Description
While some research suggests that explicit incentives to meet time budgets have recently
been reduced at audit firms, there is also evidence indicating that audit seniors and staff
still feel at least implicit pressure to meet budgets. We examine the possibility that both
of these findings tell a part of the story. Specifically, we explore whether, and under what
conditions, seniors and staff are implicitly encouraged to underreport time through future
engagement staffing decisions and the performance evaluation process. Further, we consider
the extent to which agency theory can serve as a framework for understanding
how the incentives of audit managers and partners influence how they view underreporting
by their engagement staff.
Audit team time reporting: An agency theory perspective
q
Christopher P. Agoglia
a,?
, Richard C. Hat?eld
b
, Tamara A. Lambert
c
a
Department of Accounting, Isenberg School of Management, University of Massachusetts Amherst, 121 Presidents Drive, Amherst, MA 01003, United States
b
Culverhouse School of Accountancy, The University of Alabama, United States
c
Department of Accounting, College of Business and Economics, Lehigh University, United States
a r t i c l e i n f o
Article history:
Available online 16 May 2015
a b s t r a c t
While some research suggests that explicit incentives to meet time budgets have recently
been reduced at audit ?rms, there is also evidence indicating that audit seniors and staff
still feel at least implicit pressure to meet budgets. We examine the possibility that both
of these ?ndings tell a part of the story. Speci?cally, we explore whether, and under what
conditions, seniors and staff are implicitly encouraged to underreport time through future
engagement staf?ng decisions and the performance evaluation process. Further, we con-
sider the extent to which agency theory can serve as a framework for understanding
how the incentives of audit managers and partners in?uence how they view underreport-
ing by their engagement staff. We place participants in a scenario in which they are
responsible for evaluating an engagement senior who appears to have worked more hours
than were budgeted. We manipulate the senior’s reporting accuracy (underreporting versus
accurate reporting) and the desirability of the client (more versus less desirable). We ?nd
that, when managers’ agency-related incentives con?ict more strongly with those of the ?rm
(more desirable client), they tend to tacitly encourage underreporting through their evalua-
tions of the senior’s performance. Managers are also more likely to request an underreporter
on a future engagement. In contrast, partners placed in the same setting showno evidence of
encouraging underreporting. Thus, our results suggest that managers’ tacit encouragement
of underreporting is contrary to what the ‘‘principals’’ of the ?rm (i.e., partners) appear to
want. Further, while ?rms may have reduced their emphasis on formal, explicit incentives
to underreport, it appears likely that implicit manager incentives persist.
Ó 2015 Elsevier Ltd. All rights reserved.
Introduction
Our study examines the role agency incentives play in
diminishing the effectiveness of ?rm policies aimed at
reducing underreporting of time. Speci?cally, we explore
the extent to which, and under what circumstances, supe-
riors implicitly encourage such behavior in audit staff.
Underreporting occurs when an auditor does not record
all hours worked on a particular engagement and is
believed to negatively affect audit quality, to lead to other
unethical and dysfunctional audit behaviors that can
increase audit risk, and to result in incorrect information
being used in client pricing and retention decisions
(e.g., Donnelly, Quirin, & O’Bryan, 2003; Public Oversighthttp://dx.doi.org/10.1016/j.aos.2015.03.005
0361-3682/Ó 2015 Elsevier Ltd. All rights reserved.
q
We thank Mark Peecher (editor), two anonymous reviewers, Lindsay
Andiola, Ann Backof, Jean Bedard, Bradley Bennett, Bill Brown, Bob
Cornell, Steven DeSimone, Kirsten Fanning, Jace Garrett, Ryan Guggen-
mos, Denise Hanes, Lynn Hannan, Jennifer Joe, Steve Kachelmeier,
Kathryn Kadous, Elisa Lee, Ben Luippold, Chris Nolder, Mary Parlee, Peter
Roebuck, Aaron Saiewitz, Chad Stefaniak, Hun-Tong Tan, David Wood,
and workshop participants at Bentley University, Georgia State Univer-
sity, Oklahoma State University, the 2010 International Symposium on
Audit Research, the 2011 AAA Audit Mid-Year Meeting, and the 2011 AAA
Annual Meeting, for their helpful comments and assistance.
?
Corresponding author.
E-mail address: [email protected] (C.P. Agoglia).
Accounting, Organizations and Society 44 (2015) 1–14
Contents lists available at ScienceDirect
Accounting, Organizations and Society
j our nal homepage: www. el sevi er. com/ l ocat e/ aos
Board, 2000; Shapeero, Koh, & Kilough, 2003). As a conse-
quence, audit ?rm policies expressly prohibit underreport-
ing (McNair, 1991).
Although meeting time budgets has traditionally been a
signi?cant performance evaluation focus in audit ?rms
(Lightner, Adams, & Lightner, 1982; McNair, 1991), a sur-
vey of partners and managers conducted by Buchheit,
Pasewark, and Strawser (2003) suggests that such formal,
explicit emphasis on meeting budgets has ebbed.
However, there is also evidence indicating that audit
seniors still feel pressure to underreport (Sweeney &
Pierce, 2006). One possibility is that both of these ?ndings
tell a part of the story. That is, explicit pressure to meet
budget may in fact be reduced, while implicit pressure still
exists (e.g., seniors do not want to appear less valuable or
productive than peers who report meeting budget;
Sweeney & Pierce, 2006). Our study explores whether staff
are still implicitly encouraged to underreport time through
future engagement staf?ng decisions and the performance
evaluation process and, if so, the extent to which agency
theory can serve as a framework for understanding how
incentives perpetuate the behavior, as well as a basis for
creating potential solutions.
Audit managers and partners have, at times, differing
incentives with respect to the time reporting behavior of
the engagement team. Agency theory predicts that
partners, as ‘‘principals’’ of the ?rms, have a longer-term
perspective and, thus, believe that their interests align clo-
sely with those of the ?rm as a whole (e.g., better retention
of good, honest employees who are reluctant to misreport
hours worked; more accurate costing ?gures for better cli-
ent acceptance/retention decisions and better long-run
alignment of resources). As a result, partners are likely to
prefer that their engagement teams report their time accu-
rately. However, managers are likely more in?uenced by
shorter-term incentives to complete the engagement
within the budgeted time (e.g., to avoid fee pressure on
desirable clients; to impress partners with good realization
rates) (Sweeney & Pierce, 2006). Managers also spend more
time with staff at the worksite than do partners, resulting
in managers having more accurate information regarding
staff hours (Otley & Pierce, 1996). This informational
advantage provides managers with an opportunity to act
on their incentive to implicitly encourage engagement
team underreporting. We, therefore, explore how the dif-
fering roles (and related incentive structures) of audit
managers and partners in?uence how they view the
practice of underreporting, anticipating that managers
and partners will act differently, but in a manner consis-
tent with agency theory predictions (i.e., as agents and
principals, respectively).
Agency theory predicts that, where there is strong con-
?ict between the incentives of the principal and agent, the
agent will tend to act in his/her self-interest when pro-
vided the opportunity (e.g., an informational advantage
over the principal). If no strong con?ict exists, then the
agent’s actions will tend to align more with the principal’s
incentives (Jensen & Meckling, 1976). Thus, if an agency
framework applies in the underreporting context, one
would expect that the strength of managers’ (con?icting)
incentives in?uences their tendency to encourage the
behavior. One factor that affects the strength of such a con-
?ict relates to desirability of the client. For example, if a
manager has a strong preference for a particular client
(e.g., the client is close to home or the manager gets along
well with client management), he or she is more likely to
have a stronger desire for a subordinate to underreport in
order to maintain audit fees near their current levels to
help promote client retention and, in turn, increase the
likelihood he/she remains assigned to this desirable client.
In contrast, there are generally other engagements that a
manager ?nds less attractive and is less interested in
retaining. On such less desirable clients, the strength of
the agency con?ict is lessened, and the manager is more
likely to be more accepting of budget overruns.
We conduct two experiments to investigate the in?u-
ence agency incentives have on underreporting. In sepa-
rate manager and partner experiments, we place
participants in a scenario in which an engagement senior
and her staff appear to have worked more hours than were
budgeted. We ask participants to assume they are the
senior’s immediate supervisor (i.e., the manager on the
engagement) and task them with evaluating the senior’s
performance. We manipulate reporting accuracy; that is,
whether staff underreport (i.e., report meeting the budget)
or report all the hours worked (i.e., report exceeding the
budget). We also manipulate client desirability (i.e., more
desirable versus less desirable).
Our results indicate that, through their evaluation of
staff who exceed budget, managers are more likely to tac-
itly encourage underreporting (relative to accurately
reporting exceeding budget) when client desirability is
high. When the client is less desirable, managers’ prefer-
ence for underreporters dissipates (i.e., their preferences
begin to reverse). These results are consistent with agency
theory expectations, as managers behave more like agents,
acting in their own interest when their incentives con?ict
with the ?rm’s, but acting more in the ?rm’s interest when
there is no strong con?ict between their incentives and the
?rm’s. We also ?nd that managers are more likely to
request an underreporter on a future engagement.
Further, we ?nd that reporting accuracy in?uences man-
agers’ future staf?ng decisions through its effect on staff
evaluations; however, this mediating relationship is mod-
erated by client desirability (i.e., moderated mediation).
Speci?cally, managers’ evaluations of the senior’s perfor-
mance are more predictive of their willingness to select
the senior for a future engagement when client desirability
is high than when it is low.
In contrast, partners who assume the role of the senior’s
immediate supervisor (i.e., manager) show no preference
for the underreporter, on average, either through their
evaluations of the senior or their future staf?ng
preferences. However, our results are not supportive of
our predictions that partners will react to underreporting
as ?rm guidance suggests. That is, partners in the accurate
reporting condition do not evaluate the senior signi?cantly
higher, nor are they signi?cantly more likely to request the
senior for a different engagement, than those in the under-
reporting condition.
To explore this difference between our expected results
and our observed results for partners, we examine the
2 C.P. Agoglia et al. / Accounting, Organizations and Society 44 (2015) 1–14
partner data more closely. Although partners’ manipula-
tion check responses for reporting accuracy were signi?-
cantly different by condition, we ?nd that, unlike
managers, many partners in the accurate reporting condi-
tion were not convinced that the senior accurately
reported all hours worked. In other words, there were a
number of partners who believed the accurate reporter
was likely underreporting, at least to some extent, mean-
ing that they did not see the senior as an accurate reporter.
As we are interested in whether partners prefer staff they
view as accurate reporters to staff they suspect are under-
reporters, we perform additional analyses examining the
subset of partners who were more convinced that the
accurate reporter was, in fact, reporting accurately.
1
We
?nd that, more in line with an agency framework, these
partners do prefer seniors who go over budget to accurately
report their hours. As anticipated, client desirability has no
effect on their evaluations of the senior. Further, we conduct
additional analyses that, again, utilize the full partner data
set. These additional analyses, which use partners’ beliefs
about the likelihood that the senior recorded all hours
worked (a scaled, measured variable), support the results
obtained with the subset. That is, we observe a positive asso-
ciation between partners’ perceptions about the senior’s
reporting accuracy and partner’s preference for the senior.
Our study contributes to the literature in a number of
important ways. First, our study advances the literature
on underreporting of time by taking a ?rst step toward rec-
onciling the ?ndings of prior studies that suggest both: (a)
reduced emphasis on meeting time budgets as a formal
performance metric (Buchheit et al., 2003) and (b) audit
seniors still feel pressure to underreport (Sweeney &
Pierce, 2006). We present evidence that managers still tac-
itly encourage underreporting by their engagement team,
even in a setting without explicit mention of time budgets
as a performance metric. Thus, while explicit incentives
may have reduced, it appears likely that implicit manager
incentives to underreport persist. Second, our results sug-
gest that managers’ tacit encouragement of underreporting
is contrary to what the ‘‘principals’’ of the ?rm appear to
want. That is, unlike managers, partners placed in the same
setting show no evidence of encouraging underreporting,
as our results suggest that partners do not prefer the
underreporting senior to the accurate reporter. Further,
we ?nd a positive association between partners’ beliefs
about how likely it is the senior reported all hours worked
and their preference for the senior.
