Keywords

Author name

Accounting, Organizations and Society

Volume

Issue

Pages

Accounting, Organizations and Society

SUPPORTS OPEN ACCESS

Articles in press All issues About the journal Follow journal

Volume 58
Pages 1-66 (April 2017)

Previous vol/issue Next vol/issue

5 selected: Expand all article previews Download PDFs Export

Full text access

Editorial board and publication information
Page IFC

Download PDF

Research article Full text access

Organization identity and earnings manipulation
Margaret A. Abernethy, Jan Bouwens, Peter Kroos
Pages 1-14

Download PDF Article preview

Research article Full text access

The Q&A: Under surveillance
Santhosh Abraham, Matthew Bamber
Pages 15-31

Download PDF Article preview

 Research article Full text access
Goffman's theory of frames and situated meaning-making in performance reviews. The case of a category
management approach in the French retail sector
Philippe Lorino, Damien Mourey, Géraldine Schmidt
Pages 32-49

Download PDF Article preview

Research article Full text access

Audit time pressure and earnings quality: An examination of accelerated filings
Tamara A. Lambert, Keith L. Jones, Joseph F. Brazel, D. Scott Showalter
Pages 50-66

Download PDF Article preview

ISSN: 0361-3682

Copyright © 2018 Elsevier Ltd. All rights reserved

About ScienceDirect Remote access Shopping cart Contact and support Terms and conditions Privacy policy

Cookies are used by this site. For more information, visit the cookies page.
Copyright © 2018 Elsevier B.V. or its licensors or contributors. ScienceDirect ® is a registered trademark of Elsevier B.V.
 EDITOR IN CHIEF
Keith Robson, HEC Paris, Paris, France
EDITORS
R.J. Bloomfield R. Krishnan M. Shields
Cornell University, Ithaca, USA Michigan State University, East Lansing, Michigan State University, East Lansing, USA
W.F. Chua Michigan, USA
University of Sydney, New South Wales, Australia P.B. Miller H. Tan
C. Chapman London School of Economics, London, UK Nanyang Technological University, Singapore
University of Bristol, Clifton, UK M. Power
D.J. Cooper London School of Economics and Political Science, K. Trotman
University of Alberta, Edmonton, Canada London, UK University of New South Wales, Sydney, Australia

EDITORIAL BOARD
C. Agoglia J. Fisher D. Larcker P. Quattrone
University of Massachusetts at Amherst, Indiana University, Bloomington, USA Stanford University, Stanford, USA University of Edinburgh, Edinburgh,
Amherst, Massachusetts, USA T.J. Fogarty R. Libby Scotland, UK
T. Ahrens Case Western Reserve University, Cleveland, Cornell University, Ithaca, USA
A. Richardson
United Arab Emirates University, Al-Ain, USA A.M. Lillis
University of Windsor, Windsor, Canada
United Arab Emirates C. Free University of Melbourne, Melbourne, Australia
S. Anderson University of New South Wales, Sydney, M. Lipe J. Roberts
University of California at Davis, Davis, USA New South Wales, Australia University of South Carolina, Columbia, South University of Sydney, Darlington, Australia
M. Annisette Y. Gendron Carolina, USA S. Salterio
University Laval, Quebec, Canada
York University, Toronto, Canada J. Luft Queen’s University, Kingston, Ontario,
I. Grabner Michigan State University, East Lansing, USA
M. Artz Canada
University of Maastricht, Maastricht,
Frankfurt School of Finance and Management, K. Lukka K.L. Sedatole
Netherlands
Frankfurt, Germany Turku School of Economics, Turku, Finland
C. Graham Michigan State University, East Lansing, USA
R. Balakrishnan V. Maas
York University, Toronto, Canada G. Sprinkle
University of Iowa, Iowa City, Iowa, USA Universiteit van Amsterdam, Amsterdam,
J. Hales Netherlands Indiana University, Bloomington, USA
A. Bhimani Georgia Institute of Technology, Atlanta,
London School of Economics, London, UK D. MacKenzie W. Tayler
Georgia, USA
J. Birnberg University of Edinburgh, Edinburgh, UK Brigham Young University, Provo, Utah, USA
M. Hall
University of Pittsburgh, Pittsburgh, USA B. Malsch S. Toms
London School of Economics, London, UK
J.C. Bol Queens University, Canada, Kingston, Canada
J. Hammersley University of Leeds, Leeds, UK
Tulane University, New Orleans, Louisiana, USA C. McWatters
University of Georgia, Atlanta, USA K. Towry
R. Bryer University of Ottawa, Ontario, Canada
L. Hannan Emory University, Atlanta, USA
University of Warwick, Coventry, UK A. Mennicken
Georgia State University, Atlanta, USA
London School of Economics, London, UK J. Unerman
B. Carruthers R. Hatfield
Northwestern University, Evanston, Illinois, USA K. Merchant Royal Holloway, University of London,
University of Alabama, Tuscaloosa, Alabama,
USA University of Southern California, Los Angeles, London, England, UK
C. Carter
F. Hodge USA W.A. Van der Stede
University of Edinburgh, Edinburgh, Scotland,
UK University of Washington, Seattle, Washington, M. Messner London School of Economics, London, UK
USA Leopold-Franzens-Universität Innsbruck,
C. Chen H. Vollmer
V.B. Hoffman Innsbruck, Austria
University of Illinois at Urbana-Champaign, University of Leicester, Leicester, UK
Champaign, Illinois, USA University of Pittsburgh, Pittsburgh, USA J. Meyer
P. Hopkins Stanford University, Stanford, USA M. Walker
R. Chenhall
Indiana University, Bloomington, Indiana, USA F. Moers University of Manchester, Manchester, UK
Monash University, Melbourne, Australia
M. Holzhacker University of Maastricht, Maastricht, S.P. Walker
C. Cho Netherlands
Michigan State University, Michigan, USA University of Edinburgh, Edinburgh,
York University, Toronto, Canada
C. Ittner D. Moser
J. Choi Scotland, UK
University of Pennsylvania Wharton School, University of Pittsburgh, Pittsburgh, USA
University of Pittsburgh, Pennsylvania, USA M. Wouters
Philadelphia, USA J. Mouritsen
C. Cooper Copenhagen Business School, Copenhagen, Karlsruhe Institute of Technology (KIT),
J. Jollineau
University of Strathclyde, Glasgow, UK University of San Diego (USD), San Diego, Denmark Karlsruhe, Germany
C. Dambrin California, USA M. Nelson J. Young
ESCP Europe, Paris, France K. Kadous Cornell University, Ithaca, USA University of New Mexico, Albuquerque,
H.C. Dekker Emory University, Atlanta, USA B. O’Dwyer USA
VU University, Amsterdam, Netherlands L. Koonce Universiteit van Amsterdam, Amsterdam,
S. Dikolli S. Young
University of Texas at Austin, Austin, USA Netherlands
Duke University, Durham, USA Lancaster University, Bailrigg, UK
J. Kuang M. Peecher
W. Espeland Georgia Institute of Technology, Atlanta, University of Illinois at Urbana-Champaign, M. Zimbelman
Northwestern University, Evanston, USA Georgia, USA Urbana, Illinois, USA Brigham Young University, Provo, USA

Publication information. Accounting, Organizations and Society (ISSN 0361-3682). For 2017, volumes 56–63 (8 issues) are scheduled for publication. Subscription prices are available upon request from the Publisher
or from the Elsevier Customer Service Department nearest you or from this journal’s website (http://www.elsevier.com/locate/aos). Further information is available on this journal and other Elsevier products through
Elsevier’s website: (http://www.elsevier.com). Subscriptions are accepted on a prepaid basis only and are entered on a calendar year basis. Issues are sent by standard mail (surface within Europe, air delivery outside
Europe). Priority rates are available upon request. Claims for missing issues should be made within six months of the date of dispatch.
Advertising information: If you are interested in advertising or other commercial opportunities please e-mail Commercialsales@elsevier.com and your inquiry will be passed to the correct person who will respond to
you within 48 hours.
Orders, claims, and journal inquiries: please contact the Elsevier Customer Service Department nearest you: St. Louis: Elsevier Customer Service Department, 3251 Riverport Lane, Maryland Heights, MO 63043,
USA; phone: (877) 8397126 [toll free within the USA]; (+1) (314) 4478878 [outside the USA]; fax: (+1) (314) 4478077; e-mail: JournalCustomerService-usa@elsevier.com. Oxford: Elsevier Customer Service
Department, The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK; phone: (+44) (1865) 843434; fax: (+44) (1865) 843970; e-mail: JournalsCustomerServiceEMEA@elsevier.com.
Tokyo: Elsevier Customer Service Department, 4F Higashi-Azabu, 1-Chome Bldg, 1-9-15 Higashi-Azabu, Minato-ku, Tokyo 106-0044, Japan; phone: (+81) (3) 5561 5037; fax: (+81) (3) 5561 5047; e-mail:
JournalsCustomerServiceJapan@elsevier.com. The Philippines: Elsevier Customer Service Department, 2nd Floor, Building H, UP-Ayalaland Technohub, Commonwealth Avenue, Diliman, Quezon City,
Philippines 1101; phone: (+65) 6349 0222; fax: (+63) 2 352 1394; e-mail: JournalsCustomerServiceAPAC@elsevier.com.
∞ The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper)
USA mailing notice: Accounting, Organizations and Society (ISSN 0361-3682) is published 8 issues per year (in January, February, April, May, July, August, October and November) by Elsevier Ltd
(The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK). Periodical postage paid at Jamaica, NY 11431 and additional mailing offices (not valid for journal supplements). USA POSTMASTER:
Send change of address to Accounting, Organizations and Society, Elsevier Customer Service Department, 3251 Riverport Lane, Maryland Heights, MO 63043, USA.
AIRFREIGHT AND MAILING in USA by Air Business Ltd., c/o Worldnet Shipping Inc., 156-15, 146th Avenue, 2nd Floor, Jamaica, NY 11434, USA.

© 2017 Elsevier Ltd.

 Accounting, Organizations and Society 58 (2017) 1e14

Contents lists available at ScienceDirect

Accounting, Organizations and Society

journal homepage: www.elsevier.com/locate/aos

Organization identity and earnings manipulation*

Margaret A. Abernethy a, Jan Bouwens b, *, Peter Kroos b
a
Department of Accounting, University of Melbourne, Australia
b
Department of Accounting, University of Amsterdam, The Netherlands

a r t i c l e i n f o a b s t r a c t

Article history: Management scholars are beginning to provide empirical evidence that organization identity (OI) can be
Received 13 April 2016 a powerful means of reducing agency costs. We examine whether an individual's identity with the firm
Received in revised form influences the agency costs associated with incentive contracts, namely earnings manipulation. Based on
24 March 2017
OI theory, we expect that managers who identify with the firm gain utility by taking actions that in their
Accepted 10 April 2017
Available online 18 April 2017
view benefits the firm, and experience disutility from taking actions that are harmful to the firm.
Drawing on a third-party survey database, we find that performance-based compensation is associated
with higher levels of earnings manipulation. Importantly, we also find that managers with incentive-
Keywords:
Organization identity
based compensation engage in lower levels of opportunistic earnings manipulation when they iden-
Incentives tify with the firm.
Earnings manipulation Crown Copyright © 2017 Published by Elsevier Ltd. All rights reserved.

1. Introduction recognized the importance of prosocial preferences as a motivation

Since the early 1950s accounting researchers have documented
the agency costs that emerge with the use of incentives-based
contracts (e.g., Healy, 1985).1 Where incentive contracts are imple-
mented to motivate productive effort, concerns have been raised
about the potential for managers to engage in earnings manipula-
tion (Dechow & Skinner, 2000; Jensen & Murphy, 2012). While
economics-based principal-agent models assume that managers
strictly act out of self-interest and their welfare solely relies on in-
come and effort, management scholars have for some time
*
This study received generous support from Thomas Henry Carroll-Ford Foun-
dation at Harvard Business School. We acknowledge feedback from Chung-Yu Hung,
Flora Kuang, Michal Matejka, Bo Qin, Naomi Soderstrom and Mark Wallis as well as
from the participants of seminars at the University of Melbourne, VU University
Amsterdam, Foster School of Business of the University of Washington, and the
2012 AAA MAS conference. We thank two anonymous reviewers and Chris
Chapman (editor) for their helpful comments.
* Corresponding author.
E-mail addresses: m.abernethy@unimelb.edu.au (M.A. Abernethy), JBouwens@
uva.nl (J. Bouwens), P.Kroos@uva.nl (P. Kroos).
1
Recent media about the provision of incentive-based compensation to sales
employees at retail bank Wells Fargo is a case in point. Likewise, supermarket chain
Tesco experienced an accounting fraud with Tesco's current CEO suspending
performance-based pay pending the serious fraud office investigation (The
Guardian, 29 October 2014) and the conclusion of serious fraud resulting in
charging three former Tesco executives: “as senior employees at the supermarket
they had abused their seniority for personal gain” (The Guardian, 22 September
2016).
for behavior. Researchers have more recently begun to use Akerlof
and Kranton’s (2000, 2005, 2008, 2010) notion of organization
identity (OI) to empirically assess its effect on agency costs (Boivie,
Lange, McDonald, & Westphal, 2011). Organization identity is
powerful in encouraging managers to do the right thing by the firm
as their utility increases (decreases) when they act (do not act) in
the best interest of the firm. The theoretical underpinnings of OI are
useful for understanding how OI works (Akerlof & Kranton, 2008;
Heinle, Hofmann, & Kunz, 2012) and perhaps even for under-
standing why some managers behave with integrity and some don't
(see Dikolli, Keusch, Mayew, & Steffen, 2016).
This paper examines whether a manager's OI provides a means
of mitigating some of the agency costs associated with the provi-
sion of financial incentives. The intent of incentive contracts is to
direct managers' attention to actions that will add value to the firm.
Managers are incentivized to do so as their compensation depends
on some measure of value creation. However, firms are not always
successful in designing such contracts. The problem, of course, is
that contributions of managers to firm value are imperfectly
measured. Incentive contracts can prompt opportunistic behavior
designed to improve those measures but which do not improve
firm value, what we call earnings manipulation. There is a long
history of research documenting how managers make accounting
choices to increase the proceeds of their bonus plan (Healy, 1985;
Guidry, Leone, & Rock, 1999). Roychowdhury (2006), for example,
reports managers offering price discounts to boost revenues and
reducing discretionary expenditures to improve reported

