-
ing di?erent preferences arrive at a joint decision
(Murnighan & Bazerman, 1990). Thus, understand-
ing preferences is key to understanding negotiation.
Preference di?erences can arise because of informa-
tion asymmetry between the client and the auditor
(Gibbins et al., 2001) or because the client and the
auditor operate under di?erent incentives (Zhang,
1999). Because our predictions are based on changes
to client and auditor incentives as a?ected by
mandatory rotation, we hold information to be
symmetric and allow only incentives to a?ect the
negotiators’ preferences.
8
The experiment manipulates mandatory rota-
tion at two levels (mandatory rotation imposed,
yes versus no) between subjects. Participants in
the mandatory rotation (MR) condition were told
that ‘‘Each manager–veri?er pair continues to
negotiate with each other for up to three periods
as long as the veri?er continues to accept the man-
ager’s submitted values.” In this condition, when
the same negotiating pair reached the third period
together, the computer screen automatically
showed a message reminding both that it was their
?nal period. Participants in the no mandatory
rotation (NoMR) condition were told that ‘‘Each
manager and veri?er pair continues to negotiate
with each other for an unlimited number of periods
as long as the veri?er continues to accept the man-
ager’s submitted values.”
The consequences of a negotiation impasse to
the veri?er are determined by the market for cli-
ents. Recall that the negotiations took place in a
laboratory and in groups of nine participants con-
sisting of four managers and ?ve veri?ers. The
addition of one more veri?er than manager cre-
ated a competitive market such that if a negotia-
tion fails and the relationship is terminated, the
manager is guaranteed a new veri?er but the ver-
i?er may or may not be reassigned to a new man-
ager as described below.
9
In the NoMR condition,
if a negotiation fails, the veri?er can only be reas-
signed to a new manager if another manager–ver-
i?er pair also fails to reach an agreement. The
data suggest that after an impasse in the NoMR
condition, the veri?er often had to wait three or
more periods before being reassigned to another
manager. In the MR condition, if a negotiation
fails, the veri?er has the same chance to be reas-
signed to a new manager as in the NoMR condi-
tion plus the chance to be reassigned to any
manager that had reached the three-period term
limit. The data suggest that in the MR condition,
after a failed negotiation the veri?er was immedi-
ately reassigned to another manager in two of
three instances and was reassigned in the next fol-
lowing period in the remaining instances.
The consequence of an impasse to a manager is a
40% reduction in earnings for the period, represent-
ing the political costs incurred by the manager. This
cost is imposed in all periods in the NoMR condi-
tion and the MR condition except in a third (?nal)
consecutive period in the MR condition with the
same veri?er in which political costs do not apply.
Procedures and negotiation rules
After arriving in the computer lab and receiving
their role assignments and a computer, the partici-
pants began by reading written instructions. The
instructions were pre-tested to clearly communicate
8
Some prior studies manipulate negotiator preferences by
providing less information to auditors than to clients. Arguably,
di?erences in opinion can occur with full information on both
sides. The present study assumes that the auditor performs
su?cient tests, etc. in order to arrive at the same probability
distribution as the client thus controlling for information
asymmetry.
9
The decision to have one out of ?ve auditors without a client
(i.e., 20%) is constrained by the availability of subjects and a
desire not to make the auditor’s incentive to avoid an impasse too
strong (such as two out of ?ve).
Asset Value Distribution
Value ($)
Probability
of being
true
Probability
of being
too high
500 5% 95%
450 5% 90%
400 10% 80%
350 10% 70%
300 15% 55%
250 15% 40%
200 15% 25%
150 10% 15%
100 10% 5%
50 5% 0%
Fig. 3. Asset value distribution.
230 K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243
the incentives and mechanics of the experiment. The
participants then completed two practice periods.
Practice periods did not a?ect earnings and are
excluded from hypotheses testing. Following other
negotiation studies (Roth, 1995), we set each period
to last two and a half minutes. After the two prac-
tice periods, 18 negotiation periods were conducted
for each session. In order to prevent end of game
antics, the participants were not told how many
periods the experiment would last. At the conclu-
sion, cash earnings were determined and the partic-
ipants paid and excused. Each session lasted
approximately 1 hour and 15 minutes.
To begin each period of negotiation, a manager
submitted an asset value on the computer screen
depicted in Fig. 4. After receiving the manager’s
proposal, the veri?er could choose to accept the
submitted asset value or to negotiate with the man-
ager on the computer screen depicted in Fig. 5.
Negotiations proceeded by typing messages back
and forth in a text box and by the manager propos-
ing new asset values on the computer screen. The
Fig. 4. Manager screen.
Fig. 5. Veri?er screen.
K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243 231
veri?er could accept the value submitted by the
manager at any time before the period ended. Once
the veri?er accepted a value submitted by the man-
ager, the period of negotiation ended and the actual
value of the asset as well as each party’s earnings for
that period were shown in a message box on the
computer screen.
If the veri?er did not accept a value submitted by
the manager before the negotiation time ran out, the
period ended in impasse. The manager then was
randomly paired with a new veri?er in the following
period and the veri?er was randomly reassigned to
another manager, if available, within the constraint
that veri?ers were not reassigned to the same man-
ager they had just left.
Payo?s
The experimental instructions showed payo?
trees for both managers and veri?ers correspond-
ing to the participant’s respective experimental
condition as shown in Fig. 6. Payo?s for manag-
ers are ?rst described followed by payo? for
veri?ers.
Panel A: No Mandatory Rotation
Managers’ Payoff:
Accepted value
Yes
Verifier’s Acceptance
No
60% Actual value
Verifiers’ Payoff:
$100 - 50% (Accepted Value -
Actual value)
Yes
Accepted value > Actual value
No
$100
Panel B: Mandatory Rotation
Managers’ Payoff:
Accepted value
Yes
Verifier’s Acceptance
No
60% Actual value (in any first and second
periods with the same verifier), or 100%
Actual value (in any third period with the
same verifier)
Verifiers’ Payoff:
$100 - 50% (Accepted Value -
Actual value)
Yes
Accepted value > Actual value
No
$100
Fig. 6. Payo?s.
232 K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243
A manager’s earnings in each period were deter-
mined by whether the veri?er accepted an asset
value submitted by the manager and, in the MR
condition, whether it was the third (?nal) period
with the same veri?er. If the veri?er accepted a sub-
mitted asset value, the manager earned that value
regardless of the asset’s actual value. In the MR
condition, an impasse resulted in the manager earn-
ing the asset’s actual value less the deduction for
political costs (of 40%) in all but the ?nal period.
In the NoMR condition, an impasse resulted in
the manager earning the asset’s actual value less
the deduction for political costs in all periods.
Hence, managers had an incentive to negotiate
higher asset values but not so high as to cause an
impasse.
A veri?er’s earnings in each period were deter-
mined by whether the veri?er accepted a value sub-
mitted by the manager and whether the accepted
value was greater than the actual value of the asset.
Re?ecting the assumption that auditors are typically
guaranteed their fee, in all cases, the veri?er earned a
?at service fee of 100 experimental dollars. However,
if the veri?er accepted an asset value greater than the
actual value, a penalty of 50% of the di?erence
between the accepted value and the asset’s actual
value was imposed. Furthermore, veri?ers earned
nothing when not assigned a manager. Hence, in
all but the ?nal period in the MR condition, veri?ers
had an incentive to negotiate lower asset values to
avoid a penalty but not so low as to cause an impasse
and therefore run the risk of not being assigned to a
manager on the subsequent round.
Because experimental earnings di?ered in
amount for managers and veri?ers, it was necessary
to create a cash reward system to make both parties’
payo?s comparable. This was done by making the
cash payment for each participant determined by
the rank of his or her cumulative earnings within
the group of participants who played the same role
(manager or veri?er) in each session as shown in
Table 1. This cash payment schedule was known
to all participants in advance.
Data coding
Two independent coders, rigorously trained and
blind to all hypotheses, coded all of the negotiation
protocols from the six sessions of the experiment as
evidence of integrating, obliging, inaction, or con-
tending strategies. The coding scheme was based
on Pruitt and Rubin (1986) and Rahim (1983).
Sillars’s (1986) coding manual was used as a techni-
cal reference in identifying di?erent types of mes-
sages relating to the four negotiation strategies.
10
Following prior negotiation studies, the unit of text
for coding was the semantic turn or ‘‘thought turn”
(Canary, Cupach, & Messman, 1995), in which a
change in semantic content indicates a di?erent
‘‘message” requiring coding (Sillars, 1986). Accord-
ingly, the coders were instructed to code each mes-
sage within the context of its dialogue to one of
the four negotiation strategies. This method is con-
sistent with the communication literature that
focuses on coding dialogue instead of sentences
(Donohue, 1992). The inter-coder reliability, as
measured by Cohen’s K is 0.72, which Bakeman
and Gottman (1997) classify as ‘‘very good” within
this context. The disagreements between the two
coders were reconciled by a third person who also
was blind to all hypotheses. Following a rather con-
servative approach that eliminated most messages
coded by only one coder, the reconciliation pro-
duced 3867 messages for analysis.
Results
Preliminary analyses
Veri?ers who failed to reach agreement with a
manager in no mandatory rotation condition typi-
cally experienced two or three o? periods prior to
being reassigned. Out of 60 possible periods of play
(20 periods per session times 3 sessions), at least one
negotiator pair failed to reach agreement in 26 of
those periods.
Table 2 presents the count of messages coded to
each of the four strategies in no mandatory rotation
condition and mandatory rotation condition. Both
veri?ers and managers submitted more messages in
mandatory rotation condition than in no manda-
tory rotation condition. Also, integrating messages
Table 1
Cash payment schedule
Managers Veri?ers
Highest cumulative earnings $30 $30
Second highest $24 $25
Third highest $17 $20
Fourth highest $10 $15
Fifth highest N/A $10
10
In the negotiation literature, the use of a negotiation strategy
is typically measured by the frequency of verbal messages that
indicate the use of a certain strategy.