Third, our agency framework perspective on the phe-
nomenon of underreporting suggests that agency theory
can assist researchers, regulators, and practitioners wish-
ing to understand and curb the behavior. Existing research
on principal-agent con?ict (e.g., in the context of project
continuation decisions) offers interesting avenues to
explore fresh solutions to the problem (e.g., establishing
appropriate ‘‘principal-oriented’’ incentives; incurring
monitoring costs) (Jensen & Meckling, 1976). Finally, prac-
titioners should be interested in our ?nding that manager
approval of underreporting is not uniform across audits.
In other words, client pro?tability is not uniformly affected
by underreporting. This suggests there are signi?cant costing
issues for ?rms, as the level of underreporting varies with
contextual/situational factors and the strength of agency-
related incentives in play for the manager. Thus, ?rms may
want to consider agency-related approaches to better align
audit manager incentives with long-term ?rm objectives.
The remainder of the paper is organized as follows. The
next section discusses previous literature and develops our
hypotheses. The third and fourth sections detail the
method and results for Experiments 1 and 2, respectively.
The ?nal section offers conclusions, limitations, and sug-
gestions for future research.
Background and hypothesis development
Underreporting the amount of time spent on an audit
engagement has long been labeled an unethical practice
with dysfunctional downstream effects that threaten audit
quality (e.g., it can lead to unrealistic future time budgets,
which in turn could lead to premature sign-off or docu-
mentation of work not performed) (Ponemon, 1992;
Sweeney & Pierce, 2006). Nevertheless, early survey
research indicated that auditors of all ranks believed that
underreporting resulted in better performance evaluations
and promotion opportunities (Lightner et al., 1982;
McNair, 1991). Such beliefs were likely warranted, as evi-
denced by one partner’s description of realization rates
as having been the ‘‘diamond of metrics’’ with respect to
performance evaluations (Buchheit et al., 2003).
2
However, there is more recent evidence that ?rms have
reduced emphasis on formal time budget metrics. In a sur-
vey of partners and managers, Buchheit et al. (2003) report
a reduction in budget pressure, primarily due to a reduced
emphasis on time budgets within the ?rm’s performance
evaluation system. Similarly, Sweeney and Pierce (2004,
2006) report evidence of reduced emphasis on time bud-
gets at audit ?rms, particularly for higher-ranking auditors.
While formal emphasis on meeting time budgets appears
to have been reduced, Sweeney and Pierce (2006) point
out that it has not been eliminated as a formal, explicit
metric of performance and still is a factor, albeit dimin-
ished, in promotion decisions. Further, their results suggest
that auditor perceptions of the pressure to underreport
vary by rank; staff and seniors still report that managers
often implicitly encourage the practice (Sweeney &
Pierce, 2006).
3
It is unclear then whether, in the current
audit environment, supervisors actually will still encourage
1
Focusing on partners in the accurate reporting condition who viewed
the senior as reporting accurately results in a subset of partners that is
equally as likely as our manager participants to view the accurate reporter
as reporting. We discuss this further in the Experiment 2 Partner Results
section.
2
Most audits are ?xed fee engagements in which additional hours
worked are not billed to the client (Buchheit et al., 2003). Thus, budget
overruns primarily impact auditors’ realization rates and future client
retention decisions (Akers & Eaton, 2003; McNair, 1991; Shapeero et al.,
2003), which in turn have traditionally affected performance evaluations
and promotions (Buchheit et al., 2003).
3
For a sample of more recent anecdotal evidence, see:http://www.
stuffaccountantslike.com/?p=84;http://steeplemedia.com/blogs/krupo/
archive/2009/06/04/eating-hours-is-not-cool.aspx; andhttp://www.ac-
countantbyday.com/2011/05/06/life-of-an-auditor-billable-hours/.
C.P. Agoglia et al. / Accounting, Organizations and Society 44 (2015) 1–14 3
underreporting by their engagement teams. One possibility
is that both of these ?ndings hold part of the answer: expli-
cit pressure to meet budget may in fact be reduced, but
implicit pressure may still exist (e.g., managers and seniors
do not want to appear less valuable or productive than peers
who report meeting budget; Sweeney & Pierce, 2006).
If encouragement of underreporting persists, it would
likely manifest in two in?uential ways. First, managers
might differentially evaluate staff performance based on
perceived underreporting (Otley & Pierce, 1996; Sweeney
& Pierce, 2006). For example, when meeting budget is cru-
cial to the manager, he/she has an incentive to express
(through performance evaluations) approval of staff who
appear to have underreported to meet budget, and disap-
proval of staff who report going over budget. Second, man-
agers may be more or less likely to choose staff for future
engagements based on their willingness to underreport
(Otley & Pierce, 1996; Sweeney & Pierce, 2006). That is, if
managers have similar budgetary concerns about
subsequent engagements, they are more likely to prefer
to utilize audit staff who appear willing to help them meet
future budgets as well. It is important to note that these
two decisions are key determinants of the career progres-
sion of audit staff.
Underreporting and the agency problem
An agency problem exists in all organizations and coop-
erative efforts in which a principal (e.g., an owner) utilizes
an agent (e.g., a manager) for the purpose of delegating
responsibility to him or her (Jensen & Meckling, 1976).
Agency theory suggests that, as long as the goals of the
principal and agent are aligned, the agent will make deci-
sions that maximize the goals of the principal. However,
when the goals of the principal and agent diverge and
the agent has the opportunity to act in his/her self-interest
(i.e., has relevant information that the principal lacks,
referred to as ‘‘information asymmetry’’), agency theory
predicts that the agent will make decisions that maximize
his/her self-interest over the principal’s interests (Booth &
Schulz, 2004; Jensen & Meckling, 1976). Prior research sup-
ports the notion that agents tend to act in their own inter-
est when they have both the incentive and opportunity to
do so. For example, agents have been shown to continue
failing projects in order to avoid negative reputational
effects and to manage earnings to achieve higher bonuses
(e.g., Beaudoin, Agoglia, Tsakumis, & Guggenmos, 2015;
Booth & Schulz, 2004; Guggenmos & Agoglia, 2015;
Jensen & Meckling, 1976). In this study, we employ an
agency framework to better understand the phenomenon
of underreporting of time and why current approaches
are unlikely to eliminate this behavior.
On an audit engagement, an audit manager acts as an
agent of the audit ?rm which is controlled, or ‘‘owned’’,
by the audit partners. The manager generally has a more
‘‘hands on’’ role in the day-to-day activities of the engage-
ment than the partner (Wolf, 1981). Because of this, there
is likely to be information asymmetry between the partner
and the manager regarding certain aspects of the audit (i.e.,
opportunity for the manager to act in his/her own interest).
For example, managers likely have a better idea of how
many hours their staff actually work, given that they typi-
cally spend more time on-site and interact more regularly
with the engagement team than partners (Gibbins &
Trotman, 2002). Thus, this informational advantage leaves
the manager with a dilemma: the manager can encourage
the provision of accurate information to achieve a positive
collective outcome for the ?rm or, if agency incentives
persist,
4
encourage strategic misrepresentation of the hours
worked by the engagement team in order to secure a
positive outcome for him/herself. The dilemma is less
straightforward in this context than in a typical agency set-
ting, as audit managers generally aspire to be partners (i.e.,
audit managers are somewhat ‘‘nontraditional’’ agents).
That is, as ‘‘principals in training’’, audit managers could
tend toward the behavior of a principal.
Agency con?ict and manager incentives for evaluating the
performance of underreporters and accurate reporters
Resource constraints have intensi?ed under the
Sarbanes-Oxley Act (SOX) of 2002, putting pressure on
audit fees and making retention of marginally pro?table
clients even more tenuous (Rama & Read, 2006). Given that
low realization rates lead to an increased chance of sever-
ing the relationship with a client (either due to the ?rm
severing the relationship for pro?tability reasons or the
client severing the relationship as a result of the ?rm rais-
ing fees), managers have an incentive to tacitly encourage
underreporting in order to avoid losing an attractive client.
A manager will ?nd some clients more desirable than
others due to various situational and contextual factors
that are inherent to the client. For example, a manager is
likely to prefer: a client that is convenient for the man-
ager’s work schedule, a client that is easy to get to, client
management that the manager gets along particularly well
with personally, a client that is within the manager’s
industry of interest, or an engagement that involves an
in?uential partner at the manager’s of?ce.
Since a manager working on a client he or she ?nds
more desirable will be more concerned about losing that
client, the agency con?ict (i.e., misalignment of their incen-
tives with those of the ?rm) faced by the manager will be
more pronounced (i.e., appropriately costing a less prof-
itable client versus taking action to obscure the cost of a
client the manager ?nds desirable). That is, the manager
will have a heightened sensitivity to budget overruns that
would re?ect poorly on his/her management of the
engagement or that could eventually lead to a severing of
the relationship with that client. In these circumstances,
the manager has greater incentive to tacitly encourage
strategic misrepresentation of the hours worked on the
engagement and, in turn, to evaluate underreporters sig-
ni?cantly higher than honest reporters. On the other hand,
4
As noted, such manager incentives could be implicit (e.g., to avoid fee
pressure to help with retention of desirable clients; impress partners with
good realization rates) and/or explicit (e.g., formal metrics used for
manager performance evaluations and promotion decisions). The ?nding
that audit staff report still feeling time budget pressure from managers is
consistent with the notion that such agency-related incentives persist for
managers (Sweeney & Pierce, 2006).
4 C.P. Agoglia et al. / Accounting, Organizations and Society 44 (2015) 1–14
a manager’s agency con?ict is reduced when the client is
less desirable. With a less desirable client, the manager
has less incentive to mask the true cost of the client to
the ?rm, and the strength of the principal-agent con?ict
is reduced. Without strong con?ict, we expect managers
to begin to act more in line with ?rm policy (i.e., ?rm guid-
ance would suggest evaluating accurate reporters higher
than underreporters).
Thus, consistent with agency theory, we expect man-
agers to act more in their own interest when con?ict
between their incentives and the ?rm’s is heightened, but
more in the ?rm’s interest when there is no strong con?ict
between their incentives and the ?rm’s. Speci?cally, we
expect managers to tacitly encourage underreporting by
preferring underreporters to accurate reporters when they
?nd the client more desirable (i.e., we expect a larger dis-
parity in managers’ evaluations of underreporters and
accurate reporters when staff exceed budget for a desirable
client). When managers ?nd the client less desirable, their
preference for underreporters will dissipate and likely
approach the behavior prescribed by ?rm guidance.
5
That
is, relative to when managers ?nd the client more desirable,
managers will tend to evaluate honest reporters marginally
higher than underreporters, better re?ecting the interests of
their principals. Given these expectations, we propose a
disordinal interaction of the form speci?ed below
(see Fig. 1):
H1. There will be a disordinal interaction between staff
reporting accuracy and client desirability such that man-
agers’ evaluations of staff will be highest for underre-
porters on a more desirable client, and then decrease in the
following order: accurate reporters on a less desirable
client, underreporters on a less desirable client, and
accurate reporters on a more desirable client.
Managers’ future staf?ng decisions: implications for accurate
reporters
The choice of particular team members for an engage-
ment can have signi?cant time budget implications for
the audit manager. Suppose, for example, that a manager
has worked with two audit seniors on past engagements.
Assume one of the seniors has, on occasion, missed bud-
gets and the other has always met his/her budgets
(although the manager suspects he/she is underreporting).
If, in an effort to protect realization rates on future engage-
ments, the manager is more likely to select the underre-
porter than the accurate reporter, there are potentially
serious negative implications for the accurate reporter.
For example, the accurate reporter is less likely to be
assigned to desirable engagements in the future which, in
turn, can in?uence raises, promotions, and continued
employment (Doby & Caplan, 1995). As a result, the ?rm
is more likely to lose honest, capable employees in the
process.
The choice of working with particular seniors or staff on
future engagements presents a different set of dilemmas to
managers than evaluating their performance on a com-
pleted engagement. That is, past performance evaluations
may not relate directly to a manager’s desire to select that
individual for a different future engagement (Kaplan &
Reckers, 1985). For instance, even if a manager evaluates
an accurate reporter who exceeds budget favorably (partic-
ularly with a non-preferred client), he/she is still likely to
be less inclined to select that staff member for future
engagements, as a reluctance to underreport could hurt
the manager’s chances of meeting future budgets. Having
staff willing to underreport provides the manager with dis-
cretion/?exibility to meet budget on future engagements.
6
Thus, on average, we expect managers to prefer underre-
porters (relative to accurate reporters who have gone over
budget) on future engagements regardless of the desirability
of the current client. We, therefore, propose the following
main effect hypothesis:
H2. Audit managers will be more likely to request a senior
for a different engagement when the senior underreports
than when the senior accurately reports exceeding the
budget.