http://dx.doi.org/10.1016/j.aos.2017.04.002
0361-3682/Crown Copyright © 2017 Published by Elsevier Ltd. All rights reserved.
2 M.A. Abernethy et al. / Accounting, Organizations and Society 58 (2017) 1e14
performance. We draw on the intuition provided by Akerlof and
Kranton (2000, 2005, 2008, 2010) to empirically examine the role
of OI in mitigating opportunistic reporting choices.
In our study, earnings manipulation (EM) includes choices that
result in changes in reported income, which encompasses account-
ing manipulation (e.g., shifting between accounts) and real earnings
management (e.g., postponing necessary investments). Consistent
with prior research, we expect a positive relationship between in-
centives and earnings manipulation (e.g., Healy, 1985). In addition,
we examine whether managers who are rewarded by means of
incentive contracts and who identify with the firm will engage in
less EM than those agents who do not identify with the firm. Man-
agers with higher OI will experience disutility with actions that in-
crease their own wealth but which could potentially be harmful to
the firm (Akerlof & Kranton, 2008). We expect that OI will negatively
moderate the positive relation between incentives and EM.2
Standard economic theory does not distinguish between
different sources of motivation which are just considered mani-
festations of underlying preferences (such as preferences for the
task or the reward that follows from successfully completing the
task). It is important to note that our line of argument regarding the
role of OI is different from other forms of motivation such as
intrinsic or extrinsic motivation. Psychology researchers emphasize
that intrinsic motivation come from within a person. Deci (1971)
describe intrinsic motivation as the desire “to perform an activity
when one receives no apparent reward than the activity itself” (p.
105). Organization identity is somewhat different in that those with
high OI do not necessarily have to enjoy the activity itself but
instead are motivated by a sense of connectedness with their or-
ganization (Adler & Chen, 2011; Heinle et al., 2012). It is also a
psychological preference that can be influenced or activated by the
organization itself (Akerlof & Kranton, 2008).
We test our predictions using data collected by a third-party
consulting firm from senior financial controllers. The use of finan-
cial controllers is particularly useful given that these individuals are
not only managers but also have formal training in accounting and
thus have a good working knowledge of how to manipulate earn-
ings (Dichev, Graham, Harvey, & Rajgopal, 2013; Feng, Ge, Luo, &
Shevlin, 2011; Ge, Matsumoto, & Zhang, 2011). Indeed, prior liter-
ature shows that the incentives of financial executives are partic-
ularly important in explaining earnings manipulation (e.g., Jiang,
Petroni, & Wang, 2010). The use of survey data enables us to
address some of the “vexing questions that have been difficult to
address with archival work” (Dichev et al., 2013).
In line with most prior research, the focus of our paper is on
organizational identification which is distinct from disidentification.
The concept of disidentification is not simply the opposite of orga-
nizational identification but is described as a separate variable and a
unique psychological state where individuals disconnect themselves
from their organizations (Kreiner & Ashforth, 2004). While dis-
identification may be less prevalent within organizations (due to
employee turnover), it should be recognized that our findings only
extrapolate to cases where individuals have a neutral or positive
identification with their organization.
Consistent with previous literature we find that incentives are
positively and significantly associated with EM. More importantly,
our findings also show that OI is negative and significant in
moderating the relation between incentives and EM. Our results
2
We do not claim that EM is by definition value-decreasing. Some EM is
opportunistic while other EM may improve the representational faithfulness and
predictive usefulness of accounting information. We assume that EM not induced
by incentive contracts is less likely to be motivated by self-interest and that EM
induced by incentive contracts is more likely to be opportunistic.
suggest that OI reduces the agency losses associated with incentive
contracts. Viewed differently, if an individual does not identify with
the firm, then incentives may prompt managers to engage in self-
serving behavior that can harm the firm. Given most firms use
incentive contracts, our results support the idea that it is possible to
use incentives provided the firm implements strategies to ensure
that individuals are committed to the firm.
The paper contributes to prior theoretical and empirical litera-
ture by documenting the role of OI in controlling agency costs
(Akerlof & Kranton, 2008; Boivie et al., 2011). To the extent that
incentive contracts prompt managers to take actions that benefit
themselves but may be harmful for the firm, it would appear that OI
curbs this incentive-induced behavior. We also show how OI and
incentives can complement each other. Prior research shows how
OI can motivate managers to work towards organizational objec-
tives; incentives tied to performance measures help by directing
attention of employees to those action choices that are most
desirable (e.g., Malina & Selto, 2001). We show that OI can mitigate
the potential for adverse consequences to emerge (e.g., EM) when
monetary incentives are used. Our findings also enable us to
contribute to the debate concerning incentives and EM, and pro-
vide an additional rationale for the ambiguity found in prior studies
(Armstrong, Jagolinzer, & Larcker, 2010). These authors call for
future research to “consider behavioral explanations in addition to
the traditional economic or agency rationalizations” (p. 261). We
also contribute to an emerging stream of research attempting to
assess the ‘manager’ effect on reporting choices (Bamber, Jiang., &
Wang, 2010; Dikolli et al., 2016; Geiger & North, 2006; Jia, Van
Lent, & Zeng, 2014; Schrand & Zechman, 2012).
The remainder of the paper is structured as follows. Section 2
reviews prior literature. Section 3 describes the sample and mea-
sures and Section 4 presents the results. The final section provides
concluding results.
2. Hypothesis development
For some time researchers have devoted attention to under-
standing the role of incentive contracts in misreporting and the
consequences for external stakeholders (e.g., Dechow, Ge, &
Schrand, 2010; Dichev et al., 2013; Fields, Lys, & Vincent, 2001).
Of particular concern is that incentive contracts are implemented
ex-ante to motivate productive effort, but ex-post these contracts
can elicit opportunistic reporting choices that disguise the firm's
economic performance and enhance the welfare of corporate ex-
ecutives (Badertscher, Collins, & Lys, 2012). We, first, review the
literature on the relation between managerial incentives and
earnings manipulation and subsequently explore the role of orga-
nization identity (OI) as a potentially factor that mitigates the
opportunistic earnings manipulation.
2.1. Incentives and earnings manipulation
A major stream of the accounting literature documents oppor-
tunistic reporting choices of managers, consistent with the notion
that managers use their reporting discretion to increase their own
wealth and this comes at the expense of the firm. Healy (1985) was
one of the first to document that managers use income-decreasing
accruals when they are below the threshold or above the cap of
the bonus plan, while managers select income-increasing accruals
when they are within the incentive zone. Guidry et al. (1999) show
that business-unit managers choose income-increasing accrual pol-
icies when they are in the bonus range. The more recent evidence
suggests that managers not only make opportunistic reporting
choices to maximize the proceeds from their annual bonus plans, but
also do so in response to equity-based incentives. There are
 M.A. Abernethy et al. / Accounting, Organizations and Society 58 (2017) 1e14
numerous examples of how such behavior has adverse consequences
for the firm. For example, Bhojraj, Hribar, Picconi, and McInnis (2009)
find that managers with strong equity-based incentives will cut
discretionary expenses and manage accruals to meet and beat the
analysts’ forecasts. While such actions result in higher stock market
returns in the short term (i.e., in the subsequent year) these actions
result in lower longer term returns (i.e., returns over the next three
years). Bhojraj et al. (2009) also document that for firms managing
their reported earnings numbers to marginally beat their forecasts,
there is greater insider selling in the subsequent year.
Aboody and Kasznik's (2000) findings indicate that executives
delay good news and bring forward bad news to manage investors'
expectations downwards surrounding the dates when options are
granted (to lower the exercise price of those options). Bergstresser
and Philippon (2006) find that executives who are strongly
incentivized exhibit higher levels of earnings manipulation. These
authors show how executives exercise and sell unusually large
amounts of options and shares in those years in which accruals
make up a large part of their firms' reported earnings.3 Dichev et al.
(2013) reports that 89 percent of CFOs in their survey indicate that
reported earnings misrepresent the true underlying economic
performance of the firm. And that it is done to influence executive
compensation. Graham, Harvey, and Rajgopal (2005) report that 80
percent of their surveyed CFOs indicated that they would delay
research and development, advertising, and maintenance spending
to achieve earnings benchmarks. These reporting choices introduce
bias or noise to the reported numbers and reduce the informa-
tiveness of reporting earnings numbers (Badertscher, 2011). There
is thus a significant body of literature linking incentives to earnings
manipulation. Our first hypothesis is as follows:
H1: Incentives are positively associated with earnings
manipulation.
2.2. Organization identity, incentives and earnings manipulation
Organization identity (OI) stems from social identity theory and
is defined as the extent to which an individual's self-identity is
“intertwined with the identity of her organization, or the degree to
which the individual defines herself in terms of the attributes of the
organization” (Boivie et al., 2011; Mael & Ashforth, 1992). Social
identity theory was developed by Tajfel & Turner (1979) to un-
derstand the psychological basis for intergroup discrimination (e.g.
racial discrimination). Traditional agency models, on the other
hand, assume that individual preferences are independent of social
context. Akerlof and Kranton (2000) bring these literature together
and demonstrate how the OI of employees influences their utility
and in turn their behavior. Once an individual sees herself as a part
of the organization, she derives utility by doing ‘right’ by the or-
ganization. Deviation from what is in the firm's best interests
causes disutility. Akerlof and Kranton (2008) consider OI to be an
important supplement to incentive contracts, particularly when
such contracts are costly. They argue that a firm will find it bene-
ficial to rely on OI in settings where inculcating identity comes at
low cost; when there is uncertainty; and when employee efforts are
3
Others also document a positive association between equity-based incentives
and earnings manipulation (Beneish & Vargus, 2002; Cheng & Warfield, 2005).
Armstrong et al. (2010) argue that market efficiency mitigates the potential adverse
consequences of earnings manipulation just as reputational and legal costs pre-
clude executives from engaging in misrepresentation of earnings numbers.
4
They use West Point training for their cadets as an example of the importance
of OI in the military. Others have provided examples in other industries (Gioia,
Price, Hamilton, & Thomas, 2010; Ravasi & Phillips, 2011).
3
hard to observe (Akerlof & Kranton, 2005, p. 15).4 Others provide
similar arguments (Benabou & Tirole, 2016; Jensen & Murphy,
2012).
We draw our intuition from Akerlof and Kranton (2000, 2005,
2008, 2010) who argue that incentive contracts alone are not suf-
ficient to elicit desired behavior as these contracts only rely on
contractible performance measures that are often imperfect in-
dicators of agents' effort choices creating the potential for man-
agers to game the system. Heinle et al. (2012) examine theoretically
the relation between incentives, identity, and performance mea-
surement. In a multi-task setting they demonstrate the conditions
under which an agent who identifies with the firm also receives
stronger incentives. These theoretical arguments suggest that to
the extent that OI reduces dysfunctional side-effects of incentive
contracts, OI and incentives may work as complements, as OI in-
creases the marginal benefits from the provision of incentives.
Empirical research demonstrates the value of OI in reducing agency
losses. For example, Boivie et al. (2011) show that CEOs with higher
levels of OI are less likely to exert their influence on the Board of
Directors to secure higher levels of cash compensation when the
firm's performance is low. This suggests that a ‘greater connection
between a CEO's identity and her firm's identity leads the CEO to do
what is right for the firm, because helping the firm is tantamount to
helping herself’ (Boivie et al., 2011, p. 566).
Given that most firms have some form of incentive contracts, we
do not argue that OI replaces incentive contracts altogether.
Developing employees' OI, however, can be an effective comple-
ment to the use of incentive contracts and vice versa. Where higher
levels of OI may result in managers being motivated to work to-
wards organizational objectives, incentive contracts tied to specific
performance measures can complement OI as the incentive con-
tracts may be helpful in directing managerial attention in a multi-
task framework. Most managers face a multi-task environment
and the problems of developing appropriate performance measures
in this environment have been well documented (Roberts, 2010).
However, it has long been recognized that performance measure-
ment is used not only for ‘scorekeeping’ or reward purposes but
also to direct attention to what is important (Simon, 1954). For
example, Malina and Selto (2001) demonstrate how balanced
scorecards are used as a communication device to provide strategic
guidance to division managers. Abernethy, Dekker, and Schultz
(2015) find that firms which emphasize organizational learning as
a strategic priority put a greater emphasis on employee selection as
a sorting mechanism to ensure that employee’ values and beliefs
are aligned with those of the firm. However, these organizations
also make use of incentive contracts to provide direction to em-
ployees on how to prioritize their action choices.
We focus on the role of OI in mitigating the potential for earn-
ings manipulation, induced by incentives, to emerge.5 We expect
that the relation between incentive-based pay and earnings
manipulation will be less positive when individuals identify with
the firm. Our intuition for this expectation is based on organization
identity theorists who argue that when an individual strongly
identifies with the firm they will behave in ways that benefits the
firm because doing so enhances their own concept of self (e.g.,
Dukerich, Golden, & Shortell, 2002; Dutton, Dukerich, & Harquail,
1994). When there is potential for agency costs to emerge,
namely where agents are compensated with incentive contracts,
we expect that agents with a strong OI will experience disutility
when making reporting choices that may increase their own wel-
fare, but may come at the expense of the firm (Akerlof & Kranton,
5
We use the term broadly to include all types of performance-based incentives
including increases in salary due to performance improvements.
4 M.A. Abernethy et al. / Accounting, Organizations and Society 58 (2017) 1e14