K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243 233
dominate the negotiations, consistent with a general
tendency of both parties to cooperate. However, the
table shows that contending messages are relatively
frequent and that approximately one-fourth of all
messages are non-cooperative.
Dependent measures
Analyses are performed throughout using two
dependent measures. The ?rst dependent measure
is the number of messages coded to each of the four
negotiation strategies as described above. In this
case, message counts are aggregated across all peri-
ods by each negotiator (i.e., veri?er or manager).
The second dependent measure is the negotiation
strategy used by an individual negotiator within a
speci?c period as re?ected by that strategy with
the most messages during the period. That is, the
strategy with the majority of messages within a sin-
gle negotiation script is deemed to be the dominant
strategy used by the negotiator during that period.
Hence, this measure is calculated separately for each
period of negotiation and for each negotiator. On
average, the count of messages coded to the domi-
nant strategy in each period divided by the total
number of messages in the script for the period is
0.810. This relatively high proportion lends cre-
dence to our measure of strategy use.
Table 3 reports the results of analysis showing
that the use of a particular strategy is more likely
to dominate other strategies (v
2
value = 10.3362,
p < 0.0013) in mandatory rotation condition
(89.7% of periods) than in no mandatory rotation
condition (81.8% of periods). Table 4 shows the
mean number of messages coded to each strategy
in the average script. Table 4 suggests that, when
averaged over the entire sample, most messages rely
on cooperation.
Tests of hypotheses
Hypothesis 1 predicts that mandatory rotation
will a?ect the negotiation strategies of auditors. Spe-
ci?cally, H1a suggests that auditors will send oblig-
ing messages more frequently in the absence of
mandatory rotation, and H1b suggests that auditors
will send inaction messages more frequently when
mandatory rotation is imposed. Our ?rst test of
H1 consists of a general comparison across manda-
tory and no mandatory rotation conditions of the
percentage of messages coded to each of the four
negotiation strategies. Results of this analysis are
Table 2
Negotiation messages
No mandatory rotation Mandatory rotation (MR)
Non-?nal periods Final periods All periods MR condition
Count Percent Count Percent Count Percent Count Percent
Managers
Cooperative strategies
Obliging 73 9.1 58 7.0 13 5.8 71 6.72
Integrating 548 68.4 593 71.2 137 61.4* 730 69.1
Total cooperative strategies 77.5 78.2 67.3** 75.8
Uncooperative strategies
Contending 174 21.7 144 17.3* 42 18.8 186 17.6*
Inaction 6 0.8 38 4.6*** 31 13.9*** 69 6.5***
Total managers 801 100.0 833 100.0 223 100.0 1056 100.0
Veri?ers
Cooperative strategies
Obliging 168 20.8 27 2.8*** 4 1.6*** 31 2.6***
Integrating 494 61.2 639 67.0* 130 52.2* 769 63.9
Total cooperative strategies 82.0 69.8*** 53.8*** 66.5***
Uncooperative strategies
Contending 119 14.8 217 22.8*** 47 18.9 264 22.0***
Inaction 26 3.2 71 7.4*** 68 27.3*** 139 11.6***
Total veri?ers 807 100.0 954 100.0 249 100.0 1203 100.0
*p < 0.05, **p < 0.01, ***p < 0.001. All tests compare percents in a mandatory rotation condition against the percent in the no mandatory
condition.
234 K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243
Table 4
Mean number of messages coded to each strategy within each script
a
No mandatory rotation Mandatory rotation (MR)
Non-?nal periods Final periods All periods MR condition
Panel A: Managers
Obliging 0.291 0.346 0.237 0.325
Integrating 3.196 3.712 3.342 3.639
Cooperative messages 3.487 4.059 3.579 3.963
Contending 0.930 0.797 0.921 0.822
Inaction 0.025 0.203 0.711 0.304
Uncooperative messages 0.955 1.000 1.632 1.125
Panel B: Veri?ers
Obliging 0.789 0.125 0.103 0.120
Integrating 2.783 4.000 2.795 3.754
Cooperative messages 3.572 4.125 2.897 3.874
Contending 0.507 1.303 0.974 1.236
Inaction 0.132 0.401 1.256 0.576
Uncooperative messages 0.639 1.704 2.230 1.811
Panel C: Combined
Obliging 0.535 0.236 0.169 0.225
Integrating 2.994 3.856 3.065 3.696
Cooperative messages 3.529 4.092 3.234 3.919
Contending 0.723 1.049 0.948 1.029
Inaction 0.077 0.302 0.987 0.440
Uncooperative messages 0.800 1.351 1.935 1.469
Total messages 4.329 5.443 5.168 5.387
a
A script consists of the series of messages sent by one of the negotiators during a single period.
Table 3
Strategy use as evidenced by negotiation scripts
No mandatory rotation Mandatory rotation (MR)
Non-?nal periods Final periods All periods MR condition
Panel A: Managers
Total script count 185 166 44 210
Scripts w/strategy 158 153 38 191
Percent 85.4% 92.2%
*
86.4% 91.0%
Panel B: Veri?ers
Total script count 194 167 49 216
Scripts w/strategy 152 152 39 191
Percent 78.4% 91.0%
***
79.6% 88.4%
**
Panel C: Combined
Total script count 379 333 93 426
Scripts w/strategy 310 305 77 382
Percent 81.8% 91.6%
***
82.8% 89.7%
**
*
p < 0.05,
**
p < 0.01,
***
p < 0.001. All tests compare a mandatory rotation condition against the no mandatory condition. For example,
the overall hypothesis of no association between mandatory rotation condition and the use of a negotiating strategy by the individual
participants is rejected (likelihood ratio v
2
value = 10.3362, p < 0.0013) in that a strategy is more likely to be employed in periods when
mandatory rotation is imposed (89.7% of periods) compared to when it is not imposed (81.8% of periods).
Note: In each of six di?erent sessions, 18 negotiation periods were conducted (i.e., 20 periods À 2 practice periods) with four pairs of
negotiators in each period rendering 432 individual negotiations (6  18  4) having 864 possible scripts (432 periods  2 negotiators in
each period). In this case a script consists of the series of messages sent by one of the negotiators during a single period. Of these, no
message was identi?ed with a strategy by the coding process for 59 (6.8%) negotiation scripts leaving 805 scripts available for analysis (i.e.,
379 + 426). The strategy is the strategy associated with the greatest number of messages within a script.
K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243 235
summarized in Table 2. The table shows a higher
percentage (p < 0.001) of obliging messages by veri-
?ers in no mandatory rotation condition (20.8%) as
compared to mandatory rotation condition (2.6%).
The table also shows a higher percentage
(p < 0.001) of inaction messages by veri?ers in man-
datory rotation condition (11.6%) as compared to
no mandatory rotation condition (3.2%). These
results become stronger when looking at only the
?nal period in mandatory rotation condition in
which obliging messages by veri?ers are only 1.6%
of the total but inaction messages by veri?ers are
27.3%. Thus, both H1a and H1b are supported.
These results suggest that mandatory rotation
reduces the relative importance to the auditor of
maintaining a relationship with the client. They also
suggest that the experiment design appropri-
ately re?ects our assumptions about the audit
environment.
We further examine H1 by comparing overall
strategy use by veri?ers across mandatory rotation
and no mandatory rotation conditions as shown in
Table 5. Consistent with H1a, Panel B of Table 5
shows that veri?ers are more likely (p < 0.0001) to
use an obliging strategy in no mandatory rotation
condition (14.5%) compared to mandatory rotation
condition (1.1%). This result holds for both the peri-
ods leading up to the penultimate negotiation per-
iod (1.3%) as well as the ?nal period (0.0%). In
contrast and consistent with H1b, Panel B also
shows that overall, veri?ers are more likely
(p = 0.0009) to use a strategy of inaction in manda-
tory rotation condition (7.3%) compared to no man-
datory rotation condition (0.7%). This tendency
becomes especially strong in the ?nal year of nego-
tiation in mandatory rotation condition in which
23.1% of the veri?ers employ a strategy of inaction.
Hypothesis 2 predicts that mandatory rotation
will change client negotiation strategies, particularly
in the ?nal periods of negotiation before rotation
takes place (i.e., the third consecutive period with
the same auditor). Table 2 shows that in mandatory
rotation condition, manager messages re?ect inte-
grating strategies less frequently as predicted by
H2a, but only in the ?nal period (61.4% versus
68.4%). Contrary to H2b, however, managers send
fewer contending messages in mandatory rotation
condition (17.6%) than in no mandatory condition
(21.7%, p < 0.05). Similar patterns emerge for dom-
inant strategies by managers in the ?nal period in
mandatory rotation condition as shown in Panel
A of Table 5. There are similar (but not signi?cant)
decreases in integrating messages (76.3% versus
80.4%) and increases in contending messages
(13.2% versus 11.4%) by managers in mandatory
rotation condition. However, both Tables 2 and 5
show distinct increases in inaction messages by
managers in the ?nal period in mandatory rotation
Table 5
Dominant strategies
Dominant strategy No mandatory rotation Mandatory rotation (MR)
Non-?nal periods Final periods All periods MR condition
Panel A: Managers
Total script count 158 153 38 191
Obliging 8.2% 1.3%
**
5.3% 2.1%
**
Integrating 80.4% 86.9% 76.3% 84.8%
Total cooperative 88.6% 88.2% 81.6% 86.9%
Contending 11.4% 10.5% 13.2% 11.0%
Inaction 0.0% 1.3% 5.3%
**
2.1%
Total uncooperative 11.4% 11.8% 18.5% 13.1%
Panel B: Veri?ers
Total script count 152 152 39 191
Obliging 14.5% 1.3%
***
0.0%
***
1.1%
***
Integrating 77.6% 83.6% 69.2% 80.6%
Total cooperative 92.1% 84.9% 69.2%
***
81.7%
**
Contending 7.2% 11.8% 7.7% 11.0%
Inaction 0.7% 3.3% 23.1%
***
7.3%
***
Total uncooperative 7.9% 15.1% 30.8% 18.3%
*
p < 0.05,
**
p < 0.01,
***
p < 0.001. All signi?cance tests compare a mandatory rotation condition against no mandatory rotation condition.