Partners’ interests
Our discussion of agency theory suggests that managers
will tend to act in a utility maximizing manner by tacitly
encouraging underreporting when contextual factors lead
to a misalignment of their incentives from those of their
?rms. However, agency theory implies that partners will
react differently to underreporting than managers. As
owners of the ?rm, their interests are more aligned with
the ?rm’s interests. And, given ?rm policies that explicitly
prohibit underreporting due to the wide variety of prob-
lems induced by the practice and by the related unrealistic
time budgets that result from past underreporting (e.g.,
inappropriate costing of less pro?table clients, failure to
gather suf?cient evidence/documentation of procedures
not performed, and lower employee morale and increased
5
The phenomenological strength of our client desirability manipulation
was not intended to result in the absence of an incentive (i.e., complete
removal of con?icting agency incentives), since a less desirable client is still
part of a manager’s portfolio of clients. Thus, while we expect managers’
preferences for underreporters to reverse (and begin to approach ?rm
preferences) for the less desirable client condition in our study, we do not
expect the disparity between managers’ evaluations of underreporters and
accurate reporters to be as large for less desirable clients as for more
desirable clients (i.e., we anticipate a disordinal interaction of the form
represented in Fig. 1).
6
Alternatively, a manager could evaluate the underreporter’s perfor-
mance higher on the current engagement, particularly if the client is one
the manager ?nds desirable, but not necessarily reward this staff/senior
with selection for a future engagement. That is, the manager could view the
underreporter’s performance more favorably because on this particular
engagement time needs to be eaten in order to meet budget, while on an
undetermined future audit it may be less necessary. However, agency
theory (which provides the theoretical underpinning for our hypotheses)
suggests that managers might prefer the budgeting ‘‘?exibility’’ of having
staff (on an undetermined future engagement) who are willing to
underreport if necessary. That is, agency theory predicts that an agent will
act in his/her own interest when there is both incentive (i.e., goal con?ict
with the principal) and opportunity to act in his/her interest at the expense
of the principal. Having a preference for underreporters helps allow a
manager to take advantage of these opportunities when such incentives
exist.
C.P. Agoglia et al. / Accounting, Organizations and Society 44 (2015) 1–14 5
turnover), partners should prefer that their engagement
teams report their time accurately.
Relative to managers, partners’ role and incentives will
typically cause them to have a longer-term perspective
when it comes to these negative audit quality-related ram-
i?cations of underreporting, as well as a greater concern
for obtaining a true picture of the cost of performing a par-
ticular audit. In contrast to managers (but consistent with
agency theory), when partners are asked to place them-
selves back in the role of an engagement senior’s immedi-
ate supervisor (i.e., the audit manager) and consider the
evaluation and staf?ng decisions the supervisor is respon-
sible for, we expect them to generally prefer staff to accu-
rately report hours when they exceed budget, as opposed
to staff underreporting their hours (i.e., to evaluate higher
those staff whom they believe have accurately reported
and to select them for future engagements). Therefore,
we pose the following hypotheses:
H3. Audit partners will evaluate a senior who accurately
reports exceeding the budget higher than one who under-
reports.
H4. Audit partners will be more likely to request a senior
for a different engagement when the senior accurately
reports exceeding the budget than when the senior
underreports.
Experiment 1: Managers
Method
Participants, task, and procedure
We conduct our ?rst experiment to investigate whether
audit managers impose an implicit pressure to underreport
time and, if so, whether their desire to retain the client
in?uences this pressure. We placed manager participants
in a scenario in which an engagement senior and her staff
appear to have worked more hours than were budgeted.
We asked participants to assume they are the senior’s
immediate supervisor (i.e., the manager on the engage-
ment) and tasked them with evaluating the senior’s
performance.
Manager participants were given background informa-
tion on the client and the engagement, along with current
year information on the audit and engagement team. They
were told that a clean opinion was rendered on the client
after some adjustments to the ?nancial statements were
made. Information provided to participants relating to
the senior on the engagement included generally positive
information. However, an ‘‘area for improvement’’ (i.e., a
mildly negative piece of information) was provided in
order to generate variability in participants’ responses
and to reduce the likelihood of a ceiling effect.
Speci?cally, each participant was told: the senior, Lisa
Martin, was a new member of the engagement team this
year; this was their ?rst time working with her; she
seemed to have a good rapport with the client and her
staff; she kept them informed of unusual items; she made
very few documentation errors; and she occasionally con-
tacted them with questions regarding proper ?rm proce-
dure which they hope on future engagements she will be
more prepared to apply on her own. Participants were then
given more detailed information about the engagement
and responded to questions about their perceptions of
the senior’s performance. Finally, the managers answered
a series of case-related and demographic questions, includ-
ing manipulation checks.
Participants in Experiment 1 were 100 audit managers.
7
Managers had, on average, 9.7 years of audit experience.
There are no signi?cant differences (p’s > .20) for
participants across experimental conditions for any of the
demographic measures (e.g., overall audit experience, expe-
rience evaluating subordinates, ?rm type). Including the
demographic measures as covariates does not signi?cantly
affect the results or alter the inferences drawn.
Less Desirable Client More Desirable Client
P
e
r
f
o
r
m
a
n
c
e
E
v
a
l
u
a
t
i
o
n
Underreporter
Accurate Reporter
Fig. 1. H1 Prediction: managers’ evaluations of senior’s performance.
7
We mailed instruments to 900 managers from a list provided by the
American Institute of Certi?ed Public Accountants (AICPA). We received
replies from 109 respondents and 45 were returned as undeliverable. The
resulting response rate is 12.75%. There were no differences between early
and late respondents. There were nine unusable responses resulting from
completion by inappropriately classi?ed individuals such as staff accoun-
tants, payroll clerks, and tax accountants. Our conclusions remain the same
with or without these individuals.
6 C.P. Agoglia et al. / Accounting, Organizations and Society 44 (2015) 1–14
Independent variables
We utilized a 2 Â 2 (reporting accuracy by client desir-
ability) between-participants design. We manipulated
reporting accuracy as either underreporting or accurate
reporting of hours worked by the engagement team. In
both cases, there are indications that the engagement team
worked more hours than budgeted. In order to establish
this, participants were told that they felt con?dent the
team worked beyond the budgeted hours throughout the
engagement and that engagement team members often
left voicemails and submitted electronic workpapers for
review before 8 a.m. and after 7 p.m. In the ‘‘underreport’’
condition, participants were told, ‘‘Although Lisa and the
rest of the engagement staff appeared to work long hours,
their reported hours did not exceed their time budget.’’
Because it had been previously established in both condi-
tions that the engagement team worked more hours than
budgeted, the participants were led to infer that Lisa and
the engagement team underreported their hours in the ‘‘un-
derreport’’ condition. In the ‘‘accurate reporting’’ condition,
participants were told that, ‘‘Lisa and the rest of the engage-
ment staff appeared to work long hours and their reported
hours signi?cantly exceeded their time budget (i.e., it
appears that all excess hours worked were reported).’’
For the second independent variable, the client was either
moreor less desirable. We manipulatedseveral aspects of the
engagement since a particular individual may value certain
dimensions more than others, depending on their lifestyles
and priorities. For example, participants in the more (less)
desirable client condition were informed: they ‘‘have always
(never) gotten along particularly well with client manage-
ment on a personal level’’ and that ‘‘the recent audit was no
exception’’; this engagement does not come (comes) at ahectic
time in their personal work schedules; the client is located
near (far) from their home with little (heavy) traf?c on the
commute; the partner on the engagement ‘‘is a very experi-
enced (newjunior) partner inyour of?ce’’ who has some ‘‘high
(low) pro?le engagements that you would be interested (have
little interest) inworkingoninthe future’’ (italics not included
in experimental instrument).
Dependent variables
To test H1, we examine managers’ evaluations of the
audit senior’s performance. Managers were asked to rate
the senior’s performance with respect to this engagement.
They recorded their responses for these measures on ele-
ven-point scales, where 1 = ‘‘Poor’’ and 11 = ‘‘Outstanding’’.
For H2, we asked howlikely they were to request the senior
on a different engagement in the future. Responses were
again recorded on an eleven-point scale, where 1 = ‘‘Very
Unlikely’’ and 11 = ‘‘Very Likely’’.
Manager results
Results relating to our manager hypotheses are analyzed
within a 2 Â 2 ANOVA framework (reporting accuracy by
client desirability), with participants’ evaluations of seniors
and their likelihood of requesting the senior on a different
future engagement serving as dependent variables.
Reported p-values are two-tailed unless otherwise noted.
Manipulation checks
Manipulation checks for both independent variables indi-
cate that managers understoodthe manipulations. Managers
were asked how likely they feel it is that all of the hours
worked on the engagement were recorded. They indicated
their responses on an 11-point scale where 1 = ‘‘Very
Unlikely’’ and 11 = ‘‘Very Likely’’. Managers in the underre-
porting condition reported an average response of 3.94 com-
pared to a mean of 8.67 in the accurate reporting condition
(t = 10.722, df = 94, p < 0.001, one-tailed). Managers were
also asked about the strength of their personal preference
to retain the client on an 11-point scale in which 1 = ‘‘Very
Weak’’ and 11 = ‘‘Very Strong’’. Those in the higher client
desirability condition reported an average response of 9.38
compared to 5.73 for managers in the lower desirability con-
dition (t = 8.686, df = 93, p < 0.001, one-tailed).
8
Hypotheses 1 and 2
Hypothesis 1 speci?es a disordinal interactive effect of
the two independent variables on managers’ evaluation of
the senior’s performance. That is, when client desirability
is high, managers’ incentives are in con?ict with the ?rm’s,
and we expect the difference between managers’ evalua-
tions of underreporters and accurate reporters to be large
(i.e., underreporters’ evaluations will be substantially
higher than those of accurate reporters). In contrast, when
the client is less desirable, there is less con?ict between
managers’ incentives and the ?rm’s, and we expect man-
agers to begin to approach the evaluation of the senior as
?rm guidance suggests and their preference for underre-
porters to begin to reverse (i.e., accurate reporters’ evalua-
tions should be somewhat higher than underreporters’).
Given the interaction speci?ed by H1 and depicted in
Fig. 1, we test the hypothesis with a planned contrast using
the following weights: À1 for the underreporter/less desir-
able client condition (Cell 1), +2 for the underreporter/more
desirable client condition (Cell 2), +1 for the accurate repor-
ter/less desirable client condition (Cell 3), and À2 for the
accurate reporter/more desirable client condition (Cell 4).
Table 1 presents the descriptive statistics and results from
the contrast used to test H1, along with results of the tradi-
tional ANOVA, for the senior’s overall performance evalua-
tion measure. Cell means are in a pattern consistent with
the contrast, with Cell 2 highest, followed by Cells 3, 1,
and 4 (means = 8.00, 7.70, 7.50, and 6.77, respectively).
The planned contrast is signi?cant at p < 0.001.
9
Further,
the ANOVA interaction term is signi?cant at p = 0.003.
These results support H1.
8
We note that, reassuringly, the client desirability manipulation did not
signi?cantly in?uence responses to the reporting accuracy check question
(main and interactive effects of client desirability manipulation p’s > 0.240)
and the reporting accuracy manipulation did not signi?cantly in?uence
responses to the client desirability check question (main and interactive
effects of client desirability manipulation p’s > 0.520).
9
A Jonckheere-Terpstra test con?rms the predicted ordering of cells
(p < 0.001, one-tailed). Further, all simple comparisons between cells are in
the expected direction, and all are signi?cant at p < 0.07 (one-tailed) except
for the comparisons between Cells 1 and 3 (p = 0.266) and Cells 2 and 3
(p = 0.176).
C.P. Agoglia et al. / Accounting, Organizations and Society 44 (2015) 1–14 7
Hypothesis 2 predicts that audit managers will be more
likely to request, for a different engagement, a senior who
underreports than a senior who accurately reports
exceeding the budget. Table 2 presents ANOVA results
using participants’ responses to their likelihood of request-
ing the senior on a different engagement as the dependent
variable. While neither the main effect of client desirability
nor the interactive effect is signi?cant (p = 0.738 and
p = 0.744, respectively), results indicate a highly signi?cant
main effect of hours reported on this likelihood (p < 0.002,
one-tailed). The mean responses for the underreporting
and accurate reporting conditions are 7.92 and 6.82,
respectively. These results support H2.