2000). It is the interaction between OI and incentives that can result
in “strong negative emotions such as guilt and loss of self-esteem”
and thus influence those choices that increase reported earnings
(Heinle et al., 2012, p. 1310). Akerlof and Kranton (2010, p.50) argue
that incentives in isolation are unlikely to create conditions for
agents to select actions valuable to the firm. Our expectations are
summarized as follows:
H2: The positive relation between incentives and earnings
manipulation will be negatively moderated by OI.
2.3. Does OI have a direct effect on earnings manipulation?
We argue in H2 that organization identity can play a mitigating
role on earnings manipulation when agents have performance-
based compensation contracts. That is, while OI is not directly
aimed at earnings manipulation, personal incentives may induce
managers to take opportunistic actions that benefit themselves but
which negatively impact firm value. OI curbs this behavior. Absent
personal incentives, it is not clear a priori what reasons managers
have to make opportunistic reporting choices. Organization iden-
tity might also play a role in other circumstances, e.g., if individuals
make opportunistic reporting choices due to the threat of dismissal.
In other words, there may be other personal incentives than
performance-based compensation plans that prompt managers to
manipulate earnings. If this is the case, then we expect that OI will
be negatively associated with earnings manipulation. In sum,
incentive compensation plans represent the most widely discussed
determinant of opportunistic earnings manipulation. Nonetheless,
we cannot rule out the possibility that OI will have a negative and
direct effect on earnings manipulation due to other personal in-
centives driving the opportunistic behavior that is not explicitly
accounted for in our model. So, our third hypothesis represents a
joint hypothesis of the presence of other than compensation-based
incentives that are explaining earnings manipulation and the role
of OI in curbing this incentive-induced behavior.
H3: OI is negatively associated with earnings manipulation.
3. Research method
3.1. Survey design and sample
The survey data were collected by NBA-VRC, the Dutch financial
controllers association (henceforward referred to as Association).
Membership of the Association is open to accountants and con-
trollers in business who pass their post-graduate examinations in
financial control or auditing. Our respondents typically have
reporting duties as well as decision making duties (e.g., day-to-day
management finance function, financial planning and analysis,
treasury, projects to acquisitions and disinvestments). We target
respondents working as senior financial controllers given that prior
literature shows that earnings manipulation is more sensitive to
CFO incentives relative to CEO incentives (Jiang, Petroni, & Wang
2010).6 Senior financial controllers, when surveyed, also acknowl-
edge that personal incentives play an important role in reporting
decisions (Dichev et al., 2013; Graham et al., 2005). In sum, we
6
Likewise, Chava & Purnanandam (2010) find that CFOs risk-taking incentives
are more strongly associated with earnings smoothing through accruals, relative to
CEO incentives. Kim, Li, & Zhang (2011) show how primarily CFO incentives, and
not so much CEO incentives, are positively associated with firms' future stock price
crash risk.
believe that the focus on senior financial controllers is consistent
with theory on incentive compensation and earnings manipulation.
The procedure through which the survey was developed and
executed is as follows. The Association was interested in the
compensation packages of controllers, controls in place and to learn
at what organizational positions their members were situated. We,
as the research team, were allowed to incorporate a research
question on the topic of accounting and control in the study. We
agreed that the research team would select the instruments from
existing studies. The agreement was that the Association would
administer these instruments without any adjustments. To the
extent instruments were purpose developed, the Association was
involved in formulating the questions. Some of the measures
included in this study were determined by the Association; for
example, how the position of the controller in the organization was
measured. In general, some tests to assess the convergent validity
of our constructs are performed with variables that were co-
developed with the Association such as the type of work of con-
trollers. Thus, while there were some trade-offs made in having the
Association involved in the development of the measures and in
them administering the survey, we believe that their contribution
was positive given their intimate knowledge of working conditions
of their members. The Association also put limits on the length of
the questionnaire so as not to impose on their members. All
members were invited by e-mail to participate in the online survey.
The respondents had about five weeks to participate. The Associ-
ation sent only one reminder to potential respondents one week
before the online-survey was closed.
This procedure yields an initial sample of 272 respondents.7 We
distinguish several reasons for non-response. First, potential re-
spondents may no longer be working as a financial controller.
Second, despite attempts of the Association to keep their mailing
list up to date, survey invitations did not always reach the intended
recipient. Third, as Van der Stede, Young, and Chen (2005) note the
response rate might be due to negative effect of seniority and
increased responsibilities of the survey population.8 Fourth, we
were restricted in our attempts to follow up on the initial survey
invitation. We excluded 81 respondents who were either not in
senior management positions or who worked in a unit that did not
resemble a profit center. We removed nine observations from our
sample because of missing data on one of our variables. Our final
sample is composed of 183 observations from financial controllers.9
Our respondents are relatively senior as they are, on average, 42
years of age, 44% has a CPA degree, and 39% has a CMA degree. The
respondents have been working at their current job for four years
(see Table 1). In addition, 56 percent of our respondents work at the
corporate level, 34 percent work at a division or business unit, and
7
We compute the response rate on the basis of the number of members that
meet our selection criteria. That is, according to the NBA-VRC data 5692 members
represent our target sample where the majority works in financial controlling
positions (5,067) and the remaining members work as business analysts, compli-
ance and risk manager, etc. This yields a response rate of 4.8%, consistent with
response rates of other email surveys to this type of respondent (see Graham et al.,
2005; Indjejikian & Matejka, 2009; and; Dichev et al., 2013).
8
Van der Stede et al. (2005) identify a number of reasons for declining response
rates across the social sciences.
9
We define financial controllers as employees whose activities, such as external
and internal reporting, forecasting future performance, financial analyses to sup-
port decision-making, developing business information systems, and compliance
represent the majority of their job. More than 80% of our respondents were
assigned job titles such as CFO, financial director, group controller, business-unit
controller, controller, finance manager. The financial controller may be working at
the corporate, divisional, business unit, or departmental level. The majority of our
respondents worked at the corporate level but some respondents worked at a lower
level. As robustness analyses, we control for variation in job titles and organiza-
tional levels in which the financial controller is operating.
 M.A. Abernethy et al. / Accounting, Organizations and Society 58 (2017) 1e14
10 percent work at departmental level (non-tabulated). The char-
acteristics of our sample are further discussed in section 4.1.
What is important for survey research is not so much the
response rate but rather the representativeness of the responses
(Van der Stede et al., 2005).10 We test for the presence of a response
bias in two ways. First, we compare the target population and our
final sample on demographic variables such as age and gender. We
do not find significant differences in means. Second, we compare
the responses of early and late respondents as late respondents are
more similar to non-participants (Moore & Tarnai, 2002). We use
the date of the reminder email to distinguish between early and
late respondents; we observe no significant differences in means
for our dependent variable, independent variables of interest, or
our control variables. In addition, we also do not find significant
differences between the overall experience, the experience in the
current job, and the job titles between early and late respondents,
evaluated at the 10% significance level (non-tabulated).11 In sum,
we believe that non-response bias is not likely to significantly affect
our results.
3.2. Variable measurement
We describe the measurement of each test and control variable.
For variables measured using Likert scales (using 7-point scales),
we establish the convergent validity by examining the correlation
with alternative measures (Lattin, Carrol, & Green, 2003). We heed
the advice of Ittner and Larcker (2001) and, where possible, rely on
more objective data instead of perceptual measures to assess the
convergent validity.
3.3. Test variables
The dependent variable, earnings manipulation (EM), is
measured using an adapted version of the instrument developed by
Merchant (1990).12 The instrument has been used by others (e.g.,
Chow, Kato, & Merchant, 1996). Respondents indicate how often
their unit postponed necessary expenditures, accelerated sales, and
shifted funds between accounts to manage current earnings
numbers. The survey uses a seven-point Likert scale (1 ¼ never
occurs, 4 ¼ occurs sometimes, 7 ¼ occurs frequently). We establish
convergent validity by using two alternative measures. First, we ask
the respondents to indicate the likelihood that they will leave the
organization in one year time. Given that most earnings
10
Even when response rates are low, results are generalizable if there is low non-
response bias. The issue is to what extent respondents systematically differ from
non-respondents. So, what is of interest is whether respondents and non-
respondents differ on variables that are of interest considering the research ques-
tion addressed in the survey.
11
Nonresponse depends on the authority, capacity, and motivation to respond.
The analyses suggest that there are no significant differences in the position and
access to information and knowledge that is being asked between respondents and
non-respondents.
12
We excluded one item regarding relating to the accounting treatment of
equipment purchases. We did so as “many firms may have no occasion to buy
equipment that could be produced as well inside the company” (Merchant, 1990, p.
306) and this item may therefore not be relevant for our cross-section of firms. In
addition to its low face validity, we assessed the internal consistency through item-
to-item correlations. The item had correlations of 0.16 with the item regarding
postponing necessary expenditures, below the threshold of 0.30 for inter-item
correlations (Hair et al., 2014, p. 123). As factor analysis considers the shared
variance among items, the original four-item instrument had two factor loadings
below the typical threshold of 0.50 for practical significance. Exclusion of the item
yields loadings all greater than 0.50. Removal of the item does not affect our main
inferences; the coefficient on the moderator OI*INC_FIN is negative and significant
(p ¼ 0.01 and p < 0.01 for the full models reported in Table 2) using the four-item
instrument (not tabulated).
5
manipulation, as defined here, reverse over time (e.g., accelerating
sales in the current period leads to lower next-period sales), we
expect that respondents who are more likely to leave their orga-
nization are also more likely to manipulate earnings as they do not
face the adverse consequences in future periods (e.g., Dechow &
Sloan, 1991). Indeed, we document a significant positive correla-
tion between the likelihood that they leave their organization in
the next year and earnings manipulation (r ¼ 0.14, p < 0.1). Second,
we assess the degree of target pressure by asking respondents to
what extent they experience pressure to meet or beat financial
targets (Van der Stede, 2001).13 Consistent with our expectations
that stronger target pressure is correlated with stronger earnings
manipulation, we find a positive and significant correlation
(r ¼ 0.45, p < 0.01). In sum, these findings support the convergent
validity of our earnings manipulation measure.
The first independent variable of interest, organizational iden-
tity (OI), is measured using an instrument developed by Mael and
Ashforth (1992), which has been used and validated extensively
(e.g., Boivie et al., 2011). This instrument is composed of six items
that capture the degree to which individuals define themselves as
organization members. For example, ‘when someone makes posi-
tive remarks about my organization, it feels like a personal
compliment’ or ‘when I talk about my organization, I usually say we
rather than they’. The survey uses a seven-point Likert scale
(1 ¼ not at all, 4 ¼ to some extent, 7 ¼ to a very large extent) where
higher values reflect higher degrees of organizational identity. We
establish convergent validity by using two measures composed of
hard, objective data. We ask respondents to provide their organi-
zational tenure in full years, the number of hours that they should
work according to their contract, and the actual number of hours
they work. We compute a ratio for the additional hours worked
using actual number of hours divided by the contractual number of
hours. We expect that tenure is positively correlated with OI (Boivie
et al., 2011), and that OI is positively correlated with the degree of
working more than contracted. Consistent with our expectations,
the correlation between tenure and OI is positive and significant
(r ¼ 0.15, p < 0.1) and the correlation between OI and our working
hours ratio is positive and significant (r ¼ 0.16, p < 0.05). Consistent
with Bamber and Iyer (2002), we expect that higher OI is associated
with a smaller likelihood that respondents will leave the organi-
zation. We find a negative but insignificant correlation (p ¼ 0.18). In
sum, most of the results support the convergent validity of our OI
measure.
The second independent variable of interest, incentives, should
capture the ex-ante incentives that financial managers face. In-
centives are measured by asking respondents to provide the
maximum of performance-based incentives they may earn,
expressed as a percentage of their base salary, which is set at 100%
(Abernethy, Bouwens, & Van Lent, 2013). The incentive may be a cash
bonus, stock, stock options, etc. For example, when incentives are
50%, this implies that managers at the beginning of the period know
that improving reported performance will at most lead to
performance-based compensation that is 50% of their fixed
compensation. When respondents are eligible for performance-
based compensation, this entails in 98% of the cases a cash bonus,
while 14%, 6%, and 2% of the respondents are eligible for stock, op-
tions, and other performance-related compensation respectively.
Besides the maximum of performance-based incentives, we also
asked about manager's actual performance-based compensation.
13
We distinguish eight items where respondents rate the extent to which they
experience pressure to meet or beat annual or quarterly sales targets, overall or
individual project expense budgets, etc. All items load upon one factor greater than
unity and the Cronbach alpha coefficient is 0.92.
6 M.A. Abernethy et al. / Accounting, Organizations and Society 58 (2017) 1e14