236 K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243
condition (13.9% and 5.3%) compared to no manda-
tory rotation condition (0.7% and 0.0%, p < 0.001).
The data suggest that although mandatory rotation
did a?ect the negotiation strategies used by manag-
ers as predicted by H2, the speci?c predictions of
H2a and H2b are not supported.
Hypothesis 3a predicts less cooperation by audi-
tors in negotiation with mandatory rotation than
without mandatory rotation. Table 2 shows a signif-
icant decrease (p < 0.001) in cooperative messages
by veri?ers in mandatory rotation condition
(66.5%) compared to no mandatory rotation condi-
tion (82.0%). Only 53.8% of messages by veri?ers in
the ?nal period in mandatory rotation condition are
cooperative. Likewise, Table 5 shows a signi?cant
decrease in cooperative messages by veri?ers in
mandatory rotation condition (p = 0.0042). In no
mandatory rotation condition, 92.1% of veri?er
messages are cooperative, while in mandatory rota-
tion condition, only 81.7% of veri?er messages are
cooperative. Even fewer, i.e., 69.2% of veri?er mes-
sages, in ?nal periods are cooperative. Hence, the
data support H3a that mandatory rotation leads
to less cooperation by auditors in negotiation.
Hypothesis 3b predicts less cooperation by clients
in negotiation with mandatory rotation than with-
out mandatory rotation. Table 2 shows a signi?cant
decrease (p < 0.001) in cooperative messages by
mangers, but only in the ?nal periods in mandatory
rotation condition (67.3%) compared to no manda-
tory rotation (77.5%). Table 5 shows no signi?cant
di?erences in cooperative messages by managers
across the conditions. One explanation for the lack
of results for managers can be found in Panel A of
Table 5: managers tend to use less obliging strate-
gies (2.1% versus 8.2%) but more integrating strate-
gies (84.8% versus 80.4%) in mandatory rotation
condition than in no mandatory rotation condition.
These two changes tend to cancel each other while
representing a shift from a more extreme form of
cooperation (i.e., obliging) to a less extreme form
of cooperation (i.e., integrating). Hence, although
the data do not support H3b that mandatory rota-
tion leads less cooperation by clients in negotiation,
some evidence suggests that client cooperation may
not be as strong under mandatory rotation.
Hypothesis 4 predicts that mandatory rotation
will lower agreement rate in auditor–client negotia-
tion. Consistent with this prediction, Table 6 shows
that the agreement rate in mandatory rotation con-
dition (63.7%) is signi?cantly lower (p < 0.001) than
in no mandatory rotation condition (88.7%). Agree-
ment rates drop to an even lower 44.7% in ?nal
negotiation periods in mandatory rotation condi-
tion. These ?ndings strongly support H4.
In H5 we predict that when negotiations result in
an agreement, asset values will be lower with man-
datory rotation than without mandatory rotation.
11
Tests were performed using three di?erent units of
analysis (negotiating pair, veri?er, and manager).
The results are summarized in Table 7. The mean
Table 6
Negotiation outcomes
Treatment Agreement Impasse v
2
p-Value
Count Percent Count Percent
Panel A
All periods in MR condition 130 63.7 74 36.3
No mandatory rotation 181 88.7 23 11.3
Di?erence À25.0 25.0 35.18
This paper reports the results of an experiment designed to investigate how mandatory audit firm rotation affects auditor–
client negotiations. Drawing upon process theories of negotiation, we examine the strategies used by auditors and clients
as well as the outcomes of their negotiations in alternative settings in which mandatory rotation is imposed or is not
imposed. We posit that mandatory rotation changes (1) the dynamics of the audit market by increasing the number of
clients who are in the market for a new auditor, and (2) the political costs to a client who switches auditors.
The impact of auditor rotation on auditor–client negotiation
Karl J. Wang
a,
*
, Brad M. Tuttle
b,1
a
The Patterson School of Accountancy, University of Mississippi, Oxford, MS 38677, United States
b
School of Accounting, Moore School of Business, University of South Carolina, Columbia, SC 29208, United States
Abstract
This paper reports the results of an experiment designed to investigate how mandatory audit ?rm rotation a?ects audi-
tor–client negotiations. Drawing upon process theories of negotiation, we examine the strategies used by auditors and cli-
ents as well as the outcomes of their negotiations in alternative settings in which mandatory rotation is imposed or is not
imposed. We posit that mandatory rotation changes (1) the dynamics of the audit market by increasing the number of
clients who are in the market for a new auditor, and (2) the political costs to a client who switches auditors. These changes,
in turn, alter the willingness of the auditor and the client to cooperate during negotiation. The results suggest that with
mandatory rotation auditors adopt less cooperative negotiation strategies, producing asset values that are more in line with
the auditor’s preferences than with the client’s preferences and more negotiation impasses.
Ó 2008 Elsevier Ltd. All rights reserved.
‘‘All of the technical aspects [of auditing] are
important but they can’t substitute for the pri-
mary skill, which is the art of negotiation.”
–Michael Buxbaum
The CPA Journal, 72(5), p. 80, May 2002
Introduction
This paper reports the results of an experiment
designed to investigate process di?erences in audi-
tor–client negotiation under conditions with and
without mandatory audit ?rm rotation (hereafter,
mandatory rotation). Prior experimental research
shows that client-preferred reporting by auditors is
less likely under mandatory rotation (Dopuch,
King, & Schwartz, 2001). The present study builds
on previous ?ndings by adding negotiation to the
auditor–client mix and then, using the negotiation
protocols, explores the process that produces
audited asset values. This research methodology
enables us to look inside each negotiation script to
determine the negotiation strategies used by both
parties, and to relate these strategies to the negoti-
ated outcomes.
Negotiation applies to auditing (Gibbins,
Salterio, & Webb, 2001; McNair, 1991; Nelson,
Elliott, & Tarpley, 2002; Ng & Tan, 2003) because
for many ?nancial statement accounts (1) uncer-
tainty exists about the true value to report, and (2)
di?erent incentives exist for managers and auditors
(c.f., Murnighan & Bazerman, 1990). Uncertainty
0361-3682/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aos.2008.06.003
*
Corresponding author. Tel.: +1 662 915 3980; fax: +1 662 915
7483.
E-mail addresses: [email protected](K.J. Wang), tuttle@
sc.edu (B.M. Tuttle).
1
Tel.: +1 803 777 6639.
Available online at www.sciencedirect.com
Accounting, Organizations and Society 34 (2009) 222–243
www.elsevier.com/locate/aos
implies that a range of possible and reasonable val-
ues exist so that the selection of any one particular
value is not necessarily right or necessarily wrong
ex ante. In this context, di?ering incentives can lead
client managers and auditors to prefer di?erent val-
ues within the range of possibilities. Hence, negoti-
ation in an auditor–client context is a natural
process of reconciling di?erences in incentive-
induced preferences within a range of possibilities
and not necessarily about truthful versus deceitful
reporting as has been explored in much of the exist-
ing auditing literature.
We draw upon process theories of negotiation to
predict the nature and outcome of auditor–client
negotiations. Negotiation theories developed in a
social context suggest that concerns for self and
for the other party in the relationship dominate
the negotiation process (c.f., Bazerman, Curhan,
Moore, & Valley, 2000). Extending these theories
to an economic context, Savage, Blair, and Sorensen
(1989) link concern for self to the substantive out-
come of the immediate negotiation episode and con-
cern for the other party to the economic bene?ts one
expects from a continued relationship. Applying
this model to an audit setting, the current period’s
audit outcome equates to the substantive outcome
of the immediate negotiation while the ability to
retain a client over multiple audit periods provides
continued economic bene?ts and motivates concern
for the relationship.
Mandatory rotation a?ects the auditor’s con-
cern for the relationship. The possibility of retain-
ing the same client inde?nitely provides incentives
for the auditor to prefer the same ?nancial state-
ment values as does the client in order to maintain
a good relationship with the client even when these
values con?ict with those preferred by investors
(c.f., Fisher, Frederickson, & Pe?er, 2006; Zhang,
1999). Mandatory rotation diminishes the expected
future bene?ts to the individual auditor who is
negotiating and thus alters the auditor’s concern
for the relationship from a negotiation perspec-
tive.
2
In addition, the costs of ending the relation-
ship di?er to both the auditor and the client with
mandatory rotation than without mandatory rota-
tion. Based on these theoretical considerations, we
predict the negotiation strategies that each party is
motivated to adopt. Speci?cally, we predict, and
?nd, that mandatory rotation results in the auditor
being more likely to adopt non-cooperative negoti-
ation strategies and that the negotiation is more
likely to end in impasse.
This study contributes to the auditing literature
by combining experimental economics methods
with real-time negotiation/verbal protocol analysis
methods developed in the negotiation communica-
tion literature. This combination enables us to pro-
vide a more complete image of the process that
produces the experimental outcome, the process
that is both issue and incentive driven. The adoption
of this research methodology also contributes to the
negotiation literature as it attempts to capture the
unique aspects of negotiation activity. As Davis
and Holt (1993, p. 244) point out, the rich mes-
sage-space and the real-time nature of unstructured
negotiation create a larger set of strategies than is
typically considered in traditional economic
research. Our results may stimulate negotiation
research outside the context of auditing. In addi-
tion, our study contributes to the negotiation litera-
ture also in that we ?nd support for the dual
concern theory in a highly institutionalized setting
in which rotation of the negotiation parties occurs.