Moderated mediation: The in?uence of performance
evaluation on staf?ng decisions
All else being equal, managers will typically prefer the
strongest staff and seniors for their engagements.
However, the expectations laid out in H1 and H2 suggest
a more complex relationship between performance evalu-
ations and staf?ng decisions. When taken together, our
manager hypotheses predict that client desirability
changes the in?uence reporting accuracy has on managers’
evaluations of their subordinates’ performance (H1), but
does not affect reporting accuracy’s in?uence on managers’
willingness to select a subordinate for a different future
engagement (H2). Thus, the combination of our H1 and
H2 predictions suggests a moderated mediation in which
the senior’s performance evaluation mediates the effect
of reporting accuracy on the likelihood of future engage-
ment appointment for a more desirable client, but report-
ing accuracy directly impacts this likelihood when the
client is less desirable (see Fig. 2). In other words, we
expect managers’ evaluations of the senior’s performance
to be more predictive of their willingness to select the
senior for a future engagement when client desirability is
high than when it is low. We use structural equation
modeling to test for this effect. We ?t the model using
multigroup path analysis with LISREL 8.8 software allow-
ing paths a and c
0
to vary across groups (model-?t
v
2
= 0.257).
10
As shown in Fig. 2 and Table 3, for the less desirable cli-
ent condition, the coef?cient for path a is not signi?cant
(p = 0.321) while the coef?cient for path c
0
is signi?cant
(p = 0.031), which implies no mediation of the direct effect
(i.e., no mediation of the direct path, c’, from reporting
accuracy to future engagement). That is, for the less desir-
able client, the effect of the senior’s reporting accuracy on a
manager’s likelihood of requesting the senior on a future
engagement does not go through the manager’s evaluation
of the senior’s performance. In contrast, in the higher desir-
ability condition, the coef?cient for path a is signi?cant
(p < 0.001) while the coef?cient for path c
0
is not signi?cant
(p = 0.172), which implies mediation. While the indirect
effect of the senior’s reporting accuracy on the likelihood
of requesting the senior on a future engagement (i.e., the
path from reporting accuracy to future engagement
through performance evaluation, path a  b) is not signi?-
cantly different than zero for the lower client desirability
condition, this effect accounts for 74.42 percent of the total
effect in the higher desirability condition. In other words,
when the agency con?ict is weaker (the less desirable cli-
ent condition), reporting accuracy does not affect the rela-
tionship between the senior’s performance evaluation and
future staf?ng likelihood. However, when the agency con-
?ict is stronger (the more desirable client condition),
reporting accuracy has an indirect effect on the likelihood
of the manager requesting the senior for a future engage-
ment, via the senior’s performance evaluation. Thus, we
Table 1
Managers’ evaluations of senior’s performance (Experiment 1).
Panel A: ANOVA results
Independent
variable
a
df Mean
square
F-Statistic p-
Value
b
Reporting
accuracy
1 6.509 4.664 0.033
Client
desirability
1 1.154 0.827 0.365
Interaction 1 12.722 9.115 0.003
Error 96 1.396
Total 100
Panel B: Mean (Standard Deviation)
c
Less
desirable
client
More
desirable
client
Row
means
Cell 1 Cell 2
Underreporter 7.50 (1.30) 8.00 (1.23) 7.75
[n = 26] [n = 25]
Cell 3 Cell 4
Accurate reporter 7.70 (1.03) 6.77 (1.15) 7.29
[n = 27] [n = 22]
Column
means
7.60 7.43
Panel C: Planned contrast for H1
Source of
variation
Contrast
estimate
Std. Error F-Statistic
d
p-
Value
Contrast (À1, +2,
+1, À2)
1.329 0.382 12.131 0.05); (2) RMSEA that is ‘‘close to’’ 0.09, or lower; and (3) CFI ‘‘close to’’
0.95, or higher. Further, we note that our results are robust to letting path b
vary, along with paths a and c
0
. We also compare our moderated mediation
model to the simple, non-moderated mediation model. A v
2
difference test
between the two nested models shows a reduction of minimal ?t v
2
from
23.216 for the simple model to 4.047 for our model (p < 0.01), indicating
our model’s superior ?t.
8 C.P. Agoglia et al. / Accounting, Organizations and Society 44 (2015) 1–14
conclude that client desirability moderates the mediating
effect of performance evaluation on the relationship
between time reporting accuracy and the likelihood that
the manager will request the senior for a future engage-
ment.
11
That is, when the client is less desirable, reporting
accuracy affects future staf?ng decisions directly (e.g., while
accurate reporters do not suffer from lower performance
evaluations, they do suffer from a longer-term narrowing
of their future engagement opportunities). However, when
the client is more desirable, reporting accuracy affects future
staf?ng decisions indirectly via the evaluation of the senior’s
performance (e.g., they suffer from both lower performance
evaluations and limited future engagement opportunities).
Experiment 2: Partners
Method
Participants, task, procedure, and design
Experiment 2 is designed to test our agency theory pre-
dictions that partners will have a longer-term focus and
prefer staff to accurately report going over budget rather
than underreport their hours worked. Experiment 2 utilizes
the same experimental instrument (including dependent
and independent variables) and procedures as Experiment
1. However, Experiment 2 employs partners as participants.
As with Experiment 1, we asked participants to assume
they are the senior’s immediate supervisor (i.e., the man-
ager on the engagement) and task them with evaluating
the senior’s performance. We asked partners to assume
the role of manager in order to elicit responses from part-
ners as to how they would behave if performing this man-
ager-assigned task. This allows us to test our application of
agency theory to underreporting by comparing the behav-
ior of agents (managers) to that of principals (partners) act-
ing in the role of agents.
12
In addition, this design choice
maintains a level of comparability of the task (i.e., evaluation
by an immediate supervisor). Finally, it circumvents the
dilemma of having to develop an equivalent manipulation
of client desirability for partners that similarly maintains
both the strength and salience of the manager client desir-
ability manipulation; yet, it manipulates no additional facets.
Participants were 119 audit partners.
13
Partners had, on
average, 25.8 years of audit experience. There are no signif-
icant differences (p’s > .20) for participants across experi-
mental conditions for any of the demographic measures
(e.g., overall audit experience, experience evaluating subor-
dinates, or ?rm type). Including the demographic measures
as covariates does not signi?cantly affect the results or alter
the inferences drawn.
Partner results
Manipulation checks
To test the manipulations in the partner experiment, we
use the same 11-point scales as in the manager experi-
ment. Partners’ responses were signi?cantly different
between conditions, and in the expected direction, for both
the reporting accuracy (means = 2.96 and 5.45 for the
underreporter and accurate reporter conditions, respec-
tively; t = 4.762, df = 114, p < 0.001, one-tailed) and client
desirability manipulations (means = 9.18 and 6.02 for the
more and less desirable client conditions, respectively;
p < 0.001, one-tailed).
14
Table 2
Managers’ likelihoodtorequest senior onfuture engagement (Experiment 1).
Panel A: ANOVA results
Independent
variable
a
df Mean square F-Statistic p-
Value
b
Reporting
accuracy (H2)
1 29.717 9.179 0.033
Client
desirability
1 0.366 0.113 0.738
Interaction 1 0.346 0.107 0.744
Error 94 3.237
Total 98
Panel B: Mean (Standard Deviation)
c
Less desirable
client
More
desirable
client
Row
means
Cell 1 Cell 2
Underreporter 7.80 (1.73) 8.04 (1.55) 7.92
[n = 25] [n = 24]
Cell 3 Cell 4
Accurate reporter 6.81 (2.22) 6.82 (1.53) 6.82
[n = 27] [n = 22]
Column means 7.29 7.46
a
Reporting Accuracy was manipulated such that the subordinate either
accurately reports exceeding budget or reports meeting budget (i.e.,
underreports). Client Desirability was manipulated such that the client
was either more desirable or less desirable to the participant.
b
Reported p-values are two-tailed.
c
Participants indicated their likelihood of requesting the senior on a
different future engagement on an eleven-point scale where 1 = ‘‘Very
Unlikely’’ and 11 = ‘‘Very Likely’’.
11
Results of analyses following Muller, Judd, and Yzerbyt (2005) also
suggest moderated mediation.
12
Because partners were previously managers, they are likely able to
easily place themselves back in the role of manager. Therefore, we designed
the study with the expectation that they will act as the senior’s immediate
supervisor responsible for evaluating the senior (i.e., manager), with the
exception that they are likely to bring a more layered perspective based on
their experiences and incentives as partners. In addition, we feel this design
choice is appropriate for our study because we are not interested in
drawing conclusions about higher-level evaluations a partner may be
responsible for with respect to the engagement team; instead, we are
interested in how they would evaluate staff if they, as principal, were to
perform this agent-assigned task.
13
We mailed instruments to 850 partners from a list provided by the
American Institute of Certi?ed Public Accountants (AICPA). We received
replies from 125 audit partners and 67 were returned as undeliverable. The
resulting response rate is 15.96%. There were no differences between early
and late respondents. There were six unusable responses resulting from a
failure to complete key dependent variables in the partner experiment. Our
conclusions remain the same with or without these individuals.
14
Again, reassuringly, the client desirability manipulation did not signif-
icantly in?uence responses to the reporting accuracy check question (main
and interactive effects of client desirability manipulation p’s > 0.550) and
the reporting accuracy manipulation did not signi?cantly in?uence
responses to the client desirability check question (main and interactive
effects of client desirability manipulation p’s > 0.230).
C.P. Agoglia et al. / Accounting, Organizations and Society 44 (2015) 1–14 9
Hypotheses 3 and 4
Table 4 presents the descriptive statistics and ANOVA
results for the performance evaluation variable used to test
H3. Hypothesis 3 predicts that partners will evaluate a
senior who accurately reports exceeding the budget higher
than one who underreports. However, the main effect of
reporting accuracy is not signi?cant (p = 0.395, one-tailed).
While partners’ reactions to underreporting appear to dif-
fer from managers’, they show no evidence of preferring
accurate reporters (mean = 7.47) to underreporters
(mean = 7.41). Similar to H3, H4 predicts that partners will
prefer to staff future engagements with accurate reporters
over underreporters. However, again, the main effect of
reporting accuracy is not signi?cant (p = 0.856, one-tailed),
as partners are no more likely to select an accurate repor-
ter than an underreporter (means = 7.20 and 7.53, respec-
tively; see Table 5). Thus, while partners do not exhibit
the same preference for underreporters that managers
do, they also do not appear to preferentially staff future
engagements with accurate reporters. These results are
not supportive of either H3 or H4.
To shed light on the inconsistency between these
?ndings and our agency theory expectations, we examined
the partner data more closely. Although partners’
responses to the manipulation check questions were sig-
ni?cantly different between conditions and this difference
was in the expected direction, it is still possible that
partners did not fully internalize our manipulations as
we intended. Thus, we examine whether our manipula-
tions were similarly effective for partners as they were
for managers.
15
We ?nd that, while comparisons of man-
agers’ and partners’ responses to the client desirability
manipulation check indicate that the manipulation was sim-
ilarly effective for managers and partners for both the higher
and lower desirability conditions (i.e., manager Cells 2 and 4
versus partner Cells 2 and 4, t = 0.578, df = 98, p = 0.565;
manager Cells 1 and 3 versus partner Cells 1 and 3,
t = 0.655, df = 114, p = 0.514), the manipulation check for
reporting accuracy tells a different story. A closer look
reveals that, relative to managers, partners were less likely
Fig. 2. Moderated mediation: path model. Note: Unstandardized coef?cients are provided. Coef?cients for variant paths are presented for the less desirable
client condition ?rst, followed by the more desirable client condition, separated by a slash (i.e., less/more desirable client coef?cients). Bolded path
coef?cients are signi?cant at p 6 0.05.
Table 3
Moderated Mediation: Decomposition of Effects (Experiment 1).