Table 1
Summary statistics.
Panel A reports descriptive statistics for the full sample. Panel B reports descriptive statistics for two subsamples contingent on the presence of bonus incentives. Panel C
reports the composition of the sample over industries. Panel D reports the Pearson correlations.

Panel A: Descriptive statistics (full sample)

Variable Mean Std.Dev. 10% 25% 50% 75% 90%

EM 0.000 0.738 1.141 0.512 0.013 0.541 0.943

OI 0.000 0.887 1.035 0.619 0.004 0.663 1.087
INC 22.743 37.412 0.000 0.000 15.000 25.000 50.000
INC_FIN 13.441 29.827 0.000 0.000 0.000 16.000 35.000
INC_NONFIN 8.947 22.179 0.000 0.000 0.000 12.500 25.000
INF_ASYM 0.000 0.939 1.236 0.649 0.070 0.769 1.262
ENVIRON_VOL 0.000 0.939 1.420 0.578 0.201 0.652 1.054
SIZE 7.206 2.422 4.414 5.298 7.003 9.210 10.404
GROWTH 5.678 17.792 10.000 2.000 3.000 10.000 20.000
PE_OWN 18.333 36.338 0.000 0.000 0.000 0.000 100.000
FAMILY_OWN 23.251 40.596 0.000 0.000 0.000 30.000 100.000
JOB_HORIZON 89.623 26.071 50.000 99.000 100.000 100.000 100.000
QUAL_CPA 0.443 0.498 0.000 0.000 0.000 1.000 1.000
QUAL_CMA 0.393 0.489 0.000 0.000 0.000 1.000 1.000
REL_SUP 3.016 2.904 1.000 1.000 2.000 4.000 6.000
JOB_TENURE 3.951 4.214 1.000 1.000 3.000 5.000 10.000
OVERCONF 7.158 2.295 4.000 6.000 8.000 9.000 10.000
AGE 41.819 7.796 32.000 36.000 40.000 47.000 53.000
GENDER 0.842 0.366 0.000 1.000 1.000 1.000 1.000
EM denotes the degree of earnings manipulation, INC denotes the degree of incentive compensation, INC_FIN denotes the degree of incentive pay contingent on financial
measures, INC_NONFIN denotes the degree of incentive pay contingent on nonfinancial measures, OI measures the degree in which the respondent identifies him- or
herself with the organization, INF_ASYM denotes the degree of information asymmetry between respondent and the hierarchical superior, ENVIRON_VOL denotes the
degree of environmental turbulence, SIZE is the natural logarithm of number of employees working in the organization (measured in full-time equivalents), GROWTH
denotes the degree of growth options and is measured by the change in sales in the current year relative to the prior year, PE_OWN is measured by the percentage of
shares owned by private equity firms, FAMILY_OWN is measured by the percentage of shares owned by the founding family, JOB_HORIZON denotes the likelihood
estimated by the respondent that (s)he will be in the current job at the end of the year, QUAL_CPA is an indicator variable equal to one if the respondent is a CPA, zero
otherwise, QUAL_CMA is an indicator variable equal to one if the respondent is a CMA, zero otherwise, REL_SUP is measured by the number of years that the respondent
reports to the current superior, JOB_TENURE is measured by the number of years that the respondent has been in its current position, OVERCONF denotes the degree of
overconfidence of the respondent, AGE is measured by the age of the respondent (in full years), and GENDER is an indicator variable equal to one if the respondent is
male, zero otherwise.

Panel B: descriptive statistics (by lower vs. higher organizational identity)

OI_HIGHER ¼ 0 OI_HIGHER ¼ 1 Difference tests

Variable N Mean Median N Mean Median Meana Medianb

EM 92 0.09 0.13 91 0.09 0.09 * *

INC 92 22.05 15.00 91 23.43 15.00
INC_FIN 92 14.36 0.00 91 12.51 0.00
INC_NONFIN 92 7.53 0.00 91 10.38 0.00
INF_ASYM 92 0.19 0.19 91 0.19 0.35 *** ***
ENVIRON_VOL 92 0.01 0.24 91 0.01 0.07
SIZE 92 7.28 7.13 91 7.12 6.68
GROWTH 92 4.97 4.00 91 6.38 2.00
PE_OWN 92 18.04 0.00 91 18.63 0.00
FAMILY_OWN 92 24.13 0.00 91 22.36 0.00
JOB_HORIZON 92 91.85 100.00 91 87.36 100.00
QUAL_CPA 92 0.47 0.00 91 0.40 0.00
QUAL_CMA 92 0.38 0.00 91 0.40 0.00
REL_SUP 92 2.87 2.00 91 3.16 2.00
JOB_TENURE 92 3.46 3.00 91 4.45 3.00
OVERCONF 92 7.33 8.00 91 6.99 7.00
AGE 92 40.67 40.00 91 42.97 42.00 ** **
GENDER 92 0.81 1.00 91 0.86 1.00
a
Significance levels based on t-test. b Significance levels based on Wilcoxon rank-sum (Mann-Whitney) test. ***, **, * corresponds to 1%, 5%, and 10% significance levels
(two-tailed). EM denotes the degree of earnings manipulation, INC denotes the degree of incentive pay, INC_FIN denotes the degree of incentive pay contingent on
financial measures, INC_NONFIN denotes the degree of incentive pay contingent on nonfinancial measures, OI measures the degree in which the respondent identifies
him- or herself with the organization, OI_HIGHER is an indicator variable equal to one if OI is higher than the median value, zero otherwise, INF_ASYM denotes the
degree of information asymmetry between respondent and the hierarchical superior, ENVIRON_VOL denotes the degree of environmental turbulence, SIZE is the natural
logarithm of number of employees working in the organization (measured in full-time equivalents), GROWTH denotes the degree of growth options and is measured by
the change in sales in the current year relative to the prior year, PE_OWN is measured by the percentage of shares owned by private equity firms, FAMILY_OWN is
measured by the percentage of shares owned by the founding family, JOB_HORIZON denotes the likelihood estimated by the respondent that (s)he will be in the current
job at the end of the year, QUAL_CPA is an indicator variable equal to one if the respondent is a CPA, zero otherwise, QUAL_CMA is an indicator variable equal to one if
the respondent is a CMA, zero otherwise, REL_SUP is measured by the number of years that the respondent reports to the current superior, JOB_TENURE is measured by
the number of years that the respondent has been in its current position, OVERCONF denotes the degree of overconfidence of the respondent, AGE is measured by the
age of the respondent (in full years), and GENDER is an indicator variable equal to one if the respondent is male, zero otherwise.

Panel C: Sample composition over industries (full sample and by lower vs. higher identity)

Full sample OI_HIGHER ¼ 0 OI_HIGHER ¼ 1

Agricultural, hunting, and fishing 9 (5%) 5 (5%) 4 (4%)

Traditional manufacturing 21 (11%) 12 (13%) 9 (10%)
High-tech manufacturing 16 (9%) 9 (10%) 7 (8%)
 M.A. Abernethy et al. / Accounting, Organizations and Society 58 (2017) 1e14 7

Table 1 (continued )

Panel C: Sample composition over industries (full sample and by lower vs. higher identity)

Full sample OI_HIGHER ¼ 0 OI_HIGHER ¼ 1

Production, distribution and sales of gas, electricity, or water 7 (4%) 4 (4%) 3 (3%)
Construction and building 5 (3%) 4 (4%) 1 (1%)
Consumer goods and retail 16 (9%) 5 (5%) 11 (12%)
Hotels, restaurants, and bars 1 (1%) 1 (1%) 0 (0%)
Transportation, logistics, warehousing and communication 24 (13%) 11 (12%) 13 (14%)
Financial institutions 29 (16%) 13 (14%) 16 (18%)
Real estate and professional services 12 (7%) 8 (9%) 4 (4%)
Public government and social security 8 (4%) 4 (4%) 4 (4%)
Educational services 5 (3%) 1 (1%) 4 (4%)
Health services 19 (10%) 7 (8%) 12 (13%)
Environment, culture, recreation, and other services 11 (6%) 8 (9%) 3 (3%)
Total 183 (100%) 92 (100%) 91 (100%)
OI measures the degree in which the respondent identifies him- or herself with the organization, OI_HIGHER is an indicator variable equal to one if OI is higher than the
median value, zero otherwise.

Panel D: Pearson correlations

Variable 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19.

1. EM 1.00
2. OI 0.10 1.00
3. INC 0.10 0.01 1.00
4. INC_FIN 0.10 0.03 0.80 1.00
5. INC. NONFIN 0.03 0.04 0.60 0.02 1.00
6. INF_ASYM 0.16 0.19 0.10 0.09 0.05 1.00
7. ENVIRON_VOL 0.23 0.07 0.02 0.05 0.04 0.05 1.00
8. SIZE 0.06 0.10 0.15 0.02 0.21 0.07 0.03 1.00
9. GROWTH 0.02 0.09 0.11 0.08 0.08 0.13 0.03 0.02 1.00
10. PE_OWN 0.08 0.02 0.02 0.02 0.01 0.09 0.07 0.13 0.07 1.00
11. FAMILY_OWN 0.04 0.06 0.12 0.17 0.03 0.01 0.01 ¡0.31 0.12 ¡0.24 1.00
12. JOB HORIZON ¡0.20 0.05 0.03 0.02 0.08 0.05 0.05 ¡0.21 0.10 0.04 0.06 1.00
13. QUAL_CPA ¡0.27 0.04 0.01 0.01 0.01 0.07 0.01 0.01 0.09 0.04 0.08 0.13 1.00
14. QUAL_CMA 0.18 0.03 0.09 0.03 0.09 0.09 0.02 0.24 0.07 0.02 ¡0.18 ¡0.25 ¡0.62 1.00
15. REL_SUP 0.06 0.13 0.06 0.13 0.08 0.00 0.11 ¡0.17 0.03 0.13 0.22 0.05 0.01 0.12 1.00
16. JOB_TENURE ¡0.16 0.15 0.02 0.11 0.11 0.07 0.13 ¡0.17 0.05 0.08 0.24 0.00 0.07 ¡0.19 0.55 1.00
17. OVERCONF 0.06 0.01 0.02 0.10 0.12 0.05 0.06 0.00 0.11 0.06 0.00 0.11 0.03 0.06 0.06 0.03 1.00
18. AGE 0.11 0.20 0.11 0.10 0.05 0.14 0.01 0.05 0.03 0.04 0.03 0.04 0.11 ¡0.20 0.21 0.34 0.03 1.00
19. GENDER 0.05 0.11 0.15 0.08 0.13 0.07 0.08 0.05 0.04 0.19 0.12 ¡0.17 0.07 0.13 0.10 0.09 ¡0.20 0.20 1.00
Correlations that are significant at a 5% significance level or lower are reported in bold.
EM denotes the degree of earnings manipulation, INC denotes the degree of incentive pay, INC_FIN denotes the degree of incentive pay contingent on financial measures,
INC_NONFIN denotes the degree of incentive pay contingent on nonfinancial measures, OI measures the degree in which the respondent identifies him- or herself with
the organization, INF_ASYM denotes the degree of information asymmetry between respondent and the hierarchical superior, ENVIRON_VOL denotes the degree of
environmental turbulence, SIZE is the natural logarithm of number of employees working in the organization (measured in full-time equivalents), GROWTH denotes the
degree of growth options and is measured by the change in sales in the current year relative to the prior year, PE_OWN is measured by the percentage of shares owned
by private equity firms, FAMILY_OWN is measured by the percentage of shares owned by the founding family, JOB_HORIZON denotes the likelihood estimated by the
respondent that (s)he will be in the current job at the end of the year, QUAL_CPA is an indicator variable equal to one if the respondent is a CPA, zero otherwise,
QUAL_CMA is an indicator variable equal to one if the respondent is a CMA, zero otherwise, REL_SUP is measured by the number of years that the respondent reports to
the current superior, JOB_TENURE is measured by the number of years that the respondent has been in its current position, OVERCONF denotes the degree of
overconfidence of the respondent, AGE is measured by the age of the respondent (in full years), and GENDER is an indicator variable equal to one if the respondent is
male, zero otherwise.