The remainder of this paper proceeds as follows:
The next section provides the theoretical back-
ground and develops the hypotheses. Section
‘‘Method” outlines our research method, Section
‘‘Results” presents the results, and ‘‘Discussion”
section concludes the paper with a discussion of
the results and limitations of the study.
Literature and hypotheses
Auditor–client negotiation
A number of studies investigate issues pertinent
to auditor–client negotiations (e.g., Antle &
Nalebu?, 1991; Beattie, Brandt, & Fearnley, 2000;
Demski & Frimor, 1999; Farmer, Rittenberg, &
Trompeter, 1987; Goodwin, 2002; Nelson et al.,
2002; Ng & Tan, 2003; Trotman, Wright, & Wright,
2005; Zhang, 1999). These studies generally suggest
that auditor–client negotiations can materially a?ect
?nancial statements while identifying a number of
accounting and other contextual issues that impact
auditor–client negotiations. Other studies, which
we will brie?y review below, have looked speci?cally
2
We acknowledge that with only a limited number of large
accounting ?rms, the loss of the audit engagement may result in
an increase in other non-audit revenues. While this situation
mitigates the audit incentives to the ?rm, the bene?ts of increased
non-audit revenues may not speci?cally accrue to the individual
auditor who is negotiating.
K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243 223
at the process of auditor–client negotiation with
implications for how mandatory rotation may a?ect
this process.
Gibbins et al. (2001) develop a process model of
auditor–client negotiation. They use this model to
analyze a survey of 93 partners in large audit ?rms
and draw several conclusions related to our study.
They ?nd that auditor–client negotiation is a nor-
mal part of auditing and that audit partners view
negotiation as a part of client service in which audi-
tors help produce appropriate ?nancial statements.
They also ?nd that at the start of the negotiation,
the partners believe that a range of mutually accept-
able outcomes exist and do not typically expect the
?nal outcome to be the same as their initial position.
Within this setting, both clients and auditors hold a
mutual desire to reach agreement. Interestingly,
their data suggest that retention risk a?ects the
whole process of auditor–client negotiation. Contin-
uing this line of research, Gibbins, McCracken, and
Salterio (2005) provide responses from both audi-
tors and clients and analyze the problem solving
process used by eight matched audit partner–CFO
dyads. These studies, however, re?ect a presump-
tion that the auditor–client relationship is on-going.
Most studies of auditor–client negotiation share this
assumption and so do not examine how incentives
imbedded in an on-going relationship drive audi-
tor–client negotiations or how changes in incentives
as a result of a mandatory rotation policy impact
such negotiations.
Brown and Johnstone (2005) examine the in?u-
ence of engagement risk and auditor negotiation
experience on negotiations. Using computer simu-
lated clients, they infer the negotiation strategies
of audit managers and partners based on their open-
ing and ?nal bids. Speci?cally, they infer a contend-
ing strategy when the ?nal bid is the same as the
opening bid, a concessionary strategy when the ?nal
bid is closer to the client’s initial position than to the
auditor’s opening bid, and an integrating strategy
when ?nal bids are closer to (but not the same as)
the auditor’s opening bid. They ?nd that auditors
with less client negotiation experience use more con-
cessionary negotiation strategies when engagement
risk is high. While Brown and Johnstone investigate
negotiation strategy, they only simulate the client
therefore holding its negotiation strategy constant.
Furthermore, they only infer the negotiation strat-
egy based on the negotiated outcomes. In dynamic
negotiations, i.e., those in which both sides actively
participate, it is possible for more than one strategy
to arrive at similar outcomes depending upon the
strategy of the opponent (Pruitt & Carnevale,
1993; Pruitt & Rubin, 1986). Hence, negotiation
strategy can be ascertained only by measuring the
negotiation strategy directly from the negotiation
itself.
Several studies examining certain negotiation tac-
tics of auditors and clients are also related to our
study. Using decision case methodology, Hat?eld,
Agoglia, and Sanchez (2005), for example, ?nd that
auditors use reciprocity tactics when faced with a
non-cooperative client. Sanchez, Agoglia, and
Hat?eld (2007) ?nd that reciprocity tactics in?uence
the client into accepting more proposed audit
adjustments and feeling better about the process.
Bame-Aldred and Kida (2007) investigate auditor
and client initial negotiation positions and tactics,
and ?nd that auditors, compared to clients, are less
willing to make concessions in negotiations. These
studies, however, do not consider the economic con-
ditions in which such negotiation tactics were used.
The results of these studies may not hold when
incentives that drive auditor–client negotiations
are changed.
Mandatory rotation
Several recent studies speci?cally examine the
possible e?ects of mandatory rotation. Using
decision case methods, Hat?eld, Jackson, and
Vandervelde (2007) ?nd that proposed audit adjust-
ments are more conservative under the conditions of
either audit partner rotation or audit ?rm rotation.
However, Hat?eld et al. only look at proposed
adjustments and not at the ?nal booked adjust-
ments. Proposed adjustments are ‘‘pre-negotiation”
whereas ?nal booked adjustments are likely to be
negotiated outcomes. Daniels and Booker (2006)
?nd that the imposition of mandatory rotation
alters loan o?cers’ perceptions about auditor inde-
pendence but not about audit quality. Kaplan and
Mauldin (2008) ?nd that non-professional investors
do not have the similar perception of the e?ect of
mandatory rotation on auditor independence, con-
sistent with the GAO study (2003). In contrast,
Kim, Min, and Yi (2004) ?nd that discretionary
accruals in the quasi-mandatory rotation environ-
ment actually implemented in Korea tended to be
low, suggesting a substantial impact of mandatory
rotation on ?nancial reporting. These studies, how-
ever, do not examine the e?ect of mandatory rotation
on auditor–client interactions.
224 K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243
In a seminal study, Dopuch et al. (2001) experi-
mentally test the e?ect of mandated auditor rotation
and mandated auditor retention (in non-rotating
periods) on auditors’ willingness to issue client-
preferred reports. They ?nd that mandatory rota-
tion changes the interaction between the auditor’s
reporting decision and the client’s investing deci-
sion. Speci?cally, they ?nd that mandatory rotation
reduces the willingness of the auditor-subjects to
issue client-preferred reports and the willingness of
the client-subjects to invest (which would reduce
the risk of auditor misreporting).
Unlike the present study, Dopuch et al. do not
examine the process of an auditor and a client
directly negotiating a reporting choice. This limita-
tion restricts the generalizability of Dopuch et al.’s
?ndings with respect to two processes that are criti-
cally important to auditing: client reporting decision
and negotiation. While our study helps to extend the
generalizability of Dopuch et al., its primary contri-
bution is that it directly examines the process by
which the auditor and the client reach a reporting
decision. It provides new understanding about audi-
tor–client negotiation strategies under the condition
of mandatory rotation.
Negotiation as a social process
A primary theory of negotiation as a social pro-
cess is dual concern theory (Pruitt & Carnevale,
1993). According to this theory, individuals mod-
ify their negotiating strategy by taking into
account their individual incentives and their rela-
tionship to the other party. That is, the theory
posits that concern for one’s self and concern for
the other party in?uence the negotiation strategy.
Savage et al. (1989) extend this theory to an eco-
nomic context by both recognizing that the pri-
mary motivators in a business setting are
economic outcomes and de?ning economic out-
comes in terms of relationship.
3
They characterize
two general types of negotiations. One type
includes negotiations in which ‘‘negotiators are
motivated to establish or maintain positive rela-
tionships and willingly share the pie through
mutually bene?cial collaboration.” Hence, concern
for the other party derives not so much out of
concern for the other’s well-being but out of a
desire to maintain a mutually bene?cial economic
relationship. The other type of negotiation
involves ‘‘substantive outcomes that can bene?t
one negotiator only at the expense of the other”
(p. 38). This latter concern for the substantive out-
come motivates negotiators to ‘‘discount the rela-
tionship and claim as much of the pie as
possible.” From an auditor–client engagement
point of view in which economic forces are domi-
nant, concern for the substantive outcome of the
negotiation equates to a focus on the current per-
iod’s ?nancial statements while concern for the
other party results in a focus on maintaining the
auditor–client relationship over continued engage-
ments. Because audits can be characterized as a
series of repeated negotiations from year to year,
concern for the relationship implies that auditors
and clients attempt to ensure the continuation of
engagement even if it comes at the expense of
the current year’s negotiation. Before we describe
how mandatory rotation in?uences these concerns,
?rst, we should note that both concerns for sub-
stantive outcome and for the relationship can vary
in degree, resulting in four basic combinations
leading to four basic negotiation strategies as illus-
trated in Fig. 1.
According to Savage et al. (1989), when negotia-
tors are less concerned about the relationship, they
are more likely to adopt a non-cooperative strategy.
For instance, negotiators who have low concerns for
both the substantive outcome and the relationship
3
See Bazerman et al. (2000) for a comprehensive review of
negotiation research and Gelfand, Major, Raver, Nichii, and
O’Brian (2006) for an outline of recent developments. Up until
recently, academic work in negotiations had been ‘‘the province
of mainly economists of game theory and its applications . . .”
(Lewicki, 1997, p. 589). One limitation with this body of
knowledge, according to Greenhalgh and Chapman (1995), is
its failure to address relationships – the most common element in
real-world negotiations and a particularly important element in
auditor–client negotiations.
Unilateral Negotiation Strategies
Concern for Substantive Outcome
Concern for Relationship High Low
High
A
Integrating
Strategy
B
Obliging
Strategy
Low
C
Contending
Strategy
D
Inaction
Strategy
Fig. 1. Unilateral negotiation strategies. (Adapted from Savage
et al. (1989).)