Direct effect
(path c
0
)
a
Std
Error
t-Statistic p-Value Indirect effect
(path a  b)
a
Std
Error
t-Statistic p-Value Total
effect
Less desirable client Unstandardized À0.5404 0.2348 À2.30 0.031 0.1166 0.1805 0.65 0.321 À0.4239
Standardized À0.2527 – – 0.0541 – – À0.1986
More desirable client Unstandardized À0.3711 0.2877 À1.29 0.172 À1.0522 0.2186 À4.81
While some research suggests that explicit incentives to meet time budgets have recently
been reduced at audit firms, there is also evidence indicating that audit seniors and staff
still feel at least implicit pressure to meet budgets. We examine the possibility that both
of these findings tell a part of the story. Specifically, we explore whether, and under what
conditions, seniors and staff are implicitly encouraged to underreport time through future
engagement staffing decisions and the performance evaluation process. Further, we consider
the extent to which agency theory can serve as a framework for understanding
how the incentives of audit managers and partners influence how they view underreporting
by their engagement staff.
Audit team time reporting: An agency theory perspective
q
Christopher P. Agoglia
a,?
, Richard C. Hat?eld
b
, Tamara A. Lambert
c
a
Department of Accounting, Isenberg School of Management, University of Massachusetts Amherst, 121 Presidents Drive, Amherst, MA 01003, United States
b
Culverhouse School of Accountancy, The University of Alabama, United States
c
Department of Accounting, College of Business and Economics, Lehigh University, United States
a r t i c l e i n f o
Article history:
Available online 16 May 2015
a b s t r a c t
While some research suggests that explicit incentives to meet time budgets have recently
been reduced at audit ?rms, there is also evidence indicating that audit seniors and staff
still feel at least implicit pressure to meet budgets. We examine the possibility that both
of these ?ndings tell a part of the story. Speci?cally, we explore whether, and under what
conditions, seniors and staff are implicitly encouraged to underreport time through future
engagement staf?ng decisions and the performance evaluation process. Further, we con-
sider the extent to which agency theory can serve as a framework for understanding
how the incentives of audit managers and partners in?uence how they view underreport-
ing by their engagement staff. We place participants in a scenario in which they are
responsible for evaluating an engagement senior who appears to have worked more hours
than were budgeted. We manipulate the senior’s reporting accuracy (underreporting versus
accurate reporting) and the desirability of the client (more versus less desirable). We ?nd
that, when managers’ agency-related incentives con?ict more strongly with those of the ?rm
(more desirable client), they tend to tacitly encourage underreporting through their evalua-
tions of the senior’s performance. Managers are also more likely to request an underreporter
on a future engagement. In contrast, partners placed in the same setting showno evidence of
encouraging underreporting. Thus, our results suggest that managers’ tacit encouragement
of underreporting is contrary to what the ‘‘principals’’ of the ?rm (i.e., partners) appear to
want. Further, while ?rms may have reduced their emphasis on formal, explicit incentives
to underreport, it appears likely that implicit manager incentives persist.
Ó 2015 Elsevier Ltd. All rights reserved.
Introduction
Our study examines the role agency incentives play in
diminishing the effectiveness of ?rm policies aimed at
reducing underreporting of time. Speci?cally, we explore
the extent to which, and under what circumstances, supe-
riors implicitly encourage such behavior in audit staff.
Underreporting occurs when an auditor does not record
all hours worked on a particular engagement and is
believed to negatively affect audit quality, to lead to other
unethical and dysfunctional audit behaviors that can
increase audit risk, and to result in incorrect information
being used in client pricing and retention decisions
(e.g., Donnelly, Quirin, & O’Bryan, 2003; Public Oversighthttp://dx.doi.org/10.1016/j.aos.2015.03.005
0361-3682/Ó 2015 Elsevier Ltd. All rights reserved.
q
We thank Mark Peecher (editor), two anonymous reviewers, Lindsay
Andiola, Ann Backof, Jean Bedard, Bradley Bennett, Bill Brown, Bob
Cornell, Steven DeSimone, Kirsten Fanning, Jace Garrett, Ryan Guggen-
mos, Denise Hanes, Lynn Hannan, Jennifer Joe, Steve Kachelmeier,
Kathryn Kadous, Elisa Lee, Ben Luippold, Chris Nolder, Mary Parlee, Peter
Roebuck, Aaron Saiewitz, Chad Stefaniak, Hun-Tong Tan, David Wood,
and workshop participants at Bentley University, Georgia State Univer-
sity, Oklahoma State University, the 2010 International Symposium on
Audit Research, the 2011 AAA Audit Mid-Year Meeting, and the 2011 AAA
Annual Meeting, for their helpful comments and assistance.
?
Corresponding author.
E-mail address: [email protected] (C.P. Agoglia).
Accounting, Organizations and Society 44 (2015) 1–14
Contents lists available at ScienceDirect
Accounting, Organizations and Society
j our nal homepage: www. el sevi er. com/ l ocat e/ aos
Board, 2000; Shapeero, Koh, & Kilough, 2003). As a conse-
quence, audit ?rm policies expressly prohibit underreport-
ing (McNair, 1991).
Although meeting time budgets has traditionally been a
signi?cant performance evaluation focus in audit ?rms
(Lightner, Adams, & Lightner, 1982; McNair, 1991), a sur-
vey of partners and managers conducted by Buchheit,
Pasewark, and Strawser (2003) suggests that such formal,
explicit emphasis on meeting budgets has ebbed.
However, there is also evidence indicating that audit
seniors still feel pressure to underreport (Sweeney &
Pierce, 2006). One possibility is that both of these ?ndings
tell a part of the story. That is, explicit pressure to meet
budget may in fact be reduced, while implicit pressure still
exists (e.g., seniors do not want to appear less valuable or
productive than peers who report meeting budget;
Sweeney & Pierce, 2006). Our study explores whether staff
are still implicitly encouraged to underreport time through
future engagement staf?ng decisions and the performance
evaluation process and, if so, the extent to which agency
theory can serve as a framework for understanding how
incentives perpetuate the behavior, as well as a basis for
creating potential solutions.
Audit managers and partners have, at times, differing
incentives with respect to the time reporting behavior of
the engagement team. Agency theory predicts that
partners, as ‘‘principals’’ of the ?rms, have a longer-term
perspective and, thus, believe that their interests align clo-
sely with those of the ?rm as a whole (e.g., better retention
of good, honest employees who are reluctant to misreport
hours worked; more accurate costing ?gures for better cli-
ent acceptance/retention decisions and better long-run
alignment of resources). As a result, partners are likely to
prefer that their engagement teams report their time accu-
rately. However, managers are likely more in?uenced by
shorter-term incentives to complete the engagement
within the budgeted time (e.g., to avoid fee pressure on
desirable clients; to impress partners with good realization
rates) (Sweeney & Pierce, 2006). Managers also spend more
time with staff at the worksite than do partners, resulting
in managers having more accurate information regarding
staff hours (Otley & Pierce, 1996). This informational
advantage provides managers with an opportunity to act
on their incentive to implicitly encourage engagement
team underreporting. We, therefore, explore how the dif-
fering roles (and related incentive structures) of audit
managers and partners in?uence how they view the
practice of underreporting, anticipating that managers
and partners will act differently, but in a manner consis-
tent with agency theory predictions (i.e., as agents and
principals, respectively).
Agency theory predicts that, where there is strong con-
?ict between the incentives of the principal and agent, the
agent will tend to act in his/her self-interest when pro-
vided the opportunity (e.g., an informational advantage
over the principal). If no strong con?ict exists, then the
agent’s actions will tend to align more with the principal’s
incentives (Jensen & Meckling, 1976). Thus, if an agency
framework applies in the underreporting context, one
would expect that the strength of managers’ (con?icting)
incentives in?uences their tendency to encourage the
behavior. One factor that affects the strength of such a con-
?ict relates to desirability of the client. For example, if a
manager has a strong preference for a particular client
(e.g., the client is close to home or the manager gets along
well with client management), he or she is more likely to
have a stronger desire for a subordinate to underreport in
order to maintain audit fees near their current levels to
help promote client retention and, in turn, increase the
likelihood he/she remains assigned to this desirable client.
In contrast, there are generally other engagements that a
manager ?nds less attractive and is less interested in
retaining. On such less desirable clients, the strength of
the agency con?ict is lessened, and the manager is more
likely to be more accepting of budget overruns.
We conduct two experiments to investigate the in?u-
ence agency incentives have on underreporting. In sepa-
rate manager and partner experiments, we place
participants in a scenario in which an engagement senior
and her staff appear to have worked more hours than were
budgeted. We ask participants to assume they are the
senior’s immediate supervisor (i.e., the manager on the
engagement) and task them with evaluating the senior’s
performance. We manipulate reporting accuracy; that is,
whether staff underreport (i.e., report meeting the budget)
or report all the hours worked (i.e., report exceeding the
budget). We also manipulate client desirability (i.e., more
desirable versus less desirable).
Our results indicate that, through their evaluation of
staff who exceed budget, managers are more likely to tac-
itly encourage underreporting (relative to accurately
reporting exceeding budget) when client desirability is
high. When the client is less desirable, managers’ prefer-
ence for underreporters dissipates (i.e., their preferences
begin to reverse). These results are consistent with agency
theory expectations, as managers behave more like agents,
acting in their own interest when their incentives con?ict
with the ?rm’s, but acting more in the ?rm’s interest when
there is no strong con?ict between their incentives and the
?rm’s. We also ?nd that managers are more likely to
request an underreporter on a future engagement.
Further, we ?nd that reporting accuracy in?uences man-
agers’ future staf?ng decisions through its effect on staff
evaluations; however, this mediating relationship is mod-
erated by client desirability (i.e., moderated mediation).
Speci?cally, managers’ evaluations of the senior’s perfor-
mance are more predictive of their willingness to select
the senior for a future engagement when client desirability
is high than when it is low.
In contrast, partners who assume the role of the senior’s
immediate supervisor (i.e., manager) show no preference
for the underreporter, on average, either through their
evaluations of the senior or their future staf?ng
preferences. However, our results are not supportive of
our predictions that partners will react to underreporting
as ?rm guidance suggests. That is, partners in the accurate
reporting condition do not evaluate the senior signi?cantly
higher, nor are they signi?cantly more likely to request the
senior for a different engagement, than those in the under-
reporting condition.
To explore this difference between our expected results
and our observed results for partners, we examine the
2 C.P. Agoglia et al. / Accounting, Organizations and Society 44 (2015) 1–14
partner data more closely. Although partners’ manipula-
tion check responses for reporting accuracy were signi?-
cantly different by condition, we ?nd that, unlike
managers, many partners in the accurate reporting condi-
tion were not convinced that the senior accurately
reported all hours worked. In other words, there were a
number of partners who believed the accurate reporter
was likely underreporting, at least to some extent, mean-
ing that they did not see the senior as an accurate reporter.
As we are interested in whether partners prefer staff they
view as accurate reporters to staff they suspect are under-
reporters, we perform additional analyses examining the
subset of partners who were more convinced that the
accurate reporter was, in fact, reporting accurately.
1
We
?nd that, more in line with an agency framework, these
partners do prefer seniors who go over budget to accurately
report their hours. As anticipated, client desirability has no
effect on their evaluations of the senior. Further, we conduct
additional analyses that, again, utilize the full partner data
set. These additional analyses, which use partners’ beliefs
about the likelihood that the senior recorded all hours
worked (a scaled, measured variable), support the results
obtained with the subset. That is, we observe a positive asso-
ciation between partners’ perceptions about the senior’s
reporting accuracy and partner’s preference for the senior.
Our study contributes to the literature in a number of
important ways. First, our study advances the literature
on underreporting of time by taking a ?rst step toward rec-
onciling the ?ndings of prior studies that suggest both: (a)
reduced emphasis on meeting time budgets as a formal
performance metric (Buchheit et al., 2003) and (b) audit
seniors still feel pressure to underreport (Sweeney &
Pierce, 2006). We present evidence that managers still tac-
itly encourage underreporting by their engagement team,
even in a setting without explicit mention of time budgets
as a performance metric. Thus, while explicit incentives
may have reduced, it appears likely that implicit manager
incentives to underreport persist. Second, our results sug-
gest that managers’ tacit encouragement of underreporting
is contrary to what the ‘‘principals’’ of the ?rm appear to
want. That is, unlike managers, partners placed in the same
setting show no evidence of encouraging underreporting,
as our results suggest that partners do not prefer the
underreporting senior to the accurate reporter. Further,
we ?nd a positive association between partners’ beliefs
about how likely it is the senior reported all hours worked
and their preference for the senior.