While the average value of their maximum performance-based in-
centives amounts to 22.87% of the base salary, the average value of
the actual performance-based pay that respondents received was
12.38%. We also asked respondents the performance measures used
to determine the performance-based compensation. We determined
the weight on financials (e.g., market returns, profit, cost measures,
asset and liability accounts) and the weight on nonfinancial mea-
sures (e.g., internal process measures) (Abernethy, Bouwens, & Van
Lent, 2004; 2013). For each respondent, we determined incentive
compensation tied to financial measures (INC_FIN) by multiplying
the maximum performance-based compensation by the aggregate
weight on financial measures. For example, when incentives are 50%
and the total weight on financial measures in performance-based
compensation is 80%, this implies that managers at the beginning
of the period know that improving reported financial performance
will at most lead to performance-based compensation that is 40% of
their fixed compensation. We assume that our measure INC_FIN
captures the ex-ante incentives of managers to increase financial
performance. Likewise, we determine incentive compensation tied
to nonfinancials (INC_NONFIN) by multiplying the maximum
performance-based compensation by the aggregate weight on
nonfinancial measures.
3.4. Control variables
We measure information asymmetry (INF_ASYM) using a six-
item scale developed by Dunk (1993) and used repeatedly in the
literature (e.g., Abernethy et al., 2004; 2013). When information
asymmetry is high, superiors lack sufficient information to
adequately monitor managers' actions, which may give rise to
earnings manipulation. For example, organizational complexity,
measured by geographical and product-line diversification, limits
the transparency of firm's operations and activities to investors,
which may lead to increased moral hazard problems due to the
more severe information gap between managers and investors
(Bushman, Chen, Engel, & Smith, 2004). We document satisfactory
convergent validity, given the correlations with the experience of
the respondent (r ¼ 0.15, p < 0.05), and the delegation of decision
rights (r ¼ 0.16, p < 0.05). We capture environmental volatility
(ENVIRON_VOL) using the Khandwalla’s (1972) six-item
8 M.A. Abernethy et al. / Accounting, Organizations and Society 58 (2017) 1e14
instrument that asks the respondent to describe the rate of change
in their environment. Graham et al. (2005) provides evidence
consistent with the notion that CFOs have a preference for
smoother earnings compared to the more volatile cash flows that
initially originate from their business. A more stable earnings path
(i.e. less volatile) is perceived by investors to be less risky, makes it
easier to predict future performance, and assures suppliers that
the business is stable. We assess the convergent validity by
examining the time that is devoted to new product and market
development. We find a positive and significant correlation
(r ¼ 0.14, p < 0.1). GROWTH denotes the amount of growth op-
tions and is measured by the percentage change in the current
annual sales relative to the prior year sales. It is argued that
when a greater fraction of firm value is due to growth options
(instead of assets in place), it becomes more difficult for super-
visors to assess the appropriateness of managers' actions as it
requires knowledge about available investment opportunities
(Smith & Watts, 1992). Cheng and Warfield (2005) document
how growth firms are more likely to meet or just beat analyst
forecasts, indicative of earnings manipulation. SIZE denotes firm
size and is measured by the natural logarithm of the number of
employees working for the company (in full-time equivalents).
Large firms are characterized by more complex operations indi-
cated by a greater number of products, more (international)
input and output markets, and globally dispersed production
facilities. This complexity increases the information processing
demands and therefore the difficulty in monitoring managers'
actions. Prior evidence also demonstrates that large firms are less
likely to miss analyst forecasts, indicative of earnings manipu-
lation (Cheng & Warfield, 2005).
Prior research has shown that financial expertise of financial
executives may affect the degree of earnings manipulation (Aier,
Comprix, Gunlock, & Lee, 2005; Abernethy et al., 2013). QUAL_-
CPA is an indicator variable, which is equal to one if the financial
controller has a CPA qualification, zero otherwise. QUAL_CMA is an
indicator variable, which is equal to one if the financial controller
has a CMA qualification, zero otherwise. In addition, individuals
may possess different identities where professional identity rep-
resents the cognitive connection between individuals and their
profession. Much of the research within the field of accounting
considers professional identity as the extent to which the individ-
ual is bound by rules of professional conduct, including the
requirement to maintain independence and examines whether this
allows accountants to exercise objectivity and professional skepti-
cism (Bauer, 2015). Accountants are said to exhibit professional
identification separate from employment at a specific audit firm as
individuals keep their affiliation and certification after leaving
(Bamber & Iyer, 2002). Suddaby, Gendron, & Lam (2009) show how
chartered accountants that left public practice still exhibit rela-
tively high values of professional identity. In our paper, QUAL_CPA
proxies for professional identity with the accounting profession.
We measure the number of years that the respondent is in her
current position (JOB_TENURE) and the number of years that the
respondent reports to her current superior (REL_SUP) to control
for two potentially confounding effects. First, greater experience
in the current position translates into a greater understanding of
financial reporting treatments unique to the organization and its
industry. Aier, Comprix, Gunlock, and Lee (2005) expect and find a
negative relation between CFO experience and manipulation.
Second, a recent stream of literature suggests that financial ex-
ecutives are pressured by powerful superiors into manipulation
(e.g., Feng et al., 2011; Friedman, 2014). A financial executive hired
during the CEO's tenure is expected to have less power relative to
the CEO, and more powerful superiors can be more successful in
pressuring financial executives into misreporting (Friedman,
2014). We proxy for monitoring quality by measuring the per-
centage of shares held by the founding family (FAMILY_OWN) and
the percentage of shares held by private equity firms (PE_OWN).
Wang (2006) and Ali, Chen, and Radhakrishnan (2007) find evi-
dence of higher quality earnings in firms with founding family
ownership because of their ability to directly monitor the man-
agers and because their intimate knowledge of the business ac-
tivities enables them to detect misreporting. In a similar vein, Katz
(2009) documents that because of the professional ownership,
tighter monitoring, and reputational considerations exhibited by
private equity sponsors, private equity-backed firms generally
have higher earnings quality (e.g., less EM).
We measure the current job horizon as a proxy for horizon
concerns by the estimated likelihood that the respondent will retain
the current job at the end of the year (JOB_HORIZON). Managers with
a short expected employment horizon who are rewarded on earn-
ings may be more likely to manipulate earnings to increase short-
term performance (Dechow & Sloan, 1991). Prior research has
shown that overconfident managers are more likely to exhibit an
optimistic bias and therefore are more prone to move down the
slippery slope of intentional misstatements (Schrand & Zechman,
2012). We measured overconfidence (OVERCONF) using an instru-
ment developed by Russo and Schoemaker (1989, p. 71) where re-
spondents were asked to provide a 90-percent confidence interval
on ten general-knowledge questions with numerical answers (such
as, ‘what is the length of the Nile river?’ or ‘what is the year in which
Wolfgang Amadeus Mozart was born?’). Given that respondents
were asked to provide upper and lower bounds with a 90-percent
probability of being right, on average respondents should be wrong
about 10% of the times. OVERCONF denotes the number of times (out
of 10 questions) in which our respondents provide upper and lower
bounds that do not include the right answer. AGE denotes the age of
the financial controller and GENDER is an indicator variable which is
equal to one if the financial controller is male, zero otherwise. De-
mographic characteristics such as age and gender are often associ-
ated with cognitive styles such as the degree of risk-aversion and
overconfidence and have been argued to be antecedents of EM (e.g.,
Ge et al., 2011). Finally, we control for industry effects through the
inclusion of industry dummies. We distinguish membership in 15
different industries (see Table 1, panel C).
3.5. Factor analysis
We adopt a reflective measurement model where latent con-
structs are manifested by a set of indicators. Empirical indicators are
expected to covary, and the covariation among indicators is attributed
to their common cause (latent construct). Given the covariation
among indicators, a change in a construct is expected to manifest itself
across observations in a change in all indicators (e.g., increased
earnings manipulation is expected to manifest itself in increased
delays of necessary expenditures, greater discounts to accelerate
sales, etc.) (Bisbe, Batista-Foguet, & Chenhall, 2007). We use explor-
atory factor analysis to develop our latent variables.14 We use factor
analysis with oblique rotation and retain factors with an eigenvalue
greater than unity (Nunnally & Bernstein, 1994). Variables are
measured using seven-point Likert-type scales, where all Likert-scale
variables are standardized for analysis. We conduct factor analysis on
all independent variables jointly and separately. The results are
14
With factor analysis, items can be described as a function of a small number of
latent factors and one specific factor for each item (which can partially be inter-
preted as measurement error). The variation not attributable to specific factors
must be accounted for by common factors. So, it only considers the common or
shared variance.
 M.A. Abernethy et al. / Accounting, Organizations and Society 58 (2017) 1e14
reported in the Appendix for our joint factor analysis. We construct
our composite variables using factor scores. The correlations between
our variables using factor scores and summated scales range between
0.98 and 0.99 for our four composite variables (not tabulated).
Panel A of the Appendix describes the summary statistics on the
separate items and the factor loadings for each item in our depen-
dent variable (EM).15 The average factor loading is 0.56 and the
reliability of our survey instrument is sufficient given a Cronbach's
alpha of 0.64.16 In sum, this indicates that EM represents a unidi-
mensional construct. Panel B describes the summary statistics on the
separate items, the factor loadings, and cross-loadings for our
explanatory variables, organizational identity (OI), information
asymmetry (INF_ASYM), and environmental volatility (ENVIRON_-
VOL). The results suggest that the constructs are unidimensional and
exhibit a clean factor structure as each item loads on the factor it is
theoretically associated with and not significantly on any other fac-
tor. The average factor loadings for OI, INF_ASYM, and ENVIRON_VOL
are 0.60, 0.74, and 0.69 respectively. The cross-loadings are small;
this suggests that our variables feature good discriminant validity.
The reliability of our explanatory variables exceeds the accepted
levels, given the Cronbach's alpha coefficients for OI, INF_ASYM, and
ENVIRON_VOL of 0.76, 0.88, and 0.84 respectively.
To further substantiate the validity of our factor analysis out-
comes, we, first, examine the composite reliability of our con-
structs. The composite reliability is defined as the squared sum of
the standardized loadings divided by the total of the squared sum
of standardized loadings and the sum of the error variances on each
item. The composite reliability is close to the recommended level of
about 0.70 (Hair, Anderson, Tatham, & Black, 2014) for one of our
constructs (EM with 0.64) and exceeds this level for our other three
latent constructs. Second, we examine the variance extracted, i.e.,
the amount of variance in the items captured by the construct
(Fornell & Larcker, 1981).17 The variance extracted ranges between
0.38 and 0.55, indicating that for most of the constructs close to half
of the variance in our items or more is accounted for by the pro-
posed constructs. In addition to the use of multiple alternative
measures to establish the convergent validity, we also assess the
discriminant validity of our constructs. We examine for each of our
independent variables whether the variance extracted exceeds the
covariance between the constructs (Fornell and Larcker, 1981). Our
constructs show strong discriminant validity given that for any pair
of our constructs, the variance extracted for each of these two
constructs strongly exceed the covariance.18
Finally, we employ a two-step procedure to examine whether
our results are vulnerable to common rater bias (Harman, 1967).
First, we undertake an exploratory factor analysis with all items
associated with the variables and find no indication that one factor
accounts for the majority of the covariance among the items. The
first factor accounts for 37% of the variance and all four factors
15
We report the factor pattern matrix showing the unique contribution of each
item to the factor (instead of the factor structure matrix which also includes the
correlations among factors). For convenience, we report the summary statistics on
the unstandardized items.
16
The Cronbach's alpha represents the average inter-item correlation as items of a
construct should measure the same construct and thus be highly correlated. It
reflects the reliability (internal consistency) of a latent variable composed of mul-
tiple items. Values of about 0.70 or higher are preferable (Lattin et al., 2003).
17
In confirmatory factor analysis, variance extracted is computed as the sum of
the squared standardized loadings divided by the total number of items.
18
The average variance extracted should be larger than the covariance of a
construct with any of the other constructs. Here, the covariances are 0.05, 0.07,
and 0.18 respectively.
19
Note that each model is tested against the null hypothesis that the proposed
model fits as well as a perfect model such that a significant reduction in the chi-
square statistic implies a significant better fit with the data.
9
account for more than 90% of the variance. Second, we find that our
four latent factors provide a significantly better fit to the data,
relative to a one-factor model (Dc2 ¼ 732.0, p < 0.01).19 Overall, this
suggests that common rater bias is not driving our results.
3.6. Econometric models
To test our hypotheses we propose variations of the following
regression model:
EM ¼ b0 þ b1INC_FIN þ b2INC_FIN*OI þ b3OI þ b4INF_ASYM þ
b5ENVIRON_VOL þ b6GROWTH þ b7SIZE þ b8FAMILY_OWN þ
b9PE_OWN þ b10JOB_HORIZON þ b11OUAL_CPA þ b12OUAL_CMA
þ b13JOB_TENURE þ b14REL_SUP þ b15OVERCONF þ b16AGE þ
b17GENDER þ INDUSTRY_CONTROLS þ ε [1]
First, we examine a basic model that only takes into account the
main effect of performance-based incentives on earnings manipu-
lation, while controlling for both organizational identity as well as
other firm-level and respondent-level characteristics. Here, b1 de-
notes the association between performance-based financial in-
centives and earnings manipulation. Second, the full model
examines the joint impact of performance-based incentives and
organizational identity on the degree of earnings manipulation. The
relationship between the presence of incentives and earnings
manipulation is reflected by: b1 þ b2*OI.20 Our model thus includes
an interaction term to assess whether OI moderates the relation-
ship between incentives and manipulation. Based on our hypoth-
esis that OI mitigates the positive relationship between incentives
and earnings management, we expect that b2 < 0. In the full model,
main effects are included to partial out the lower-order effects
when testing for an interaction effect. As all survey-items are
standardized and we mean-center INC_FIN, the coefficient of the
main effects represent the effect of one variable at the (sample)
average of the other (Hartmann & Moers, 1999). These trans-
formations imply that b1 represents the effect of incentives on
manipulation at the sample average of OI.
To substantiate our findings, we also employ variations of a
second regression model:
EM ¼ b0 þ b1INC_FIN þ b2INC_FIN*OI_HIGHER þ b3OI_HIGHER þ
b4INF_ASYM þ b5ENVIRON_VOL þ b6GROWTH þ b7SIZE þ
b8FAMILY_OWN þ b9PE_OWN þ b10JOB_HORIZON þ b11OUAL_CPA
þ b12OUAL_CMA þ b13JOB_TENURE þ b14REL_SUP þ b15OVERCONF
þ b16AGE þ b17GENDER þ INDUSTRY_CONTROLS þ ε [2]
We perform a median-split and define an indicator variable
(OI_HIGHER) equal to one if the degree of organizational identity is
higher than the median value for the sample (OI_HIGHER ¼ 1 for 91
observations), zero otherwise (OI_HIGHER ¼ 0 for 92 observations).
Again, we examine a basic model that only takes into account the
main effect of performance-based financial incentives and the full
model with both main effects and the interaction effect. In equation
(2), we define two subgroups contingent on whether they score
higher or lower on OI. In the full model, the relationship between
incentives and manipulation for subjects with lower (higher) OI is
given by b1 (b1 þ b2). By implication, the coefficient b2 represents
the difference in the relationship between incentives and mis-
reporting for those agents characterized by higher OI vis- a-vis
lower OI. We expect b2 < 0.
20
The relation between organizational identity and manipulation is represented
by: b3 þ b2*INC_FIN.
10 M.A. Abernethy et al. / Accounting, Organizations and Society 58 (2017) 1e14
4. Empirical results
4.1. Descriptive statistics
In Table 1 we present descriptive statistics and some pre-
liminary evidence of relations among our main variables of interest.
Table 1, panel A reports the descriptive statistics for the full sample.
The mean level of incentives represents about 23% of respondents’
base salary. The main part of the performance-based compensation
is tied to financial measures (13.4%). On average, the firms in our
sample employ about 1350 employees (expressed in full-time
equivalents) and 18 (23) percent of their equity is owned by
founding families (private equity parties). The mean growth in
annual sales is about 6%, 84% of the respondents are male, 44 (39)%
of the respondents have a CPA (CMA) qualification, and the average
age of the respondents is 42 years. On average, respondents have
been working for four years in their current job, while reporting to
their current superior for three years.
Table 1, panel B provides the descriptive statistics for the sub-
samples of respondents with a lower versus higher degree of
organizational identity. We report the mean and median values for
the subsamples of lower (OI_HIGHER ¼ 0) and higher organiza-
tional identity (OI_HIGHER ¼ 1), and performance difference-tests
based on t-tests and Wilcoxon rank-sum tests. Panel B shows that
respondents characterized by a higher degree of organizational
identity are older, work in settings characterized by a higher degree
of information asymmetry, and exhibit more misreporting.
Table 1, panel C reports the composition of the sample across
industries. Our sample appears to be concentrated in financial in-
stitutions; transportation, logistics, warehousing, and communi-
cation; and traditional manufacturing. We also report the degree of
OI across industries. Notwithstanding some small differences, we
do not observe notable differences between industries in the de-
gree to which respondents identify with the firm. Finally, Table 1,
panel D reports the Pearson correlations (with significant relations
at 5% or lower in bold). Our univariate analysis indicates that there
is no significant relation between OI and earnings manipulation
(p ¼ 0.18), nor between INC_FIN and earnings manipulation
(p ¼ 0.19). We also find an insignificant relation between OI and
INC_FIN. We will further explore the relation between financial
incentives and earnings manipulation, as well as the moderating
role of OI in our multivariate hypotheses tests.
4.2. Does organizational identity moderate earnings manipulation?
Table 2 reports the estimates of robust regressions of equation [1].
Robust regressions exclude observations with Cook's D > 1 and sub-
sequently address the potential impact of remaining influential ob-
servations through the downward weighting of such observations in
an iterative process to avoid bias of the regression estimates. Leone,
Minutti-Meza, and Wasley (2013) compare various methods
employed by accounting researchers (e.g., winsorizing) to address any
bias that may originate from unusual observations and conclude that
robust regressions are most appropriate. We employ models where
we control for both firm-level and respondent-level characteristics, as
well as industry-effects. The first model examines the interaction with
the continuous specification of organizational identity (OI), while the
second model uses the dichotomous specification of identity
(OI_HIGHER). In both models, the left column presents the results of
the main effect of financial incentives on earnings manipulation. In
the right column of each of the two models, we focus on how OI
moderates the relation between performance-based incentives and
the degree of earnings manipulation.
For our first model, we observe that the presence of incentive
contracts tied to financial measures is positively associated with
earnings manipulation as the coefficient on INC_FIN is positive and
significant for both models.21 This result is consistent with our first
hypothesis and confirms prior research documenting a positive
relation between financial executives’ incentives and earnings
management (e.g., Jiang et al., 2010). In the full model (including
the interaction term), the coefficient on INC_FIN is again positive
and significant.22 Our second hypothesis predicts that OI will
negatively moderate the positive relationship between
performance-based incentives and earnings manipulation. The
coefficient on the interaction term OI*INC_FIN is negative and sig-
nificant (p < 0.01), indicating that OI mitigates the relation between
financial incentives and earnings manipulation. In other words,
organizational identity appears to address the agency losses asso-
ciated with misreporting when financial controllers are rewarded
on the basis of financial measures. For our second model with the
dichotomous specification of OI, we, again, find evidence consistent
with our hypotheses. For the full model (including the interaction),
the coefficient on INC_FIN reflects the relation between financial
incentives and EM for those with lower OI. This coefficient is pos-
itive and significant (p < 0.01). INC_FIN*OI_HIGHER reflects the
moderating role of organizational identity. That is, the coefficient
on INC_FIN*OI_HIGHER represents the difference in the relation
between financial incentives and EM between higher OI relative to
lower OI. As expected, the coefficient is negative and significant
(p < 0.01); that is, higher OI seems to mitigate the incentive-driven
misreporting. The sum of coefficients on INC_FIN and INC_FI-
N*OI_HIGHER represents the relation between financial incentives
and manipulation for those with higher OI. An F-test for the sum of
coefficients is insignificant (p ¼ 0.45), which indicates that higher
values of OI address the incentive-driven opportunistic manipula-
tion. Finally, we do not find a significant effect of OI and OI_HIGHER
on EM. This may be consistent with the observation that
performance-based compensation plans represent the most widely
discussed determinant of opportunistic earnings manipulation.
Consistent with prior literature, we find significant results on
some of our control variables. For example, in the Graham et al.
(2005) study, 97% of the financial executives indicated that they
prefer to smooth earnings; one interviewed executive said that
‘businesses are more volatile than what their earnings numbers
would suggest.’ Indeed, we find that both environmental volatility
and information asymmetry are positively associated with the de-
gree of earnings manipulation. We also find that employees holding
a CPA qualification are less likely to engage in earnings manipula-
tion. This is consistent with findings of Aier et al. (2005), as well as
in accordance with prior research on professional identity that
focused on the cognitive connection between public accountants
and the rules of professional conduct embedded in the accounting
profession.23 Finally, the results suggest that employees who expect
to leave the firm are more likely to resort to misreporting.
We run a number of (non-tabulated) robustness tests. First, we
21
In the left hand column of model 1, the coefficient on INC_FIN is interpreted in a
straightforward manner; however, the coefficient on INC_FIN for the right hand
column (that allows for a moderating effect) should be interpreted at the sample
average of OI. See also section 3.4.
22
In the full model, the coefficient on INC_FIN should be evaluated at the sample
average of OI. The positive and significant coefficient here suggests that average
values of OI do not address the incentive-driven misreporting.
23
Note that professional identity is expected to have a direct negative impact on
earnings manipulation (regardless of managers' incentives) given their connect-
edness with the rules of professional conduct of the accounting profession. On the
contrary, OI is not directly aimed at earnings manipulation, but given the
connectedness to their firm, is expected to play a role in the presence of incentives
as incentives induce managers to take actions that benefit them at the expense of
the firm. Thus, OI is expected to negatively moderate the relation between in-
centives and EM.
 M.A. Abernethy et al. / Accounting, Organizations and Society 58 (2017) 1e14 11