K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243 225
tend to adopt a strategy of inaction as shown in cell
D of Fig. 1. Inaction strategies are typically associ-
ated with the goal to avoid a loss. Such strategies
may exhibit wait-and-see behavior that results in
agreement only if something viewed as a gain is
o?ered. For example, the auditor may ignore the cli-
ent’s request for negotiating a mutually agreeable
asset value to report and stick to the most conserva-
tive approach when client punishment is unlikely
(Antle & Nalebu?, 1991). Likewise, negotiators
who are unconcerned about the relationship but
who are concerned about the substantive outcome
are more likely to adopt a contending strategy as
shown in cell C of Fig. 1. Negotiators who adopt
a contending strategy actively argue for their own
positions without considering the costs to the other
party or parties. For example, the client may press
the auditor to accept an aggressive accounting treat-
ment by using, explicitly or implicitly, a dismissal
threat. Inaction and contending strategies do not
rely on cooperation to achieve agreement.
On the other hand, when negotiators are more
concerned about the relationship, they will be more
likely to adopt a cooperative strategy. For instance,
as shown in cell A of Fig. 1, negotiators who have
high concerns for both the substantive outcome and
the relationship are more likely to adopt an integrat-
ing strategy. Negotiators who adopt an integrating
strategy may seek a mutually compromising, ‘‘fair”
agreement that is somewhere between the two oppos-
ing initial positions. Alternatively, they may propose
‘‘win–win” solutions to the negotiation by introduc-
ing a new issue and/or a new solution that neither
party initially considered. For example, the auditor
and the client may both agree to concessions within
the current period or over a series of periods as sug-
gested by Dopuch et al. (2001). In like manner, nego-
tiators who have low concerns for the substantive
outcome but high concern for the relationship tend
to adopt an obliging strategy. Negotiators who adopt
an obliging strategy seek agreement to maintain the
relationshipwith less regard toits cost totheir current
payo?s. Integrating and obliging strategies rely on
cooperation to achieve agreement.
To implement the dual concern model in an audi-
tor–client negotiation setting, we identify the eco-
nomic incentives associated with each party’s
actions concerning ?nancial reporting in the rota-
tion and non-rotation regimes. That is, we de?ne
how the auditor’s and the client’s concerns change
as a function of the economic conditions that varies
with whether mandatory rotation is imposed. In
absolute terms, auditor–client relationships and spe-
ci?c audit outcomes are very important to both cli-
ents and auditors. In relationship to the various
negotiation contexts manipulated in the present
study, however, the two dimensions di?er in relative
importance. Hence, we frame the discussion in
terms of one negotiation setting relative to other
negotiation settings. We ?rst outline the e?ect of
mandatory rotation on the negotiation strategies
of the auditor followed by a discussion of its e?ects
on the client’s negotiation strategies.
Auditor negotiation strategies
We ?rst assume that whether or not mandatory
rotation is imposed, the auditor is guaranteed the
audit fee for the current year and that the current
fee amount is una?ected by any negotiation regard-
ing the ?nancial statements. We further assume a
healthy client without risk of insolvency so that
the risk of litigation is relatively low. While these
assumptions do not describe every client, we believe
that they represent the general audit environment.
Together, these assumptions work to lower an audi-
tor’s concern for the immediate substantive
outcome.
On the other hand, the current audit environ-
ment contains incentives for auditors to be con-
cerned about their long-term relationship with the
client. The Government Accounting O?ce (GAO,
2003) reports that the average auditor tenure for
Fortune 1000 companies is 22 years. Likewise, a
study published by Fulcrum Financial Group
(2003) reports that 10% of the companies in their
sample have had the same auditor for 50 or more
years, with the average tenure of this group being
75 years. Although the opportunity costs to a large
audit ?rm from losing a single client may be mini-
mal, individual partners and managers (who are
doing the negotiation) face very high personal
opportunity costs related to their reputations and
to possible lost long-term pro?ts for their o?ces if
negotiations fail (Defond, Francis, & Carcello,
2005; Francis, Maydew, & Sparks, 1999; Reynolds
& Francis, 2000). Under these economic incentives,
auditors may be willing to concede some items in
the short term in order to preserve the long-term
relationship with their clients, as suggested by some
proponents of mandatory rotation (Benson, 2002;
Imho?, 2003; Wolf, Tackett, & Claypool, 1999).
Hence, we argue that the auditor’s concern for the
substantive outcome is relatively low compared to
226 K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243
their concern for the on-going relationship.
4
Given
that the concern for this relationship is higher with-
out mandatory rotation and lower with mandatory
rotation, as we will explain below, we propose
‘‘Obliging” as the dominating strategy for auditors
without mandatory rotation and ‘‘Inaction” as the
dominating strategy for auditors with mandatory
rotation, as displayed in Fig. 2.
Mandatory rotation potentially a?ects an audi-
tor’s incentives during a negotiation in three ways:
First, in the ?nal audit year before the rotation
occurs, it removes much of the possible reputation
e?ects to the auditor arising from an auditor
switch.
5
This freedom from reputation e?ects
accrues to the individual auditor both from inside
and outside of the ?rms, reducing the auditor’s con-
cern for the relationship with the client. Second, in
non-?nal years, mandatory rotation reduces the
prospect of long-term rents arising from subsequent
audits beyond the limit. As a result, the auditor’s
concern for the relationship is reduced, which is
the argument typically employed by mandatory
rotation proponents (e.g., Gietzmann & Sen, 2002;
Imho?, 2003; SEC, 1994; US Senate, 1976, 1977).
Third, mandatory rotation alters the market for cli-
ents by increasing the number of clients looking for
new auditors in any given year, thus dramatically
reducing the auditor’s concern for the relationship.
6
In summary, we argue that without mandatory
rotation, auditors have relatively low concern for
the substantive outcome and relatively high concern
for the relationship compared to auditors with man-
datory rotation. This in turn leads to the greater
adoption of obliging strategies during negotiation.
We also argue that, with mandatory rotation, audi-
tors have relatively low concerns for both substan-
tive outcome and the relationship, which leads to
a greater use of inaction strategies. These arguments
are formalized in the following hypotheses regard-
ing auditor strategies in auditor–client negotiations:
H1 : Mandatory rotation a?ects the negotiation
strategies used by auditors.
H1a : Auditors will use an obliging negotiation
strategy more frequently without mandatory
rotation than with mandatory rotation.
H1b : Auditors will use an inaction negotiation
strategy more frequently with mandatory
rotation than without mandatory rotation.
Client negotiation strategies
When negotiating with the auditor, the client
management (client) has strong reasons to be con-
cerned with the substantive outcome. Unlike the
auditor, who receives a guaranteed audit fee, cur-
rent period ?nancial statements can in?uence the
client’s immediate compensation. Hence, the client
has strong incentives to in?uence the outcome of
the current negotiation as set forth in the earnings
management literature (c.f., Defond & Jiambalvo,
1993). Imposing auditor term limits has no impact
on these incentives as they are endogenous to the
auditor–client relationship. We, therefore, assert
that client concern for substance in the negotiated
outcome is relatively high regardless of mandatory
rotation.
In terms of concern for the relationship, without
mandatory rotation the client generally bene?ts
from a good relationship by avoiding switching
costs associated with training a new auditor. These
switching costs are thought to be substantial
(GAO, 1996) thus leading to a high concern for
the relationship. In addition, clients that switch
auditors can incur political costs associated with
market perceptions of opinion shopping.
7
We assert
4
If one alternatively assumes that the auditor’s concern about
the substantive outcome of the current negotiation is high, the
auditors’ speci?c negotiation strategy changes from contending to
inaction. As pointed out by a reviewer, neither strategy involves
cooperation and the gist of our predictions holds.
5
We implement our experiment in an anonymous setting, thus
precluding reputation e?ects while allowing us to focus on the
other e?ects of mandatory rotation. Analysis of the negotiation
scripts revealed that no one ever made an e?ort to identify who
their opponent was.
6
Mandatory rotation increasing the negotiating auditor’s
ability to replace a lost client is a critical distinction between an
audit environment with mandatory rotation and the current audit
environment. Assuming that the average auditor tenure is 22
years or more, it follows that less than 5% of clients are in the
market for new auditors in any given year. Hence, at the
individual o?ce level, a partner involved in the negotiation
knows that the likelihood of replacing the client is very small.
However, should the auditor term be limited to just four years as
has been advocated by some (US Senate, 2002), one out of four
clients every year will be looking for a new auditor. As a result,
the prospects of replacing a lost client when mandatory rotation
is imposed are improved. The fact that audit ?rm rotation
dramatically alters the availability of replacement clients is a key
di?erence between audit ?rm rotation and audit partner rotation
that has not been considered in previous studies.
7
There may be, however, some limited circumstances, such as a
switch from a non-Big-Four auditor to a Big-Four auditor, in
which this may not be the case.
K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243 227
that market perceptions of opinion shopping by the
client are more likely in the early years of an audi-
tor–client relationship and are more likely for more
frequent switches. These political costs are consid-
ered to be substantial and to be present without
mandatory rotation and in the non-?nal years with
mandatory rotation. This suggests that in all these
circumstances, the client’s concern for the relation-
ship will be high.
In the ?nal year under mandatory rotation, a
change in auditor after the current period is auto-
matic so the client incurs the same switching costs
(i.e., those related to e?ciency) regardless of the
outcome of negotiations. However, the automatic
change in auditor prevents investors from attribut-
ing this event to opinion shopping without addi-
tional information. Hence, the client is unlikely to
experience political costs in this year regardless of
the reason for the change. This suggests that in
the ?nal year under mandatory rotation, the client’s
concern for maintaining a long-term relationship
with the auditor is lower than in non-?nal years,
as illustrated in Fig. 2.