Third, our agency framework perspective on the phe-
nomenon of underreporting suggests that agency theory
can assist researchers, regulators, and practitioners wish-
ing to understand and curb the behavior. Existing research
on principal-agent con?ict (e.g., in the context of project
continuation decisions) offers interesting avenues to
explore fresh solutions to the problem (e.g., establishing
appropriate ‘‘principal-oriented’’ incentives; incurring
monitoring costs) (Jensen & Meckling, 1976). Finally, prac-
titioners should be interested in our ?nding that manager
approval of underreporting is not uniform across audits.
In other words, client pro?tability is not uniformly affected
by underreporting. This suggests there are signi?cant costing
issues for ?rms, as the level of underreporting varies with
contextual/situational factors and the strength of agency-
related incentives in play for the manager. Thus, ?rms may
want to consider agency-related approaches to better align
audit manager incentives with long-term ?rm objectives.
The remainder of the paper is organized as follows. The
next section discusses previous literature and develops our
hypotheses. The third and fourth sections detail the
method and results for Experiments 1 and 2, respectively.
The ?nal section offers conclusions, limitations, and sug-
gestions for future research.
Background and hypothesis development
Underreporting the amount of time spent on an audit
engagement has long been labeled an unethical practice
with dysfunctional downstream effects that threaten audit
quality (e.g., it can lead to unrealistic future time budgets,
which in turn could lead to premature sign-off or docu-
mentation of work not performed) (Ponemon, 1992;
Sweeney & Pierce, 2006). Nevertheless, early survey
research indicated that auditors of all ranks believed that
underreporting resulted in better performance evaluations
and promotion opportunities (Lightner et al., 1982;
McNair, 1991). Such beliefs were likely warranted, as evi-
denced by one partner’s description of realization rates
as having been the ‘‘diamond of metrics’’ with respect to
performance evaluations (Buchheit et al., 2003).
2
However, there is more recent evidence that ?rms have
reduced emphasis on formal time budget metrics. In a sur-
vey of partners and managers, Buchheit et al. (2003) report
a reduction in budget pressure, primarily due to a reduced
emphasis on time budgets within the ?rm’s performance
evaluation system. Similarly, Sweeney and Pierce (2004,
2006) report evidence of reduced emphasis on time bud-
gets at audit ?rms, particularly for higher-ranking auditors.
While formal emphasis on meeting time budgets appears
to have been reduced, Sweeney and Pierce (2006) point
out that it has not been eliminated as a formal, explicit
metric of performance and still is a factor, albeit dimin-
ished, in promotion decisions. Further, their results suggest
that auditor perceptions of the pressure to underreport
vary by rank; staff and seniors still report that managers
often implicitly encourage the practice (Sweeney &
Pierce, 2006).
3
It is unclear then whether, in the current
audit environment, supervisors actually will still encourage
1
Focusing on partners in the accurate reporting condition who viewed
the senior as reporting accurately results in a subset of partners that is
equally as likely as our manager participants to view the accurate reporter
as reporting. We discuss this further in the Experiment 2 Partner Results
section.
2
Most audits are ?xed fee engagements in which additional hours
worked are not billed to the client (Buchheit et al., 2003). Thus, budget
overruns primarily impact auditors’ realization rates and future client
retention decisions (Akers & Eaton, 2003; McNair, 1991; Shapeero et al.,
2003), which in turn have traditionally affected performance evaluations
and promotions (Buchheit et al., 2003).
3
For a sample of more recent anecdotal evidence, see:http://www.
stuffaccountantslike.com/?p=84;http://steeplemedia.com/blogs/krupo/
archive/2009/06/04/eating-hours-is-not-cool.aspx; andhttp://www.ac-
countantbyday.com/2011/05/06/life-of-an-auditor-billable-hours/.
C.P. Agoglia et al. / Accounting, Organizations and Society 44 (2015) 1–14 3
underreporting by their engagement teams. One possibility
is that both of these ?ndings hold part of the answer: expli-
cit pressure to meet budget may in fact be reduced, but
implicit pressure may still exist (e.g., managers and seniors
do not want to appear less valuable or productive than peers
who report meeting budget; Sweeney & Pierce, 2006).
If encouragement of underreporting persists, it would
likely manifest in two in?uential ways. First, managers
might differentially evaluate staff performance based on
perceived underreporting (Otley & Pierce, 1996; Sweeney
& Pierce, 2006). For example, when meeting budget is cru-
cial to the manager, he/she has an incentive to express
(through performance evaluations) approval of staff who
appear to have underreported to meet budget, and disap-
proval of staff who report going over budget. Second, man-
agers may be more or less likely to choose staff for future
engagements based on their willingness to underreport
(Otley & Pierce, 1996; Sweeney & Pierce, 2006). That is, if
managers have similar budgetary concerns about
subsequent engagements, they are more likely to prefer
to utilize audit staff who appear willing to help them meet
future budgets as well. It is important to note that these
two decisions are key determinants of the career progres-
sion of audit staff.
Underreporting and the agency problem
An agency problem exists in all organizations and coop-
erative efforts in which a principal (e.g., an owner) utilizes
an agent (e.g., a manager) for the purpose of delegating
responsibility to him or her (Jensen & Meckling, 1976).
Agency theory suggests that, as long as the goals of the
principal and agent are aligned, the agent will make deci-
sions that maximize the goals of the principal. However,
when the goals of the principal and agent diverge and
the agent has the opportunity to act in his/her self-interest
(i.e., has relevant information that the principal lacks,
referred to as ‘‘information asymmetry’’), agency theory
predicts that the agent will make decisions that maximize
his/her self-interest over the principal’s interests (Booth &
Schulz, 2004; Jensen & Meckling, 1976). Prior research sup-
ports the notion that agents tend to act in their own inter-
est when they have both the incentive and opportunity to
do so. For example, agents have been shown to continue
failing projects in order to avoid negative reputational
effects and to manage earnings to achieve higher bonuses
(e.g., Beaudoin, Agoglia, Tsakumis, & Guggenmos, 2015;
Booth & Schulz, 2004; Guggenmos & Agoglia, 2015;
Jensen & Meckling, 1976). In this study, we employ an
agency framework to better understand the phenomenon
of underreporting of time and why current approaches
are unlikely to eliminate this behavior.
On an audit engagement, an audit manager acts as an
agent of the audit ?rm which is controlled, or ‘‘owned’’,
by the audit partners. The manager generally has a more
‘‘hands on’’ role in the day-to-day activities of the engage-
ment than the partner (Wolf, 1981). Because of this, there
is likely to be information asymmetry between the partner
and the manager regarding certain aspects of the audit (i.e.,
opportunity for the manager to act in his/her own interest).
For example, managers likely have a better idea of how
many hours their staff actually work, given that they typi-
cally spend more time on-site and interact more regularly
with the engagement team than partners (Gibbins &
Trotman, 2002). Thus, this informational advantage leaves
the manager with a dilemma: the manager can encourage
the provision of accurate information to achieve a positive
collective outcome for the ?rm or, if agency incentives
persist,
4
encourage strategic misrepresentation of the hours
worked by the engagement team in order to secure a
positive outcome for him/herself. The dilemma is less
straightforward in this context than in a typical agency set-
ting, as audit managers generally aspire to be partners (i.e.,
audit managers are somewhat ‘‘nontraditional’’ agents).
That is, as ‘‘principals in training’’, audit managers could
tend toward the behavior of a principal.
Agency con?ict and manager incentives for evaluating the
performance of underreporters and accurate reporters
Resource constraints have intensi?ed under the
Sarbanes-Oxley Act (SOX) of 2002, putting pressure on
audit fees and making retention of marginally pro?table
clients even more tenuous (Rama & Read, 2006). Given that
low realization rates lead to an increased chance of sever-
ing the relationship with a client (either due to the ?rm
severing the relationship for pro?tability reasons or the
client severing the relationship as a result of the ?rm rais-
ing fees), managers have an incentive to tacitly encourage
underreporting in order to avoid losing an attractive client.
A manager will ?nd some clients more desirable than
others due to various situational and contextual factors
that are inherent to the client. For example, a manager is
likely to prefer: a client that is convenient for the man-
ager’s work schedule, a client that is easy to get to, client
management that the manager gets along particularly well
with personally, a client that is within the manager’s
industry of interest, or an engagement that involves an
in?uential partner at the manager’s of?ce.
Since a manager working on a client he or she ?nds
more desirable will be more concerned about losing that
client, the agency con?ict (i.e., misalignment of their incen-
tives with those of the ?rm) faced by the manager will be
more pronounced (i.e., appropriately costing a less prof-
itable client versus taking action to obscure the cost of a
client the manager ?nds desirable). That is, the manager
will have a heightened sensitivity to budget overruns that
would re?ect poorly on his/her management of the
engagement or that could eventually lead to a severing of
the relationship with that client. In these circumstances,
the manager has greater incentive to tacitly encourage
strategic misrepresentation of the hours worked on the
engagement and, in turn, to evaluate underreporters sig-
ni?cantly higher than honest reporters. On the other hand,
4
As noted, such manager incentives could be implicit (e.g., to avoid fee
pressure to help with retention of desirable clients; impress partners with
good realization rates) and/or explicit (e.g., formal metrics used for
manager performance evaluations and promotion decisions). The ?nding
that audit staff report still feeling time budget pressure from managers is
consistent with the notion that such agency-related incentives persist for
managers (Sweeney & Pierce, 2006).
4 C.P. Agoglia et al. / Accounting, Organizations and Society 44 (2015) 1–14
a manager’s agency con?ict is reduced when the client is
less desirable. With a less desirable client, the manager
has less incentive to mask the true cost of the client to
the ?rm, and the strength of the principal-agent con?ict
is reduced. Without strong con?ict, we expect managers
to begin to act more in line with ?rm policy (i.e., ?rm guid-
ance would suggest evaluating accurate reporters higher
than underreporters).
Thus, consistent with agency theory, we expect man-
agers to act more in their own interest when con?ict
between their incentives and the ?rm’s is heightened, but
more in the ?rm’s interest when there is no strong con?ict
between their incentives and the ?rm’s. Speci?cally, we
expect managers to tacitly encourage underreporting by
preferring underreporters to accurate reporters when they
?nd the client more desirable (i.e., we expect a larger dis-
parity in managers’ evaluations of underreporters and
accurate reporters when staff exceed budget for a desirable
client). When managers ?nd the client less desirable, their
preference for underreporters will dissipate and likely
approach the behavior prescribed by ?rm guidance.
5
That
is, relative to when managers ?nd the client more desirable,
managers will tend to evaluate honest reporters marginally
higher than underreporters, better re?ecting the interests of
their principals. Given these expectations, we propose a
disordinal interaction of the form speci?ed below
(see Fig. 1):
H1. There will be a disordinal interaction between staff
reporting accuracy and client desirability such that man-
agers’ evaluations of staff will be highest for underre-
porters on a more desirable client, and then decrease in the
following order: accurate reporters on a less desirable
client, underreporters on a less desirable client, and
accurate reporters on a more desirable client.
Managers’ future staf?ng decisions: implications for accurate
reporters
The choice of particular team members for an engage-
ment can have signi?cant time budget implications for
the audit manager. Suppose, for example, that a manager
has worked with two audit seniors on past engagements.
Assume one of the seniors has, on occasion, missed bud-
gets and the other has always met his/her budgets
(although the manager suspects he/she is underreporting).
If, in an effort to protect realization rates on future engage-
ments, the manager is more likely to select the underre-
porter than the accurate reporter, there are potentially
serious negative implications for the accurate reporter.
For example, the accurate reporter is less likely to be
assigned to desirable engagements in the future which, in
turn, can in?uence raises, promotions, and continued
employment (Doby & Caplan, 1995). As a result, the ?rm
is more likely to lose honest, capable employees in the
process.
The choice of working with particular seniors or staff on
future engagements presents a different set of dilemmas to
managers than evaluating their performance on a com-
pleted engagement. That is, past performance evaluations
may not relate directly to a manager’s desire to select that
individual for a different future engagement (Kaplan &
Reckers, 1985). For instance, even if a manager evaluates
an accurate reporter who exceeds budget favorably (partic-
ularly with a non-preferred client), he/she is still likely to
be less inclined to select that staff member for future
engagements, as a reluctance to underreport could hurt
the manager’s chances of meeting future budgets. Having
staff willing to underreport provides the manager with dis-
cretion/?exibility to meet budget on future engagements.
6
Thus, on average, we expect managers to prefer underre-
porters (relative to accurate reporters who have gone over
budget) on future engagements regardless of the desirability
of the current client. We, therefore, propose the following
main effect hypothesis:
H2. Audit managers will be more likely to request a senior
for a different engagement when the senior underreports
than when the senior accurately reports exceeding the
budget.