Table 2
Does organizational identity moderate earnings manipulation?
This table reports the estimates of robust regressions of variations of the following models:
EM ¼ b0 þ b1INC_FIN þ b2OI*INC_FIN þ b3OI þ CONTROLS þ ε (1),
EM ¼ b0 þ b1INC_FIN þ b2OI_HIGHER*INC_FIN þ b3OI_HIGHER þ CONTROLS þ ε. (2).

EM

Model 1 Model 2

Variables Pred.

INC_FIN (þ) 0.004** 0.004*** 0.004** 0.008***

(<0.01) (<0.01) (0.02) (<0.01)
OI*INC_FIN () e 0.004*** e e
(<0.01)
OI () 0.092 0.101 e e
(0.86) (0.90)
OI_HIGHER*INC_FIN () e e e 0.007***
(<0.01)
OI_HIGHER () e e 0.137 0.158
(0.89) (0.93)
INF_ASYM 0.139** 0.135** 0.139** 0.126**
(0.02) (0.04) (0.02) (0.03)
ENVIRON_VOL 0.185*** 0.184*** 0.178*** 0.177***
(<0.01) (<0.01) (<0.01) (<0.01)
SIZE 0.002 0.002 0.003 0.005
(0.93) (0.95) (0.91) (0.83)
GROWTH 0.001 0.001 0.001 0.002
(0.82) (0.75) (0.73) (0.61)
PE_OWN 0.001 0.001 0.001 0.002
(0.43) (0.32) (0.42) (0.30)
FAMILY_OWN 0.000 0.000 0.000 0.000
(0.92) (0.92) (0.88) (0.84)
JOB_HORIZON 0.004** 0.005** 0.005** 0.004**
(0.02) (0.03) (0.03) (0.04)
QUAL_CPA 0.459*** 0.433*** 0.449*** 0.410***
(<0.01) (<0.01) (<0.01) (<0.01)
QUAL_CMA 0.118 0.091 0.120 0.082
(0.46) (0.50) (0.37) (0.54)
REL_SUP 0.011 0.016 0.012 0.018
(0.59) (0.34) (0.47) (0.30)
JOB_TENURE 0.026* 0.028** 0.026** 0.029**
(0.02) (0.02) (0.02) (0.01)
OVERCONF 0.024 0.021 0.020 0.018
(0.29) (0.36) (0.39) (0.45)
AGE 0.008 0.012 0.009 0.011
(0.26) (0.12) (0.24) (0.14)
GENDER 0.000 0.003 0.028 0.053
(0.99) (0.98) (0.86) (0.74)
INTERCEPT 1.025** 1.093** 0.943* 0.919*
(0.04) (0.02) (0.05) (0.06)
Industry controls Yes Yes Yes Yes
Test (b1 þ b2) ¼ 0 e e e (0.44)
N 183 183 183 183
The p-values of robust regressions are reported in parentheses. ***, **, * denotes 1%, 5%, and 10% significance levels (one-tailed when coefficient sign is predicted, two-tailed
otherwise). EM denotes the degree of earnings manipulation, INC denotes the degree of incentive pay, INC_FIN denotes the degree of incentive pay contingent on
financial measures, OI measures the degree in which the respondent identifies him- or herself with the organization, OI_HIGHER is an indicator variable equal to one if
OI is higher than the median value, zero otherwise, INF_ASYM denotes the degree of information asymmetry between respondent and the hierarchical superior,
ENVIRON_VOL denotes the degree of environmental turbulence, SIZE is the natural logarithm of number of employees working in the organization (measured in full-
time equivalents), GROWTH denotes the degree of growth options and is measured by the change in sales in the current year relative to the prior year, PE_OWN is
measured by the percentage of shares owned by private equity firms, FAMILY_OWN is measured by the percentage of shares owned by the founding family,
JOB_HORIZON denotes the likelihood estimated by the respondent that (s)he will be in the current job at the end of the year, QUAL_CPA is an indicator variable equal to
one if the respondent is a CPA, zero otherwise, QUAL_CMA is an indicator variable equal to one if the respondent is a CMA, zero otherwise, REL_SUP is measured by the
number of years that the respondent reports to the current superior, JOB_TENURE is measured by the number of years that the respondent has been in its current
position, OVERCONF denotes the degree of overconfidence of the respondent, AGE is measured by the age of the respondent (in full years), AGE is measured by the age of
the respondent (in full years), and GENDER is an indicator variable equal to one if the respondent is male, zero otherwise.