As illustrated in Fig. 1, clients who have high
concerns for the substantive outcome and high con-
cerns for the relationship are more likely to adopt
an integrating strategy of negotiation. This occurs
when mandatory rotation is not imposed and in
non-?nal years when it is imposed. On the other
hand, in the ?nal year under mandatory rotation,
with high concern for the substantive outcome but
low concern for the relationship, clients are more
likely to adopt a contending strategy. These predic-
tions are summarized in the following hypotheses:
H2 : Mandatory rotation a?ects the negotiation
strategies used by clients.
H2a : Clients will use an integrating negotiation
strategy more frequently without mandatory
rotationthanwithmandatoryrotation, particu-
larly inthe ?nal year before rotationtakes place.
H2b : Clients will use a contending negotiation
strategy more frequently with mandatory
rotation, particularly in the ?nal year before
rotation takes place, than without mandatory
rotation.
Relative Concern for Relationship and Substantive Outcome
And the Tendency to Adopt Specific Negotiation Strategies
a
Mandatory Rotation
No Mandatory Rotation Non-Final Periods Final Periods
Auditor Client Auditor Client Auditor Client
Panel A: Unilateral Strategies
Concern for
Relationship
Higher Higher Lower Higher Lower Lower
Concern for
Substantive
Outcome
Lower Higher Lower Higher Lower Higher
Unilateral
Negotiation
Strategy
More
Obliging
More
Integrating
More
Inaction
More
Integrating
More
Inaction
More
Contending
Cooperation Higher Higher Lower Higher Lower Lower
Panel B: Interactive Strategies
Interactive
Negotiation
Strategy
Trusting
Collaboration
Trusting
Collaboration
Principled
Collaboration
& Soft
Competition
Focused
Subordination
Responsive
Avoidance
& Firm
Competition
Principled
Collaboration
& Soft
Competition
Fig. 2. Relative concern for relationship and substantive outcome and the tendency to adopt speci?c negotiation strategies. (In absolute
terms, auditor–client relationships and speci?c audit outcomes are very important to both clients and auditors. In relationship to the
various negotiation contexts manipulated in the present study, however, the two dimensions di?er in relative importance thus leading to
the prediction that the use of the various negotiation strategies will di?er across conditions.)
228 K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243
Negotiation outcomes
Particularly in the ?nal year under mandatory
rotation, the auditor appears to wield complete
power over the client. The economic prediction in
this case is that no negotiation will take place and
that auditors will simply dictate their positions to
their clients. However, the literature on ultimatum
and dictator games, in which one party has com-
plete power over the other (c.f., Davis & Holt,
1993; Kagel & Roth, 1995), suggests that people
do not strictly adhere to economic predictions but
often follow social norms related to fair play even
when given complete power over another individ-
ual. Hence, we make comparative predictions, rec-
ognizing the relative economic power rather than
the absolute economic power that dictates equilib-
rium predictions without considering the impact of
social factors. Notice from Fig. 2 that both auditors
and clients tend to use cooperative strategies (oblig-
ing and integrating) without mandatory rotation
but tend to use non-cooperative strategies (inaction
and contending) when mandatory rotation is
imposed. To the extent that cooperation leads to
successful negotiation, the following hypotheses
are proposed:
H3a : Mandatory rotation leads to less cooperation
by auditors in negotiation.
H3b : Mandatory rotation leads to less cooperation
by clients in negotiation.
H4 : More negotiations will end in impasse with
mandatory rotation than without mandatory
rotation.
When an agreement is reached, the party that
uses an obliging strategy is more likely to receive a
lower payo? and the party that uses an inaction
strategy is more likely to receive a higher payo? than
the party that uses other strategies (Pruitt &
Carnevale, 1993). This ?nding, when applied to
Fig. 2, leads us to predict
H5 : Negotiation outcomes will be closer to audi-
tors’ preferences with mandatory rotation and
closer to the clients’ preferences without man-
datory rotation.
Negotiation theory recognizes that individual
negotiators may tend towards certain unilateral
strategies but that these strategies can change over
time after interacting with the opponent as shown
in Panel B of Fig. 2. For instance, Savage et al.
(1989, Exhibit 3) predict that when one negotiator
unilaterally adopts an obliging strategy while the
opponent adopts an integrating strategy, they both
may possibly evolve into using a strategy of collab-
oration based on trust. The key point from Fig. 2 is
that the various interactive strategies that are perti-
nent to this study retain their cooperative or non-
cooperative nature from their root unilateral strate-
gies (Greenhalgh & Chapman, 1995). For instance,
all low cooperation unilateral strategies depicted in
Fig. 2 may evolve into interactive strategies that
emphasize competition. The high cooperation uni-
lateral strategies shown in Fig. 2 may evolve into
trusting collaboration or subordination. Because
people evolve into using integrative strategies that
are either cooperative or non-cooperative in a man-
ner that re?ects their cooperation under their initial
unilateral strategies, predictions about negotiated
outcomes are likely to be the same. Furthermore,
to design a study that investigates the use of integra-
tive strategy requires complex negotiations over an
extended time. For these reasons, we only examine
unilateral strategies.
Method
Participants and experimental task
Fifty-four graduate business students partici-
pated in a laboratory negotiation experiment con-
ducted over six sessions with nine participants in
each session. The typical participant was 26.2 years
old with 3.7 years of work experience. About 44% of
the participants were women.
Within each session, four participants were ran-
domly assigned the role of manager (i.e., client)
and ?ve participants were assigned the role of veri-
?er (i.e., auditor). Together in pairs, one manager
and one veri?er negotiated a value to be reported
for an asset. Hence, each session consisted of four
pairs of negotiators with one veri?er in each period
sitting out and thus earning zero fees for that
period.
In all negotiations, managers and veri?ers shared
the same set of information indicating the range and
associated probabilities of the actual value of the
asset, as shown in Fig. 3. The actual value of the
asset for each negotiation period was pre-deter-
mined using a computer-generated random process
consistent with the probabilities shown in Fig. 3.
The actual value of the asset was revealed to both
K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243 229
managers and veri?ers after each negotiation period
was over.
Experimental manipulations
Negotiation, including auditor–client negotia-
tion, is a process by which two or more parties hav-
ing di?erent preferences arrive at a joint decision
(Murnighan & Bazerman, 1990). Thus, understand-
ing preferences is key to understanding negotiation.
Preference di?erences can arise because of informa-
tion asymmetry between the client and the auditor
(Gibbins et al., 2001) or because the client and the
auditor operate under di?erent incentives (Zhang,
1999). Because our predictions are based on changes
to client and auditor incentives as a?ected by
mandatory rotation, we hold information to be
symmetric and allow only incentives to a?ect the
negotiators’ preferences.
8
The experiment manipulates mandatory rota-
tion at two levels (mandatory rotation imposed,
yes versus no) between subjects. Participants in
the mandatory rotation (MR) condition were told
that ‘‘Each manager–veri?er pair continues to
negotiate with each other for up to three periods
as long as the veri?er continues to accept the man-
ager’s submitted values.” In this condition, when
the same negotiating pair reached the third period
together, the computer screen automatically
showed a message reminding both that it was their
?nal period. Participants in the no mandatory
rotation (NoMR) condition were told that ‘‘Each
manager and veri?er pair continues to negotiate
with each other for an unlimited number of periods
as long as the veri?er continues to accept the man-
ager’s submitted values.”
The consequences of a negotiation impasse to
the veri?er are determined by the market for cli-
ents. Recall that the negotiations took place in a
laboratory and in groups of nine participants con-
sisting of four managers and ?ve veri?ers. The
addition of one more veri?er than manager cre-
ated a competitive market such that if a negotia-
tion fails and the relationship is terminated, the
manager is guaranteed a new veri?er but the ver-
i?er may or may not be reassigned to a new man-
ager as described below.
9
In the NoMR condition,
if a negotiation fails, the veri?er can only be reas-
signed to a new manager if another manager–ver-
i?er pair also fails to reach an agreement. The
data suggest that after an impasse in the NoMR
condition, the veri?er often had to wait three or
more periods before being reassigned to another
manager. In the MR condition, if a negotiation
fails, the veri?er has the same chance to be reas-
signed to a new manager as in the NoMR condi-
tion plus the chance to be reassigned to any
manager that had reached the three-period term
limit. The data suggest that in the MR condition,
after a failed negotiation the veri?er was immedi-
ately reassigned to another manager in two of
three instances and was reassigned in the next fol-
lowing period in the remaining instances.
The consequence of an impasse to a manager is a
40% reduction in earnings for the period, represent-
ing the political costs incurred by the manager. This
cost is imposed in all periods in the NoMR condi-
tion and the MR condition except in a third (?nal)
consecutive period in the MR condition with the
same veri?er in which political costs do not apply.
Procedures and negotiation rules
After arriving in the computer lab and receiving
their role assignments and a computer, the partici-
pants began by reading written instructions. The
instructions were pre-tested to clearly communicate
8
Some prior studies manipulate negotiator preferences by
providing less information to auditors than to clients. Arguably,
di?erences in opinion can occur with full information on both
sides. The present study assumes that the auditor performs
su?cient tests, etc. in order to arrive at the same probability
distribution as the client thus controlling for information
asymmetry.
9
The decision to have one out of ?ve auditors without a client
(i.e., 20%) is constrained by the availability of subjects and a
desire not to make the auditor’s incentive to avoid an impasse too
strong (such as two out of ?ve).
Asset Value Distribution
Value ($)
Probability
of being
true
Probability
of being
too high
500 5% 95%
450 5% 90%
400 10% 80%
350 10% 70%
300 15% 55%
250 15% 40%
200 15% 25%
150 10% 15%
100 10% 5%
50 5% 0%
Fig. 3. Asset value distribution.
230 K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243
the incentives and mechanics of the experiment. The
participants then completed two practice periods.