Partners’ interests
Our discussion of agency theory suggests that managers
will tend to act in a utility maximizing manner by tacitly
encouraging underreporting when contextual factors lead
to a misalignment of their incentives from those of their
?rms. However, agency theory implies that partners will
react differently to underreporting than managers. As
owners of the ?rm, their interests are more aligned with
the ?rm’s interests. And, given ?rm policies that explicitly
prohibit underreporting due to the wide variety of prob-
lems induced by the practice and by the related unrealistic
time budgets that result from past underreporting (e.g.,
inappropriate costing of less pro?table clients, failure to
gather suf?cient evidence/documentation of procedures
not performed, and lower employee morale and increased
5
The phenomenological strength of our client desirability manipulation
was not intended to result in the absence of an incentive (i.e., complete
removal of con?icting agency incentives), since a less desirable client is still
part of a manager’s portfolio of clients. Thus, while we expect managers’
preferences for underreporters to reverse (and begin to approach ?rm
preferences) for the less desirable client condition in our study, we do not
expect the disparity between managers’ evaluations of underreporters and
accurate reporters to be as large for less desirable clients as for more
desirable clients (i.e., we anticipate a disordinal interaction of the form
represented in Fig. 1).
6
Alternatively, a manager could evaluate the underreporter’s perfor-
mance higher on the current engagement, particularly if the client is one
the manager ?nds desirable, but not necessarily reward this staff/senior
with selection for a future engagement. That is, the manager could view the
underreporter’s performance more favorably because on this particular
engagement time needs to be eaten in order to meet budget, while on an
undetermined future audit it may be less necessary. However, agency
theory (which provides the theoretical underpinning for our hypotheses)
suggests that managers might prefer the budgeting ‘‘?exibility’’ of having
staff (on an undetermined future engagement) who are willing to
underreport if necessary. That is, agency theory predicts that an agent will
act in his/her own interest when there is both incentive (i.e., goal con?ict
with the principal) and opportunity to act in his/her interest at the expense
of the principal. Having a preference for underreporters helps allow a
manager to take advantage of these opportunities when such incentives
exist.
C.P. Agoglia et al. / Accounting, Organizations and Society 44 (2015) 1–14 5
turnover), partners should prefer that their engagement
teams report their time accurately.
Relative to managers, partners’ role and incentives will
typically cause them to have a longer-term perspective
when it comes to these negative audit quality-related ram-
i?cations of underreporting, as well as a greater concern
for obtaining a true picture of the cost of performing a par-
ticular audit. In contrast to managers (but consistent with
agency theory), when partners are asked to place them-
selves back in the role of an engagement senior’s immedi-
ate supervisor (i.e., the audit manager) and consider the
evaluation and staf?ng decisions the supervisor is respon-
sible for, we expect them to generally prefer staff to accu-
rately report hours when they exceed budget, as opposed
to staff underreporting their hours (i.e., to evaluate higher
those staff whom they believe have accurately reported
and to select them for future engagements). Therefore,
we pose the following hypotheses:
H3. Audit partners will evaluate a senior who accurately
reports exceeding the budget higher than one who under-
reports.
H4. Audit partners will be more likely to request a senior
for a different engagement when the senior accurately
reports exceeding the budget than when the senior
underreports.
Experiment 1: Managers
Method
Participants, task, and procedure
We conduct our ?rst experiment to investigate whether
audit managers impose an implicit pressure to underreport
time and, if so, whether their desire to retain the client
in?uences this pressure. We placed manager participants
in a scenario in which an engagement senior and her staff
appear to have worked more hours than were budgeted.
We asked participants to assume they are the senior’s
immediate supervisor (i.e., the manager on the engage-
ment) and tasked them with evaluating the senior’s
performance.
Manager participants were given background informa-
tion on the client and the engagement, along with current
year information on the audit and engagement team. They
were told that a clean opinion was rendered on the client
after some adjustments to the ?nancial statements were
made. Information provided to participants relating to
the senior on the engagement included generally positive
information. However, an ‘‘area for improvement’’ (i.e., a
mildly negative piece of information) was provided in
order to generate variability in participants’ responses
and to reduce the likelihood of a ceiling effect.
Speci?cally, each participant was told: the senior, Lisa
Martin, was a new member of the engagement team this
year; this was their ?rst time working with her; she
seemed to have a good rapport with the client and her
staff; she kept them informed of unusual items; she made
very few documentation errors; and she occasionally con-
tacted them with questions regarding proper ?rm proce-
dure which they hope on future engagements she will be
more prepared to apply on her own. Participants were then
given more detailed information about the engagement
and responded to questions about their perceptions of
the senior’s performance. Finally, the managers answered
a series of case-related and demographic questions, includ-
ing manipulation checks.
Participants in Experiment 1 were 100 audit managers.
7
Managers had, on average, 9.7 years of audit experience.
There are no signi?cant differences (p’s > .20) for
participants across experimental conditions for any of the
demographic measures (e.g., overall audit experience, expe-
rience evaluating subordinates, ?rm type). Including the
demographic measures as covariates does not signi?cantly
affect the results or alter the inferences drawn.
Less Desirable Client More Desirable Client
P
e
r
f
o
r
m
a
n
c
e
E
v
a
l
u
a
t
i
o
n
Underreporter
Accurate Reporter
Fig. 1. H1 Prediction: managers’ evaluations of senior’s performance.
7
We mailed instruments to 900 managers from a list provided by the
American Institute of Certi?ed Public Accountants (AICPA). We received
replies from 109 respondents and 45 were returned as undeliverable. The
resulting response rate is 12.75%. There were no differences between early
and late respondents. There were nine unusable responses resulting from
completion by inappropriately classi?ed individuals such as staff accoun-
tants, payroll clerks, and tax accountants. Our conclusions remain the same
with or without these individuals.
6 C.P. Agoglia et al. / Accounting, Organizations and Society 44 (2015) 1–14
Independent variables
We utilized a 2 Â 2 (reporting accuracy by client desir-
ability) between-participants design. We manipulated
reporting accuracy as either underreporting or accurate
reporting of hours worked by the engagement team. In
both cases, there are indications that the engagement team
worked more hours than budgeted. In order to establish
this, participants were told that they felt con?dent the
team worked beyond the budgeted hours throughout the
engagement and that engagement team members often
left voicemails and submitted electronic workpapers for
review before 8 a.m. and after 7 p.m. In the ‘‘underreport’’
condition, participants were told, ‘‘Although Lisa and the
rest of the engagement staff appeared to work long hours,
their reported hours did not exceed their time budget.’’
Because it had been previously established in both condi-
tions that the engagement team worked more hours than
budgeted, the participants were led to infer that Lisa and
the engagement team underreported their hours in the ‘‘un-
derreport’’ condition. In the ‘‘accurate reporting’’ condition,
participants were told that, ‘‘Lisa and the rest of the engage-
ment staff appeared to work long hours and their reported
hours signi?cantly exceeded their time budget (i.e., it
appears that all excess hours worked were reported).’’
For the second independent variable, the client was either
moreor less desirable. We manipulatedseveral aspects of the
engagement since a particular individual may value certain
dimensions more than others, depending on their lifestyles
and priorities. For example, participants in the more (less)
desirable client condition were informed: they ‘‘have always
(never) gotten along particularly well with client manage-
ment on a personal level’’ and that ‘‘the recent audit was no
exception’’; this engagement does not come (comes) at ahectic
time in their personal work schedules; the client is located
near (far) from their home with little (heavy) traf?c on the
commute; the partner on the engagement ‘‘is a very experi-
enced (newjunior) partner inyour of?ce’’ who has some ‘‘high
(low) pro?le engagements that you would be interested (have
little interest) inworkingoninthe future’’ (italics not included
in experimental instrument).
Dependent variables
To test H1, we examine managers’ evaluations of the
audit senior’s performance. Managers were asked to rate
the senior’s performance with respect to this engagement.
They recorded their responses for these measures on ele-
ven-point scales, where 1 = ‘‘Poor’’ and 11 = ‘‘Outstanding’’.
For H2, we asked howlikely they were to request the senior
on a different engagement in the future. Responses were
again recorded on an eleven-point scale, where 1 = ‘‘Very
Unlikely’’ and 11 = ‘‘Very Likely’’.
Manager results
Results relating to our manager hypotheses are analyzed
within a 2 Â 2 ANOVA framework (reporting accuracy by
client desirability), with participants’ evaluations of seniors
and their likelihood of requesting the senior on a different
future engagement serving as dependent variables.
Reported p-values are two-tailed unless otherwise noted.
Manipulation checks
Manipulation checks for both independent variables indi-
cate that managers understoodthe manipulations. Managers
were asked how likely they feel it is that all of the hours
worked on the engagement were recorded. They indicated
their responses on an 11-point scale where 1 = ‘‘Very
Unlikely’’ and 11 = ‘‘Very Likely’’. Managers in the underre-
porting condition reported an average response of 3.94 com-
pared to a mean of 8.67 in the accurate reporting condition
(t = 10.722, df = 94, p < 0.001, one-tailed). Managers were
also asked about the strength of their personal preference
to retain the client on an 11-point scale in which 1 = ‘‘Very
Weak’’ and 11 = ‘‘Very Strong’’. Those in the higher client
desirability condition reported an average response of 9.38
compared to 5.73 for managers in the lower desirability con-
dition (t = 8.686, df = 93, p < 0.001, one-tailed).
8
Hypotheses 1 and 2
Hypothesis 1 speci?es a disordinal interactive effect of
the two independent variables on managers’ evaluation of
the senior’s performance. That is, when client desirability
is high, managers’ incentives are in con?ict with the ?rm’s,
and we expect the difference between managers’ evalua-
tions of underreporters and accurate reporters to be large
(i.e., underreporters’ evaluations will be substantially
higher than those of accurate reporters). In contrast, when
the client is less desirable, there is less con?ict between
managers’ incentives and the ?rm’s, and we expect man-
agers to begin to approach the evaluation of the senior as
?rm guidance suggests and their preference for underre-
porters to begin to reverse (i.e., accurate reporters’ evalua-
tions should be somewhat higher than underreporters’).
Given the interaction speci?ed by H1 and depicted in
Fig. 1, we test the hypothesis with a planned contrast using
the following weights: À1 for the underreporter/less desir-
able client condition (Cell 1), +2 for the underreporter/more
desirable client condition (Cell 2), +1 for the accurate repor-
ter/less desirable client condition (Cell 3), and À2 for the
accurate reporter/more desirable client condition (Cell 4).
Table 1 presents the descriptive statistics and results from
the contrast used to test H1, along with results of the tradi-
tional ANOVA, for the senior’s overall performance evalua-
tion measure. Cell means are in a pattern consistent with
the contrast, with Cell 2 highest, followed by Cells 3, 1,
and 4 (means = 8.00, 7.70, 7.50, and 6.77, respectively).
The planned contrast is signi?cant at p < 0.001.
9
Further,
the ANOVA interaction term is signi?cant at p = 0.003.
These results support H1.
8
We note that, reassuringly, the client desirability manipulation did not
signi?cantly in?uence responses to the reporting accuracy check question
(main and interactive effects of client desirability manipulation p’s > 0.240)
and the reporting accuracy manipulation did not signi?cantly in?uence
responses to the client desirability check question (main and interactive
effects of client desirability manipulation p’s > 0.520).
9
A Jonckheere-Terpstra test con?rms the predicted ordering of cells
(p < 0.001, one-tailed). Further, all simple comparisons between cells are in
the expected direction, and all are signi?cant at p < 0.07 (one-tailed) except
for the comparisons between Cells 1 and 3 (p = 0.266) and Cells 2 and 3
(p = 0.176).
C.P. Agoglia et al. / Accounting, Organizations and Society 44 (2015) 1–14 7
Hypothesis 2 predicts that audit managers will be more
likely to request, for a different engagement, a senior who
underreports than a senior who accurately reports
exceeding the budget. Table 2 presents ANOVA results
using participants’ responses to their likelihood of request-
ing the senior on a different engagement as the dependent
variable. While neither the main effect of client desirability
nor the interactive effect is signi?cant (p = 0.738 and
p = 0.744, respectively), results indicate a highly signi?cant
main effect of hours reported on this likelihood (p < 0.002,
one-tailed). The mean responses for the underreporting
and accurate reporting conditions are 7.92 and 6.82,
respectively. These results support H2.