run the same models but now replace INC_FIN by INC_NONFIN. The
relation between INC_NONFIN and EM is nonsignificant in both
models (p ¼ 0.63 and p ¼ 0.76). In addition, the coefficient on the
interactions INC_NONFIN*OI and INC_NONFIN*OI_HIGHER are both
insignificant (p ¼ 0.69 and p ¼ 0.61). Second, we assess to what
extent the results are similar when we do not address the potential
impact of influential observation by means of robust regressions. We
replicate our earlier analyses using OLS and find similar results (non-
tabulated). For our full models, the coefficient on INC_FIN *OI and
INC_FIN*OI_HIGHER are both negative and significant
(coefficient: 0.004, p ¼ 0.04 and coefficient: 0.007, p ¼ 0.03
respectively). Third, we repeat the same analyses but now each time
replace our dependent variable EM by one of the three survey in-
dicators. Out of six models, INC_FIN is positive and significant in all
models and the interaction coefficient INC_FIN*OI_HIGHER) is
negative and significant in five out of six models (with the insignif-
icant model reporting a negative coefficient for INC_FIN*OI_HIGHER,
p ¼ 0.17). Fourth, our respondents are financial controllers who are
operating at different hierarchical levels. To address the potential
that our results are mainly driven by observations from the corporate
12 M.A. Abernethy et al. / Accounting, Organizations and Society 58 (2017) 1e14
or business unit level, we include an indicator variable (CORP_LEVEL)
that is equal to one if the respondent operates at the corporate level,
zero otherwise. Again, our findings are similar with regard to the
sign, magnitude, and significance levels of our variables of interest.
That is, the coefficients on the interaction term for both the
continuous and dichotomous specification of OI is 0.004 (p < 0.01)
and 0.007 (p < 0.01). The coefficient on CORP_LEVEL is not signif-
icant (p-values ranging between 0.70 and 0.78).
Fifth, given that all respondents have responsibilities for financial
control, but have various job titles, we include an indicator variable
(OTHER_JOBTITLE) equal to one if the respondent did not have a job
title equal to CFO, financial director, group controller, business-unit
controller, controller, or finance manager. Our inferences are not
altered. The coefficient on INC_FIN*OI is 0.004 (p < 0.01) and the
coefficient on INC_FIN*OI_HIGHER is 0.007 (p < 0.01). The coeffi-
cient on OTHER_JOBTITLE is not significant (p-values exceed 0.72).
Sixth, we repeat the analysis but include a proxy for financial
distress. We define an indicator variable DISTRESS equal to one if the
firm in the lowest decile on the profit change in the current year
relative to the prior year (i.e., profit decreased by 25% or more), zero
otherwise. We run the same models, but now include DISTRESS as an
additional control variable. The sign, magnitude and significance
levels of the coefficients of interest are not affected. For example, the
coefficients on INC_FIN*OI and INC_FIN*OI_HIGHER are both nega-
tive and significant (coefficient: 0.004, p < 0.01 and
coefficient: 0.007, p < 0.01). The coefficient on DISTRESS is positive
but insignificant for both full models (p ¼ 0.15 and p ¼ 0.14).
5. Discussion and conclusions
Our objective in this study was to assess how an individual's
identity with the firm influences their propensity to engage in
opportunistic earnings manipulation to increase the proceeds from
their incentive contracts. Considerable research has looked at the
question of what motivates earnings management. It describes the
role of incentive contracts and the extent to which managers ex-
ercise their financial reporting discretion opportunistically. The
question becomes what options does a firm have to minimize self-
interested behavior? Consistent with the extant literature we as-
sume and find that incentives are related to misreporting. But this
does not mean that incentives should be abandoned altogether. Our
findings suggest that when incentives become stronger, agents on
average are more likely to misreport. But not all agents misreport.
Apparently other motives can mitigate the inclination to manipu-
late earnings in the presence of incentives. Then the question is;
can firms create conditions that motivate agents not to misreport?
Drawing on intuition provided by Akerlof and Kranton (2000, 2005,
2008, 2010), and empirical findings from the management litera-
ture (Boivie et al., 2011) our findings indicate that organization
identity can mitigate the agency losses associated with incentives.
In other words, the relation between incentives and manipulation
is less positive when individuals strongly identify with the firm.
This suggests that identification with the organization provides
agents with motives to refrain from misreporting even when
monetary incentives are strong.
Our study has implications for practice and for future research.
There continues to be considerable interest from practitioners, reg-
ulators and accounting academics on the determinants of mis-
reporting. Researchers are now going beyond the traditional
economic explanation of earnings manipulation to examine how
individual characteristics such as honesty; leadership style; CEO
integrity; overconfidence, physical or biological factors influence
accounting reporting choices (Bamber et al., 2010; Dikolli et al., 2016;
Jia et al., 2014; Schrand & Zechman, 2012). We contribute to this
nascent literature by attempting to better understand how an
individual's identity with the firm influences this behavior. Our
findings suggest that firms need not necessarily do away with
incentive-based pay. They can invest in selection, recruitment and
socialization processes to ensure that individuals are selected and
socialized so that they identify with the firm. Prior research has
demonstrated empirically and theoretically the importance of se-
lection processes in reducing agency costs (Abernethy et al., 2015;
Campbell, 2012; Prendergast, 2011). Our study suggests that the
firm can mitigate agency costs associated with the use of incentive
contracts when their employees identify with the firm. Our findings
also indicate that a firm with employees who identify with their firm
can increase their use of incentive contracts as OI will mitigate the
potential for agents to go down the slippery slope (Schrand &
Zechman, 2012).
That said, we only offer a first exploration of how identity and
incentives relate to the phenomenon of earnings manipulation. In
the first place we examine this relation for a specific type of
employee, namely the financial controller. Given that financial
controllers through their education and professional association are
subject to a code of conduct that precludes misreporting, the extent
of earnings manipulation may be different for employees with
different backgrounds. We also are not able to discern benign
within-GAAP earnings management from more aggressive ac-
counting practices; although the findings of Schrand and Zechman
(2012) suggest that we should prevent any form of EM as there is a
nontrivial chance that it will lead to accounting fraud. Differenti-
ating between types of earnings management would provide an
opportunity to increase our knowledge of the working of incentive
contracts. While financial incentives may deter agents from outside-
GAAP manipulation (Armstrong et al., 2010; Mazar, Amir, & Ariely,
2008), this may be less so for within-GAAP manipulation. Howev-
er, it is still an open question whether individuals themselves can
always distinguish between outside GAAP manipulation from more
benign actions (e.g., Bazerman & Banaji, 2004). Furthermore, our
study is focused on organization identity where individuals have a
neutral or positive identification with the firm. Our findings do not
extrapolate to cases where individuals disconnect themselves from
the organization. Nonetheless, these cases may be less prevalent as
individuals with a negative identification are more likely to seek
employment in another organization.
In addition, some of the items in our empirical measure of
earnings manipulation specifically focus on income-increasing ac-
tions (e.g., postponing necessary expenditures). While we agree
that managers may also engage in income-decreasing actions, prior
research documents that most earnings management is income
increasing.24 Notwithstanding these caveats we believe that this
study adds to our knowledge of the behavioral explanations of
earnings manipulation.
Appendix. Descriptives and statistical properties of latent
constructs
Panel A reports the summary statistics and factor loadings of the
three survey items that measure the degree of earnings manipu-
lation. All items are measured on a scale from 1 (never occurs) to 7
(occurs frequently). Panel B reports the summary statistics and
cross loadings of all survey items that measure our latent con-
structs: identity, information asymmetry, and environmental
volatility. Items for identity are measured on a scale from 1
(strongly disagree) to 7 (strongly agree), items for information
asymmetry are measured on a scale from 1 (my superior is much
24
For example, Dichev et al. (2013) finds that their surveyed CFOs believe that
about 60 percent of within-GAAP manipulation is income-increasing.
 M.A. Abernethy et al. / Accounting, Organizations and Society 58 (2017) 1e14 13

better informed) to 7 (I am much better informed), and items for

environmental volatility are measured on a scale from 1 (highly
stable, infrequent change) to 7 (highly volatile, constant change).

Panel A
Summary statistics and factor loadings on earnings manipulation (EM).

Mean Std.Dev. Min. Max. Factor pattern

Please indicate how often your unit postpones necessary expenditures to shift future profits into the current period. 4.03 1.56 1 7 0.51
Please indicate how often your unit accelerates sales to shift future profits into the current period. 2.96 1.62 1 7 0.60
Please indicate how often your unit shifts funds between accounts to avoid budget overruns. 2.56 1.48 1 7 0.58
Cronbach alpha: 0.64

Panel B
Summary statistics and cross loadings on latent independent variables.

Summary Cross-loadings
statistics

Mean Std.Dev. 1 2 3

1. Organizational identity (OI)

When someone makes positive remarks about my organization, it feels like a personal compliment. 5.32 1.14 0.63 0.06 0.01
When someone criticizes my organization, it feels like a personal insult, even if I do not know the person. 4.66 1.40 0.59 0.01 0.00
I am very interested in what people think about my organization. 5.29 1.19 0.61 0.04 0.08
When I talk about my organization, I often say ‘we’ rather than ‘they’. 6.03 1.04 0.47 0.08 0.01
This organization's successes are my successes. 5.11 1.21 0.70 0.03 0.04
If a story in the media criticizes my organization, I feel embarrassed. 4.56 1.42 0.57 0.06 0.01
Cronbach alpha: 0.76
2. Information asymmetry (INF_ASYM)
Relative to your superior, who is better informed about the activities of your unit? 5.15 1.61 0.02 0.72 0.06
Relative to your superior, who is better informed about input-output relations inherent in the internal operations of your unit? 5.28 1.39 0.03 0.69 0.09
Relative to your superior, who is more certain about the performance potential of your unit? 4.93 1.49 0.03 0.81 0.01
Relative to your superior, who is better able to estimate the impact of external conditions on the activities of your unit? 4.41 1.50 0.00 0.76 0.06
Relative to your superior, who has a better understanding of what can be achieved within your unit? 4.84 1.51 0.05 0.80 0.10
Relative to your superior, who is more familiar, in a technical fashion, with the work of your unit? 5.25 1.69 0.08 0.63 0.11
Cronbach alpha: 0.88
3. Environmental volatility (ENVIRON_VOL)
What is the rate of change in the buying patterns and requirements of customers? 4.54 1.56 0.00 0.00 0.82
What is the rate of change in distributors' attitudes? 3.62 1.62 0.11 0.02 0.64
What is the rate of change in industry buying patterns? 4.23 1.61 0.05 0.02 0.83
What is the rate of change in competitor strategies? 4.37 1.51 0.06 0.08 0.80
What is the rate of change in technical developments relevant to your unit's business? 4.32 1.52 0.10 0.13 0.49
What is the rate of change of the changes in the (service) production process? 4.35 1.48 0.09 0.04 0.56
Cronbach alpha: 0.84

References management mechanisms. The Accounting Review, 86(5), 1491e1518.