Practice periods did not a?ect earnings and are
excluded from hypotheses testing. Following other
negotiation studies (Roth, 1995), we set each period
to last two and a half minutes. After the two prac-
tice periods, 18 negotiation periods were conducted
for each session. In order to prevent end of game
antics, the participants were not told how many
periods the experiment would last. At the conclu-
sion, cash earnings were determined and the partic-
ipants paid and excused. Each session lasted
approximately 1 hour and 15 minutes.
To begin each period of negotiation, a manager
submitted an asset value on the computer screen
depicted in Fig. 4. After receiving the manager’s
proposal, the veri?er could choose to accept the
submitted asset value or to negotiate with the man-
ager on the computer screen depicted in Fig. 5.
Negotiations proceeded by typing messages back
and forth in a text box and by the manager propos-
ing new asset values on the computer screen. The
Fig. 4. Manager screen.
Fig. 5. Veri?er screen.
K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243 231
veri?er could accept the value submitted by the
manager at any time before the period ended. Once
the veri?er accepted a value submitted by the man-
ager, the period of negotiation ended and the actual
value of the asset as well as each party’s earnings for
that period were shown in a message box on the
computer screen.
If the veri?er did not accept a value submitted by
the manager before the negotiation time ran out, the
period ended in impasse. The manager then was
randomly paired with a new veri?er in the following
period and the veri?er was randomly reassigned to
another manager, if available, within the constraint
that veri?ers were not reassigned to the same man-
ager they had just left.
Payo?s
The experimental instructions showed payo?
trees for both managers and veri?ers correspond-
ing to the participant’s respective experimental
condition as shown in Fig. 6. Payo?s for manag-
ers are ?rst described followed by payo? for
veri?ers.
Panel A: No Mandatory Rotation
Managers’ Payoff:
Accepted value
Yes
Verifier’s Acceptance
No
60% Actual value
Verifiers’ Payoff:
$100 - 50% (Accepted Value -
Actual value)
Yes
Accepted value > Actual value
No
$100
Panel B: Mandatory Rotation
Managers’ Payoff:
Accepted value
Yes
Verifier’s Acceptance
No
60% Actual value (in any first and second
periods with the same verifier), or 100%
Actual value (in any third period with the
same verifier)
Verifiers’ Payoff:
$100 - 50% (Accepted Value -
Actual value)
Yes
Accepted value > Actual value
No
$100
Fig. 6. Payo?s.
232 K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243
A manager’s earnings in each period were deter-
mined by whether the veri?er accepted an asset
value submitted by the manager and, in the MR
condition, whether it was the third (?nal) period
with the same veri?er. If the veri?er accepted a sub-
mitted asset value, the manager earned that value
regardless of the asset’s actual value. In the MR
condition, an impasse resulted in the manager earn-
ing the asset’s actual value less the deduction for
political costs (of 40%) in all but the ?nal period.
In the NoMR condition, an impasse resulted in
the manager earning the asset’s actual value less
the deduction for political costs in all periods.
Hence, managers had an incentive to negotiate
higher asset values but not so high as to cause an
impasse.
A veri?er’s earnings in each period were deter-
mined by whether the veri?er accepted a value sub-
mitted by the manager and whether the accepted
value was greater than the actual value of the asset.
Re?ecting the assumption that auditors are typically
guaranteed their fee, in all cases, the veri?er earned a
?at service fee of 100 experimental dollars. However,
if the veri?er accepted an asset value greater than the
actual value, a penalty of 50% of the di?erence
between the accepted value and the asset’s actual
value was imposed. Furthermore, veri?ers earned
nothing when not assigned a manager. Hence, in
all but the ?nal period in the MR condition, veri?ers
had an incentive to negotiate lower asset values to
avoid a penalty but not so low as to cause an impasse
and therefore run the risk of not being assigned to a
manager on the subsequent round.
Because experimental earnings di?ered in
amount for managers and veri?ers, it was necessary
to create a cash reward system to make both parties’
payo?s comparable. This was done by making the
cash payment for each participant determined by
the rank of his or her cumulative earnings within
the group of participants who played the same role
(manager or veri?er) in each session as shown in
Table 1. This cash payment schedule was known
to all participants in advance.
Data coding
Two independent coders, rigorously trained and
blind to all hypotheses, coded all of the negotiation
protocols from the six sessions of the experiment as
evidence of integrating, obliging, inaction, or con-
tending strategies. The coding scheme was based
on Pruitt and Rubin (1986) and Rahim (1983).
Sillars’s (1986) coding manual was used as a techni-
cal reference in identifying di?erent types of mes-
sages relating to the four negotiation strategies.
10
Following prior negotiation studies, the unit of text
for coding was the semantic turn or ‘‘thought turn”
(Canary, Cupach, & Messman, 1995), in which a
change in semantic content indicates a di?erent
‘‘message” requiring coding (Sillars, 1986). Accord-
ingly, the coders were instructed to code each mes-
sage within the context of its dialogue to one of
the four negotiation strategies. This method is con-
sistent with the communication literature that
focuses on coding dialogue instead of sentences
(Donohue, 1992). The inter-coder reliability, as
measured by Cohen’s K is 0.72, which Bakeman
and Gottman (1997) classify as ‘‘very good” within
this context. The disagreements between the two
coders were reconciled by a third person who also
was blind to all hypotheses. Following a rather con-
servative approach that eliminated most messages
coded by only one coder, the reconciliation pro-
duced 3867 messages for analysis.
Results
Preliminary analyses
Veri?ers who failed to reach agreement with a
manager in no mandatory rotation condition typi-
cally experienced two or three o? periods prior to
being reassigned. Out of 60 possible periods of play
(20 periods per session times 3 sessions), at least one
negotiator pair failed to reach agreement in 26 of
those periods.
Table 2 presents the count of messages coded to
each of the four strategies in no mandatory rotation
condition and mandatory rotation condition. Both
veri?ers and managers submitted more messages in
mandatory rotation condition than in no manda-
tory rotation condition. Also, integrating messages
Table 1
Cash payment schedule
Managers Veri?ers
Highest cumulative earnings $30 $30
Second highest $24 $25
Third highest $17 $20
Fourth highest $10 $15
Fifth highest N/A $10
10
In the negotiation literature, the use of a negotiation strategy
is typically measured by the frequency of verbal messages that
indicate the use of a certain strategy.
K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243 233
dominate the negotiations, consistent with a general
tendency of both parties to cooperate. However, the
table shows that contending messages are relatively
frequent and that approximately one-fourth of all
messages are non-cooperative.
Dependent measures
Analyses are performed throughout using two
dependent measures. The ?rst dependent measure
is the number of messages coded to each of the four
negotiation strategies as described above. In this
case, message counts are aggregated across all peri-
ods by each negotiator (i.e., veri?er or manager).
The second dependent measure is the negotiation
strategy used by an individual negotiator within a
speci?c period as re?ected by that strategy with
the most messages during the period. That is, the
strategy with the majority of messages within a sin-
gle negotiation script is deemed to be the dominant
strategy used by the negotiator during that period.
Hence, this measure is calculated separately for each
period of negotiation and for each negotiator. On
average, the count of messages coded to the domi-
nant strategy in each period divided by the total
number of messages in the script for the period is
0.810. This relatively high proportion lends cre-
dence to our measure of strategy use.
Table 3 reports the results of analysis showing
that the use of a particular strategy is more likely
to dominate other strategies (v
2
value = 10.3362,
p < 0.0013) in mandatory rotation condition
(89.7% of periods) than in no mandatory rotation
condition (81.8% of periods). Table 4 shows the
mean number of messages coded to each strategy
in the average script. Table 4 suggests that, when
averaged over the entire sample, most messages rely
on cooperation.
Tests of hypotheses
Hypothesis 1 predicts that mandatory rotation
will a?ect the negotiation strategies of auditors. Spe-
ci?cally, H1a suggests that auditors will send oblig-
ing messages more frequently in the absence of
mandatory rotation, and H1b suggests that auditors
will send inaction messages more frequently when
mandatory rotation is imposed. Our ?rst test of
H1 consists of a general comparison across manda-
tory and no mandatory rotation conditions of the
percentage of messages coded to each of the four
negotiation strategies. Results of this analysis are
Table 2
Negotiation messages
No mandatory rotation Mandatory rotation (MR)
Non-?nal periods Final periods All periods MR condition
Count Percent Count Percent Count Percent Count Percent
Managers
Cooperative strategies
Obliging 73 9.1 58 7.0 13 5.8 71 6.72
Integrating 548 68.4 593 71.2 137 61.4* 730 69.1
Total cooperative strategies 77.5 78.2 67.3** 75.8
Uncooperative strategies
Contending 174 21.7 144 17.3* 42 18.8 186 17.6*
Inaction 6 0.8 38 4.6*** 31 13.9*** 69 6.5***
Total managers 801 100.0 833 100.0 223 100.0 1056 100.0
Veri?ers
Cooperative strategies
Obliging 168 20.8 27 2.8*** 4 1.6*** 31 2.6***
Integrating 494 61.2 639 67.0* 130 52.2* 769 63.9
Total cooperative strategies 82.0 69.8*** 53.8*** 66.5***
Uncooperative strategies
Contending 119 14.8 217 22.8*** 47 18.9 264 22.0***
Inaction 26 3.2 71 7.4*** 68 27.3*** 139 11.6***
Total veri?ers 807 100.0 954 100.0 249 100.0 1203 100.0
*p < 0.05, **p < 0.01, ***p < 0.001. All tests compare percents in a mandatory rotation condition against the percent in the no mandatory
condition.