Moderated mediation: The in?uence of performance
evaluation on staf?ng decisions
All else being equal, managers will typically prefer the
strongest staff and seniors for their engagements.
However, the expectations laid out in H1 and H2 suggest
a more complex relationship between performance evalu-
ations and staf?ng decisions. When taken together, our
manager hypotheses predict that client desirability
changes the in?uence reporting accuracy has on managers’
evaluations of their subordinates’ performance (H1), but
does not affect reporting accuracy’s in?uence on managers’
willingness to select a subordinate for a different future
engagement (H2). Thus, the combination of our H1 and
H2 predictions suggests a moderated mediation in which
the senior’s performance evaluation mediates the effect
of reporting accuracy on the likelihood of future engage-
ment appointment for a more desirable client, but report-
ing accuracy directly impacts this likelihood when the
client is less desirable (see Fig. 2). In other words, we
expect managers’ evaluations of the senior’s performance
to be more predictive of their willingness to select the
senior for a future engagement when client desirability is
high than when it is low. We use structural equation
modeling to test for this effect. We ?t the model using
multigroup path analysis with LISREL 8.8 software allow-
ing paths a and c
0
to vary across groups (model-?t
v
2
= 0.257).
10
As shown in Fig. 2 and Table 3, for the less desirable cli-
ent condition, the coef?cient for path a is not signi?cant
(p = 0.321) while the coef?cient for path c
0
is signi?cant
(p = 0.031), which implies no mediation of the direct effect
(i.e., no mediation of the direct path, c’, from reporting
accuracy to future engagement). That is, for the less desir-
able client, the effect of the senior’s reporting accuracy on a
manager’s likelihood of requesting the senior on a future
engagement does not go through the manager’s evaluation
of the senior’s performance. In contrast, in the higher desir-
ability condition, the coef?cient for path a is signi?cant
(p < 0.001) while the coef?cient for path c
0
is not signi?cant
(p = 0.172), which implies mediation. While the indirect
effect of the senior’s reporting accuracy on the likelihood
of requesting the senior on a future engagement (i.e., the
path from reporting accuracy to future engagement
through performance evaluation, path a  b) is not signi?-
cantly different than zero for the lower client desirability
condition, this effect accounts for 74.42 percent of the total
effect in the higher desirability condition. In other words,
when the agency con?ict is weaker (the less desirable cli-
ent condition), reporting accuracy does not affect the rela-
tionship between the senior’s performance evaluation and
future staf?ng likelihood. However, when the agency con-
?ict is stronger (the more desirable client condition),
reporting accuracy has an indirect effect on the likelihood
of the manager requesting the senior for a future engage-
ment, via the senior’s performance evaluation. Thus, we
Table 1
Managers’ evaluations of senior’s performance (Experiment 1).
Panel A: ANOVA results
Independent
variable
a
df Mean
square
F-Statistic p-
Value
b
Reporting
accuracy
1 6.509 4.664 0.033
Client
desirability
1 1.154 0.827 0.365
Interaction 1 12.722 9.115 0.003
Error 96 1.396
Total 100
Panel B: Mean (Standard Deviation)
c
Less
desirable
client
More
desirable
client
Row
means
Cell 1 Cell 2
Underreporter 7.50 (1.30) 8.00 (1.23) 7.75
[n = 26] [n = 25]
Cell 3 Cell 4
Accurate reporter 7.70 (1.03) 6.77 (1.15) 7.29
[n = 27] [n = 22]
Column
means
7.60 7.43
Panel C: Planned contrast for H1
Source of
variation
Contrast
estimate
Std. Error F-Statistic
d
p-
Value
Contrast (À1, +2,
+1, À2)
1.329 0.382 12.131 0.05); (2) RMSEA that is ‘‘close to’’ 0.09, or lower; and (3) CFI ‘‘close to’’
0.95, or higher. Further, we note that our results are robust to letting path b
vary, along with paths a and c
0
. We also compare our moderated mediation
model to the simple, non-moderated mediation model. A v
2
difference test
between the two nested models shows a reduction of minimal ?t v
2
from
23.216 for the simple model to 4.047 for our model (p < 0.01), indicating
our model’s superior ?t.
8 C.P. Agoglia et al. / Accounting, Organizations and Society 44 (2015) 1–14
conclude that client desirability moderates the mediating
effect of performance evaluation on the relationship
between time reporting accuracy and the likelihood that
the manager will request the senior for a future engage-
ment.
11
That is, when the client is less desirable, reporting
accuracy affects future staf?ng decisions directly (e.g., while
accurate reporters do not suffer from lower performance
evaluations, they do suffer from a longer-term narrowing
of their future engagement opportunities). However, when
the client is more desirable, reporting accuracy affects future
staf?ng decisions indirectly via the evaluation of the senior’s
performance (e.g., they suffer from both lower performance
evaluations and limited future engagement opportunities).
Experiment 2: Partners
Method
Participants, task, procedure, and design
Experiment 2 is designed to test our agency theory pre-
dictions that partners will have a longer-term focus and
prefer staff to accurately report going over budget rather
than underreport their hours worked. Experiment 2 utilizes
the same experimental instrument (including dependent
and independent variables) and procedures as Experiment
1. However, Experiment 2 employs partners as participants.
As with Experiment 1, we asked participants to assume
they are the senior’s immediate supervisor (i.e., the man-
ager on the engagement) and task them with evaluating
the senior’s performance. We asked partners to assume
the role of manager in order to elicit responses from part-
ners as to how they would behave if performing this man-
ager-assigned task. This allows us to test our application of
agency theory to underreporting by comparing the behav-
ior of agents (managers) to that of principals (partners) act-
ing in the role of agents.
12
In addition, this design choice
maintains a level of comparability of the task (i.e., evaluation
by an immediate supervisor). Finally, it circumvents the
dilemma of having to develop an equivalent manipulation
of client desirability for partners that similarly maintains
both the strength and salience of the manager client desir-
ability manipulation; yet, it manipulates no additional facets.
Participants were 119 audit partners.
13
Partners had, on
average, 25.8 years of audit experience. There are no signif-
icant differences (p’s > .20) for participants across experi-
mental conditions for any of the demographic measures
(e.g., overall audit experience, experience evaluating subor-
dinates, or ?rm type). Including the demographic measures
as covariates does not signi?cantly affect the results or alter
the inferences drawn.
Partner results
Manipulation checks
To test the manipulations in the partner experiment, we
use the same 11-point scales as in the manager experi-
ment. Partners’ responses were signi?cantly different
between conditions, and in the expected direction, for both
the reporting accuracy (means = 2.96 and 5.45 for the
underreporter and accurate reporter conditions, respec-
tively; t = 4.762, df = 114, p < 0.001, one-tailed) and client
desirability manipulations (means = 9.18 and 6.02 for the
more and less desirable client conditions, respectively;
p < 0.001, one-tailed).
14
Table 2
Managers’ likelihoodtorequest senior onfuture engagement (Experiment 1).
Panel A: ANOVA results
Independent
variable
a
df Mean square F-Statistic p-
Value
b
Reporting
accuracy (H2)
1 29.717 9.179 0.033
Client
desirability
1 0.366 0.113 0.738
Interaction 1 0.346 0.107 0.744
Error 94 3.237
Total 98
Panel B: Mean (Standard Deviation)
c
Less desirable
client
More
desirable
client
Row
means
Cell 1 Cell 2
Underreporter 7.80 (1.73) 8.04 (1.55) 7.92
[n = 25] [n = 24]
Cell 3 Cell 4
Accurate reporter 6.81 (2.22) 6.82 (1.53) 6.82
[n = 27] [n = 22]
Column means 7.29 7.46
a
Reporting Accuracy was manipulated such that the subordinate either
accurately reports exceeding budget or reports meeting budget (i.e.,
underreports). Client Desirability was manipulated such that the client
was either more desirable or less desirable to the participant.
b
Reported p-values are two-tailed.
c
Participants indicated their likelihood of requesting the senior on a
different future engagement on an eleven-point scale where 1 = ‘‘Very
Unlikely’’ and 11 = ‘‘Very Likely’’.
11
Results of analyses following Muller, Judd, and Yzerbyt (2005) also
suggest moderated mediation.
12
Because partners were previously managers, they are likely able to
easily place themselves back in the role of manager. Therefore, we designed
the study with the expectation that they will act as the senior’s immediate
supervisor responsible for evaluating the senior (i.e., manager), with the
exception that they are likely to bring a more layered perspective based on
their experiences and incentives as partners. In addition, we feel this design
choice is appropriate for our study because we are not interested in
drawing conclusions about higher-level evaluations a partner may be
responsible for with respect to the engagement team; instead, we are
interested in how they would evaluate staff if they, as principal, were to
perform this agent-assigned task.
13
We mailed instruments to 850 partners from a list provided by the
American Institute of Certi?ed Public Accountants (AICPA). We received
replies from 125 audit partners and 67 were returned as undeliverable. The
resulting response rate is 15.96%. There were no differences between early
and late respondents. There were six unusable responses resulting from a
failure to complete key dependent variables in the partner experiment. Our
conclusions remain the same with or without these individuals.
14
Again, reassuringly, the client desirability manipulation did not signif-
icantly in?uence responses to the reporting accuracy check question (main
and interactive effects of client desirability manipulation p’s > 0.550) and
the reporting accuracy manipulation did not signi?cantly in?uence
responses to the client desirability check question (main and interactive
effects of client desirability manipulation p’s > 0.230).
C.P. Agoglia et al. / Accounting, Organizations and Society 44 (2015) 1–14 9
Hypotheses 3 and 4
Table 4 presents the descriptive statistics and ANOVA
results for the performance evaluation variable used to test
H3. Hypothesis 3 predicts that partners will evaluate a
senior who accurately reports exceeding the budget higher
than one who underreports. However, the main effect of
reporting accuracy is not signi?cant (p = 0.395, one-tailed).
While partners’ reactions to underreporting appear to dif-
fer from managers’, they show no evidence of preferring
accurate reporters (mean = 7.47) to underreporters
(mean = 7.41). Similar to H3, H4 predicts that partners will
prefer to staff future engagements with accurate reporters
over underreporters. However, again, the main effect of
reporting accuracy is not signi?cant (p = 0.856, one-tailed),
as partners are no more likely to select an accurate repor-
ter than an underreporter (means = 7.20 and 7.53, respec-
tively; see Table 5). Thus, while partners do not exhibit
the same preference for underreporters that managers
do, they also do not appear to preferentially staff future
engagements with accurate reporters. These results are
not supportive of either H3 or H4.
To shed light on the inconsistency between these
?ndings and our agency theory expectations, we examined
the partner data more closely. Although partners’
responses to the manipulation check questions were sig-
ni?cantly different between conditions and this difference
was in the expected direction, it is still possible that
partners did not fully internalize our manipulations as
we intended. Thus, we examine whether our manipula-
tions were similarly effective for partners as they were
for managers.
15
We ?nd that, while comparisons of man-
agers’ and partners’ responses to the client desirability
manipulation check indicate that the manipulation was sim-
ilarly effective for managers and partners for both the higher
and lower desirability conditions (i.e., manager Cells 2 and 4
versus partner Cells 2 and 4, t = 0.578, df = 98, p = 0.565;
manager Cells 1 and 3 versus partner Cells 1 and 3,
t = 0.655, df = 114, p = 0.514), the manipulation check for
reporting accuracy tells a different story. A closer look
reveals that, relative to managers, partners were less likely
Fig. 2. Moderated mediation: path model. Note: Unstandardized coef?cients are provided. Coef?cients for variant paths are presented for the less desirable
client condition ?rst, followed by the more desirable client condition, separated by a slash (i.e., less/more desirable client coef?cients). Bolded path
coef?cients are signi?cant at p 6 0.05.
Table 3
Moderated Mediation: Decomposition of Effects (Experiment 1).
Direct effect
(path c
0
)
a
Std
Error
t-Statistic p-Value Indirect effect
(path a  b)
a
Std
Error
t-Statistic p-Value Total
effect
Less desirable client Unstandardized À0.5404 0.2348 À2.30 0.031 0.1166 0.1805 0.65 0.321 À0.4239
Standardized À0.2527 – – 0.0541 – – À0.1986
More desirable client Unstandardized À0.3711 0.2877 À1.29 0.172 À1.0522 0.2186 À4.81