234 K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243
Table 4
Mean number of messages coded to each strategy within each script
a
No mandatory rotation Mandatory rotation (MR)
Non-?nal periods Final periods All periods MR condition
Panel A: Managers
Obliging 0.291 0.346 0.237 0.325
Integrating 3.196 3.712 3.342 3.639
Cooperative messages 3.487 4.059 3.579 3.963
Contending 0.930 0.797 0.921 0.822
Inaction 0.025 0.203 0.711 0.304
Uncooperative messages 0.955 1.000 1.632 1.125
Panel B: Veri?ers
Obliging 0.789 0.125 0.103 0.120
Integrating 2.783 4.000 2.795 3.754
Cooperative messages 3.572 4.125 2.897 3.874
Contending 0.507 1.303 0.974 1.236
Inaction 0.132 0.401 1.256 0.576
Uncooperative messages 0.639 1.704 2.230 1.811
Panel C: Combined
Obliging 0.535 0.236 0.169 0.225
Integrating 2.994 3.856 3.065 3.696
Cooperative messages 3.529 4.092 3.234 3.919
Contending 0.723 1.049 0.948 1.029
Inaction 0.077 0.302 0.987 0.440
Uncooperative messages 0.800 1.351 1.935 1.469
Total messages 4.329 5.443 5.168 5.387
a
A script consists of the series of messages sent by one of the negotiators during a single period.
Table 3
Strategy use as evidenced by negotiation scripts
No mandatory rotation Mandatory rotation (MR)
Non-?nal periods Final periods All periods MR condition
Panel A: Managers
Total script count 185 166 44 210
Scripts w/strategy 158 153 38 191
Percent 85.4% 92.2%
*
86.4% 91.0%
Panel B: Veri?ers
Total script count 194 167 49 216
Scripts w/strategy 152 152 39 191
Percent 78.4% 91.0%
***
79.6% 88.4%
**
Panel C: Combined
Total script count 379 333 93 426
Scripts w/strategy 310 305 77 382
Percent 81.8% 91.6%
***
82.8% 89.7%
**
*
p < 0.05,
**
p < 0.01,
***
p < 0.001. All tests compare a mandatory rotation condition against the no mandatory condition. For example,
the overall hypothesis of no association between mandatory rotation condition and the use of a negotiating strategy by the individual
participants is rejected (likelihood ratio v
2
value = 10.3362, p < 0.0013) in that a strategy is more likely to be employed in periods when
mandatory rotation is imposed (89.7% of periods) compared to when it is not imposed (81.8% of periods).
Note: In each of six di?erent sessions, 18 negotiation periods were conducted (i.e., 20 periods À 2 practice periods) with four pairs of
negotiators in each period rendering 432 individual negotiations (6  18  4) having 864 possible scripts (432 periods  2 negotiators in
each period). In this case a script consists of the series of messages sent by one of the negotiators during a single period. Of these, no
message was identi?ed with a strategy by the coding process for 59 (6.8%) negotiation scripts leaving 805 scripts available for analysis (i.e.,
379 + 426). The strategy is the strategy associated with the greatest number of messages within a script.
K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243 235
summarized in Table 2. The table shows a higher
percentage (p < 0.001) of obliging messages by veri-
?ers in no mandatory rotation condition (20.8%) as
compared to mandatory rotation condition (2.6%).
The table also shows a higher percentage
(p < 0.001) of inaction messages by veri?ers in man-
datory rotation condition (11.6%) as compared to
no mandatory rotation condition (3.2%). These
results become stronger when looking at only the
?nal period in mandatory rotation condition in
which obliging messages by veri?ers are only 1.6%
of the total but inaction messages by veri?ers are
27.3%. Thus, both H1a and H1b are supported.
These results suggest that mandatory rotation
reduces the relative importance to the auditor of
maintaining a relationship with the client. They also
suggest that the experiment design appropri-
ately re?ects our assumptions about the audit
environment.
We further examine H1 by comparing overall
strategy use by veri?ers across mandatory rotation
and no mandatory rotation conditions as shown in
Table 5. Consistent with H1a, Panel B of Table 5
shows that veri?ers are more likely (p < 0.0001) to
use an obliging strategy in no mandatory rotation
condition (14.5%) compared to mandatory rotation
condition (1.1%). This result holds for both the peri-
ods leading up to the penultimate negotiation per-
iod (1.3%) as well as the ?nal period (0.0%). In
contrast and consistent with H1b, Panel B also
shows that overall, veri?ers are more likely
(p = 0.0009) to use a strategy of inaction in manda-
tory rotation condition (7.3%) compared to no man-
datory rotation condition (0.7%). This tendency
becomes especially strong in the ?nal year of nego-
tiation in mandatory rotation condition in which
23.1% of the veri?ers employ a strategy of inaction.
Hypothesis 2 predicts that mandatory rotation
will change client negotiation strategies, particularly
in the ?nal periods of negotiation before rotation
takes place (i.e., the third consecutive period with
the same auditor). Table 2 shows that in mandatory
rotation condition, manager messages re?ect inte-
grating strategies less frequently as predicted by
H2a, but only in the ?nal period (61.4% versus
68.4%). Contrary to H2b, however, managers send
fewer contending messages in mandatory rotation
condition (17.6%) than in no mandatory condition
(21.7%, p < 0.05). Similar patterns emerge for dom-
inant strategies by managers in the ?nal period in
mandatory rotation condition as shown in Panel
A of Table 5. There are similar (but not signi?cant)
decreases in integrating messages (76.3% versus
80.4%) and increases in contending messages
(13.2% versus 11.4%) by managers in mandatory
rotation condition. However, both Tables 2 and 5
show distinct increases in inaction messages by
managers in the ?nal period in mandatory rotation
Table 5
Dominant strategies
Dominant strategy No mandatory rotation Mandatory rotation (MR)
Non-?nal periods Final periods All periods MR condition
Panel A: Managers
Total script count 158 153 38 191
Obliging 8.2% 1.3%
**
5.3% 2.1%
**
Integrating 80.4% 86.9% 76.3% 84.8%
Total cooperative 88.6% 88.2% 81.6% 86.9%
Contending 11.4% 10.5% 13.2% 11.0%
Inaction 0.0% 1.3% 5.3%
**
2.1%
Total uncooperative 11.4% 11.8% 18.5% 13.1%
Panel B: Veri?ers
Total script count 152 152 39 191
Obliging 14.5% 1.3%
***
0.0%
***
1.1%
***
Integrating 77.6% 83.6% 69.2% 80.6%
Total cooperative 92.1% 84.9% 69.2%
***
81.7%
**
Contending 7.2% 11.8% 7.7% 11.0%
Inaction 0.7% 3.3% 23.1%
***
7.3%
***
Total uncooperative 7.9% 15.1% 30.8% 18.3%
*
p < 0.05,
**
p < 0.01,
***
p < 0.001. All signi?cance tests compare a mandatory rotation condition against no mandatory rotation condition.
236 K.J. Wang, B.M. Tuttle / Accounting, Organizations and Society 34 (2009) 222–243
condition (13.9% and 5.3%) compared to no manda-
tory rotation condition (0.7% and 0.0%, p < 0.001).
The data suggest that although mandatory rotation
did a?ect the negotiation strategies used by manag-
ers as predicted by H2, the speci?c predictions of
H2a and H2b are not supported.
Hypothesis 3a predicts less cooperation by audi-
tors in negotiation with mandatory rotation than
without mandatory rotation. Table 2 shows a signif-
icant decrease (p < 0.001) in cooperative messages
by veri?ers in mandatory rotation condition
(66.5%) compared to no mandatory rotation condi-
tion (82.0%). Only 53.8% of messages by veri?ers in
the ?nal period in mandatory rotation condition are
cooperative. Likewise, Table 5 shows a signi?cant
decrease in cooperative messages by veri?ers in
mandatory rotation condition (p = 0.0042). In no
mandatory rotation condition, 92.1% of veri?er
messages are cooperative, while in mandatory rota-
tion condition, only 81.7% of veri?er messages are
cooperative. Even fewer, i.e., 69.2% of veri?er mes-
sages, in ?nal periods are cooperative. Hence, the
data support H3a that mandatory rotation leads
to less cooperation by auditors in negotiation.
Hypothesis 3b predicts less cooperation by clients
in negotiation with mandatory rotation than with-
out mandatory rotation. Table 2 shows a signi?cant
decrease (p < 0.001) in cooperative messages by
mangers, but only in the ?nal periods in mandatory
rotation condition (67.3%) compared to no manda-
tory rotation (77.5%). Table 5 shows no signi?cant
di?erences in cooperative messages by managers
across the conditions. One explanation for the lack
of results for managers can be found in Panel A of
Table 5: managers tend to use less obliging strate-
gies (2.1% versus 8.2%) but more integrating strate-
gies (84.8% versus 80.4%) in mandatory rotation
condition than in no mandatory rotation condition.
These two changes tend to cancel each other while
representing a shift from a more extreme form of
cooperation (i.e., obliging) to a less extreme form
of cooperation (i.e., integrating). Hence, although
the data do not support H3b that mandatory rota-
tion leads less cooperation by clients in negotiation,
some evidence suggests that client cooperation may
not be as strong under mandatory rotation.
Hypothesis 4 predicts that mandatory rotation
will lower agreement rate in auditor–client negotia-
tion. Consistent with this prediction, Table 6 shows
that the agreement rate in mandatory rotation con-
dition (63.7%) is signi?cantly lower (p < 0.001) than
in no mandatory rotation condition (88.7%). Agree-
ment rates drop to an even lower 44.7% in ?nal
negotiation periods in mandatory rotation condi-
tion. These ?ndings strongly support H4.
In H5 we predict that when negotiations result in
an agreement, asset values will be lower with man-
datory rotation than without mandatory rotation.
11
Tests were performed using three di?erent units of
analysis (negotiating pair, veri?er, and manager).
The results are summarized in Table 7. The mean
Table 6
Negotiation outcomes
Treatment Agreement Impasse v
2
p-Value
Count Percent Count Percent
Panel A
All periods in MR condition 130 63.7 74 36.3
No mandatory rotation 181 88.7 23 11.3
Di?erence À25.0 25.0 35.18