Badertscher, B. A., Collins, D. W., & Lys, T. Z. (2012). Discretionary accounting choices
and the predictive ability of accruals with respect to future cash flows. Journal of
Abernethy, M. A., Bouwens, J., & Van Lent, L. (2004). Determinants of control system
Accounting and Economics, 53(1), 330e352.
design in divisionalized firms. The Accounting Review, 79(3), 545e570.
Bamber, E. M., & Iyer, V. M. (2002). Big 5 auditors' professional and organizational
Abernethy, M. A., Bouwens, J., & Van Lent, L. (2013). The role of performance
identification: Consistency or conflict? Auditing: A Journal of Practice & Theory,
measures in the intertemporal decisions of business unit managers. Contem-
21(2), 21e38.
porary Accounting Research, 30(3), 925e961.
Bamber, L. S., Jiang, J., & Wang, I. Y. (2010). What is my style? The influence of top
Abernethy, M. A., Dekker, H., & Schultz, A. (2015). Are employee selection and
managers on voluntary corporate financial disclosure. The Accounting Review,
incentive contracts complements or substitutes? Journal of Accounting Research,
85(4), 1131e1162.
53(4), 633e668.
Bauer, T. D. (2015). The effects of client identity strength and professional identity
Aboody, D., & Kasznik, R. (2000). CEO stock option awards and the timing of
salience on auditor judgments. The Accounting Review, 90(1), 95e114.
corporate voluntary disclosures. Journal of Accounting and Economics, 29(1),
Bazerman, M. H., & Banaji, M. R. (2004). The social psychology of ordinary ethical
73e100.
failures. Social Justice Research, 17(2), 111e115.
Adler, P. S., & Chen, C. X. (2011). Combining creativity and control: Understanding
Benabou, R., & Tirole, J. (2016). Bonus culture: Competitive pay, screening, and
individual motivation in large-scale collaborative creativity. Accounting, Orga-
multitasking. Journal of Political Economy, 124(2), 305e370.
nizations and Society, 36(1), 63e85.
Beneish, M. D., & Vargus, M. E. (2002). Insider trading, earnings quality, and accrual
Aier, J. K., Comprix, J., Gunlock, M. T., & Lee, D. (2005). The financial expertise of
mispricing. The Accounting Review, 77(4), 755e791.
CFOs and accounting restatements. Accounting Horizons, 19(3), 123e135.
Bergstresser, D., & Philippon, T. (2006). CEO incentives and earnings management.
Akerlof, G. A., & Kranton, R. E. (2000). Economics and identity. The Quarterly Journal
Journal of Financial Economics, 80(3), 511e529.
of Economics, 115(3), 715e753.
Bhojraj, S., Hribar, P., Picconi, M., & McInnis, J. (2009). Making sense of cents: An
Akerlof, G. A., & Kranton, R. E. (2005). Identity and the economics of organizations.
examination of firms that marginally miss or beat analyst forecasts. The Journal
Journal of Economic Perspectives, 19(1), 9e32.
of Finance, 64(5), 2361e2388.
Akerlof, G. A., & Kranton, R. E. (2008). Identity, supervision, and work groups.
Bisbe, J., Batista-Foguet, J. M., & Chenhall, R. (2007). Defining management ac-
American Economic Review, 98(2), 212e217.
counting constructs: A methodological note on the risks of conceptual mis-
Akerlof, G. A., & Kranton, R. E. (2010). Identity economics. Princeton: Princeton
specification. Accounting, Organizations and Society, 32(7), 789e820.
University Press.
Boivie, S., Lange, D., McDonald, M. L., & Westphal, J. D. (2011). Me or we: The effects
Ali, A., Chen, T. Y., & Radhakrishnan, S. (2007). Corporate disclosures by family firms.
of CEO organizational identity on agency costs. Academy of Management Journal,
Journal of Accounting and Economics, 44(1), 238e286.
54(3), 551e576.
Armstrong, C. S., Jagolinzer, A. D., & Larcker, D. F. (2010). Chief executive officer
Bushman, R., Chen, Q., Engel, E., & Smith, A. (2004). Financial accounting infor-
equity incentives and accounting irregularities. Journal of Accounting Research,
mation, organizational complexity and corporate governance systems. Journal
48(2), 225e271.
of Accounting and Economics, 37(2), 167e201.
Badertscher, B. A. (2011). Overvaluation and the choice of alternative earnings
Campbell, D. (2012). Employee selection as a control system. Journal of Accounting
14 M.A. Abernethy et al. / Accounting, Organizations and Society 58 (2017) 1e14
Research, 50(4), 931e966.
Chava, S., & Purnanandam, A. (2010). CEOs versus CFOs: Incentives and corporate
policies. Journal of Financial Economics, 97(2), 263e278.
Cheng, Q., & Warfield, T. D. (2005). Equity incentives and earnings management. The
Accounting Review, 80(2), 441e476.
Chow, C. W., Kato, Y., & Merchant, K. (1996). The use of organizational controls and
their effects on data manipulation and management myopia: A Japan vs US
comparison. Accounting, Organization, and Society, 21(2), 175e192.
Dechow, P. M., Ge, W., & Schrand, C. (2010). Understanding earnings quality: A
review of the proxies, their determinants and their consequences. Journal of
Accounting and Economics, 50(2e3), 344e401.
Dechow, P. M., & Skinner, D. J. (2000). Earnings management: Reconciling the views
of accounting academics, practitioners, and regulators. Accounting Horizons,
14(2), 235e250.
Dechow, P. M., & Sloan, R. G. (1991). Executive incentives and the horizon problem.
Journal of Accounting and Economics, 14(1), 51e89.
Deci, E. L. (1971). Effects of externally mediated rewards on intrinsic motivation.
Journal of Personality and Social Psychology, 18(1), 105e115.
Dichev, I. D., Graham, J. R., Harvey, C. R., & Rajgopal, S. (2013). Earnings quality:
Evidence from the field. Journal of Accounting and Economics, 56(2), 1e33.
Dikolli, S., Keusch, T., Mayew, W. J., & Steffen, T. D. (2016). A linguistic-based approach
to measuring innate executive traits: The case of CEO integrity. Working paper.
Duke University, INSEAD, and Yale University.
Dukerich, J. M., Golden, B. R., & Shortell, S. M. (2002). Beauty is in the eye of the
beholder: The impact of organization identity, and image on the cooperative
behaviors of physicians. Administrative Science Quarterly, 47(3), 507e533.
Dunk, A. S. (1993). The effect of budget emphasis and information asymmetry on
the relation between budgetary participation and slack. The Accounting Review,
68(2), 400e410.
Dutton, J. E., Dukerich, J. M., & Harquail, C. V. (1994). Organizational images and
member identity. Administrative Science Quarterly, 39(2), 239e263.
Feng, M., Ge, W., Luo, S., & Shevlin, T. (2011). Why do CFOs become involved in
material earnings managements? Journal of Accounting and Economics, 51(1),
21e36.
Fields, T. D., Lys, T. Z., & Vincent, L. (2001). Empirical research on accounting choice.
Journal of Accounting and Economics, 31(1), 255e307.
Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models with un-
observable variables and measurement error. Journal of Marketing Research,
18(1), 39e50.
Friedman, H. (2014). Implications of power: When the CEO can pressure the CFO to
bias reports. Journal of Accounting and Economics, 58(1), 117e141.
Geiger, M. A., & North, D. S. (2006). Does hiring a new CFO change things? An
investigation of changes in discretionary accruals. The Accounting Review, 81(4),
781e809.
Ge, W., Matsumoto, D., & Zhang, J. L. (2011). Do CFOs have Style? An empirical
investigation of the effect of individual CFOs on accounting practices. Contem-
porary Accounting Research, 28(4), 1141e1179.
Gioia, D. A., Price, K. N., Hamilton, A. L., & Thomas, J. B. (2010). Forging an identity:
An insider-outsider study of processes involved in the formation of organiza-
tional identity. Administrative Science Quarterly, 55(1), 1e46.
Graham, J. R., Harvey, C. R., & Rajgopal, S. (2005). The economic implications of
corporate financial reporting. Journal of Accounting and Economics, 40(1), 3e73.
The Guardian. (29 October 2014). Criminal investigation launched into Tesco's
accounting.
The Guardian. (22 September 2016). Former Tesco directors plead not guilty to fraud,
false accounting charges.
Guidry, F., Leone, A. J., & Rock, S. (1999). Earnings-based bonus plans and earnings
management by business-unit managers. Journal of Accounting and Economics,
26(1e3), 113e142.
Hair, G. F., Anderson, R. E., Tatham, R. L., & Black, W. C. (2014). Multivariate data
analysis. Prentice-hall international.
Harman, H. (1967). Modern factor analysis. Chicago: IL. University of Chicago Press.
Hartmann, F. G. H., & Moers, F. (1999). Testing contingency hypotheses in budgetary
research: An evaluation of the use of moderated regression analysis. Accounting,
Organizations and Society, 24(4), 291e315.
Healy, P. M. (1985). The effect of bonus schemes on accounting decisions. Journal of
Accounting and Economics, 7(1e3), 85e107.
Heinle, M. S., Hofmann, C., & Kunz, A. H. (2012). Identity, incentives and the value of
information. The Accounting Review, 87(4), 1309e1334.
Indjejikian, R., & Mate jka, M. (2009). CFO fiduciary responsibilities and annual
bonus incentives. Journal of Accounting Research, 47(4), 1061e1093.
Ittner, C. D., & Larcker, D. F. (2001). Assessing empirical research in managerial
accounting: A value-based management perspective. Journal of Accounting and
Economics, 32(1), 349e410.
Jensen, M. C., & Murphy, K. J. (2012). The earnings management game: It's time to stop
it. Working paper. Harvard Business School and University of Southern
California.
Jiang, J., Petroni, K. R., & Wang, I. Y. (2010). CFOs and CEOs: Who have the most
influence on earnings management? Journal of Financial Economics, 96(3),
513e526.
Jia, Y., Van Lent, L., & Zeng, Y. (2014). Testosterone and financial misreporting.
Journal of Accounting Research, 52(5), 1195e1246.
Katz, S. (2009). Earnings quality and ownership structure: The role of private equity
sponsors. The Accounting Review, 84(3), 623e658.
Khandwalla, P. (1972). The effect of different types of competition on the use of
management controls. Journal of Accounting Research, 10(2), 275e285.
Kim, J. B., Li, Y., & Zhang, L. (2011). CFOs versus CEOs: Equity incentives and crashes.
Journal of Financial Economics, 101(3), 713e730.
Kreiner, G. E., & Ashforth, B. E. (2004). Evidence toward an expanded model of
organizational identification. Journal of Organizational Behavior, 25(1), 1e27.
Lattin, J. M., Carrol, J. D., & Green, P. E. (2003). Analyzing multivariate data. Duxbury
Press.
Leone, A., Minutti-Meza, M., & Wasley, C. (2013). Influential observations and infer-
ence in accounting research. Working paper. University of Miami and University
of Rochester.
Mael, F., & Ashforth, B. E. (1992). Alumni and their alma mater: A partial test of the
reformulated model of organizational identity. Journal of Organization Behavior,
13(2), 103e123.
Malina, M. A., & Selto, F. H. (2001). Communicating and controlling strategy: An
empirical study of the effectiveness of the balanced scorecard. Journal of
Management Accounting Research, 13(1), 47e90.
Mazar, N., Amir, O., & Ariely, D. (2008). The dishonesty of honest people: A theory of
self-concept maintenance. Journal of Marketing Research, 45(6), 633e644.
Merchant, K. A. (1990). The effects of financial controls on data manipulation and
management myopia. Accounting Organizations and Society, 15(4), 297e313.
Moore, D., & Tarnai, J. (2002). Evaluating nonresponse error in mail surveys. In
R. M. Groves, D. A. Dillman, J. L. Eltinge, & R. J. A. Little (Eds.), Survey Nonre-
sponse. New York NY: John Wiley & Sons, 197-211.
Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory. New York, NY:
McGraw-Hill.
Prendergast, C. (2011). Contracts and conflict in organizations. Working paper. Uni-
versity of Chicago.
Ravasi, D., & Phillips, N. (2011). Strategies of alignment: Organizational identity
management and strategic change at Bang & Olufsen. Organization, 9(2),
103e135.
Roberts, J. (2010). Designing incentives in organizations. Journal of Institutional
Economics, 6(1), 125e132.
Roychowdhury, S. (2006). Earnings management through real activities manipu-
lation. Journal of Accounting and Economics, 42(3), 335e370.
Russo, J. E., & Schoemaker, P. J. H. (1989). Decision traps. UK: Doubleday Publishing
Co.
Schrand, C., & Zechman, S. (2012). Executive overconfidence and the slippery slope
to fraud. Journal of Accounting and Economics, 53(1e2), 311e329.
Simon, H. A. (1954). Centralization vs. decentralization in organizing the controller's
department. Scholars Book Co.
Smith, C. W., & Watts, R. L. (1992). The investment opportunity set and corporate
financing, dividend, and compensation policies. Journal of Financial Economics,
32(3), 263e292.
Suddaby, R., Gendron, Y., & Lam, H. (2009). The organizational context of profes-
sionalism in accounting. Accounting, Organization, and Society, 34(3e4),
409e427.
Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. In
W. G. Austin, & S. Worchel (Eds.), The social psychology of intergroup relations
(pp. 33e47). California: Brooks & Cole.
Van der Stede, W. A. (2001). Measuring ‘tight budgetary control’. Management Ac-
counting Research, 12(1), 117e139.
Van der Stede, W. A., Young, S. M., & Chen, C. X. (2005). Assessing the quality of
evidence in empirical management accounting research: The case of survey
studies. Accounting, Organization and Society, 30(7e8), 655e684.
Wang, D. (2006). Founding family ownership and earnings quality. Journal of Ac-
counting Research, 44(3), 619e656.
 Accounting, Organizations and Society 58 (2017) 15e31

Contents lists available at ScienceDirect

Accounting, Organizations and Society

journal homepage: www.elsevier.com/locate/aos

The Q&A: Under surveillance

Santhosh Abraham a, Matthew Bamber b, *
a
School of Business & Technology, Excelsior College, 7 Columbia Circle, Albany, NY 12203, USA
b
University of Toronto, Rotman School of Management, 105 St George Street, Toronto, Ontario, M5S 3E6, Canada

a r t i c l e i n f o a b s t r a c t

Article history: Drawing on theories of surveillance and interaction ritual, we explore the incentives (disincentives) to
Received 26 August 2015 analyst participation during the question-and-answer session (Q&A) which concludes firms’ results
Received in revised form presentations. Analysis of our qualitative data shows that interrogation strategies and behaviours are
31 March 2017
influenced by a combination of regulatory and ritual codes. Furthermore, the presence of surveillance
Accepted 8 April 2017
Available online 18 April 2017
technologies and networks exacerbate the risks and rewards faced by analysts during this interactive
information exchange. In turn, we find that the common conceptualisation of the Q&A as an ostensibly
economic event, underpinned by information retrieval, is overly simplistic. The gaze of surveillance
Keywords:
Surveillance
transforms the Q&A into a dramaturgical encounter, where impression management techniques are
Interaction ritual important. From this, we develop a descriptive framework to explain public interrogation strategies and
Impression management behaviours. Our work will help future researchers better understand investor-manager meetings.
Performance Furthermore, we propose that our descriptive framework has extensions to similar public interrogation
Results presentation settings.
Q&A © 2017 Elsevier Ltd. All rights reserved.
Qualitative research

1. Introduction synchronised by accounting period-ends. It provides a rare and

A firm's results presentation usually concludes with an analyst-
management question-and-answer session (Q&A). Strikingly,
despite the fundamental importance of analyst participation, little
is known about what incentivises or disincentivises their interro-
gation strategies and behaviours. Relatedly but unsurprisingly, the
literature is silent on what types of questions are asked, and why.
Brown, Call, Clement, and Sharp (2015) suggest that analysts pur-
posefully avoid asking questions in this forum for fear of revealing
private information to peers, but this explanation over-simplifies
the Q&A, which is a complex and ambiguous interactive
encounter. Indeed, analysis of our data shows evidence to the
contrary. Instead, we find that analysts are keen to be seen to ask
good questions. Our extensive interview and observational evi-
dence suggest that whilst economic motives are important, the
function of the Q&A goes far beyond information retrieval; it is also
a social and political event. As such, participation offers a set of
unique risks and rewards, costs and benefits.
The Q&A is underpinned by accounting information and
* Corresponding author.
E-mail addresses: SAbraham1@excelsior.edu (S. Abraham), matt.bamber@
rotman.utoronto.ca (M. Bamber).
timely opportunity for conversation between investors and man-
agement about the firm's position, performance and strategy
(Barker, 1998). Surveys determining which information sources are
used and valued by professional investors reveal that contact with
senior company management is crucial to their overall decision-
making; and in many cases, meetings with management are the
most important source of information (Barker, 1998; Cascino et al.,
2013; Clatworthy, 2005). It is unclear from the extant literature,
however, why this might be the case. Nonetheless, these claims
appear to be supported by findings which show that results pre-
sentations and the Q&A are economically valuable (e.g. Allee &
Deangelis, 2015; Davis, Ge, Matsumoto, & Zhang, 2014; Frankel,
Johnson, & Skinner, 1999; Li, Minnis, Nagar, & Rajan, 2014;
Matsumoto, Pronk, & Roelofsen, 2011; Mayew, Sharp, &
Venkatachalam, 2012). To address this issue, the focus of our
research is on interrogators' strategies and behaviour. Specifically,
our research question addresses what factors (dis)incentivise ana-
lyst participation during the Q&A.
It is often assumed that analysts are rational decision-makers in
the neo-classical economic sense and therefore would devote their
resources to undertaking tasks and attending events where useful
information might emerge (e.g. Brown et al., 2015; Fama, 1970;
Kothari, 2001). However, during a firm's results presentation,
managerial disclosure choices are bound and controlled by

http://dx.doi.org/10.1016/j.aos.2017.04.001
0361-3682/© 2017 Elsevier Ltd. All rights reserved.
16 S. Abraham, M. Bamber / Accounting, Organizations and Society 58 (2017) 15e31
complex social, structural and regulatory disciplinary mechanisms.
For example, fair disclosure regulation (Financial Conduct
Authority, 2014; Financial Services Authority, 2000, 1996)1 pro-
hibits management from releasing price-sensitive, inside-infor-
mation. Therefore, this creates the possibility of a conflict of
motives, i.e. why would an analyst choose to attend an event e at
this time-critical moment on results day e where (s)he is unlikely
to glean useful information (cf. Barker, Hendry, Roberts &
Sanderson, 2012,2; Brown et al., 2015). Yet, analysts compete to
attend these events (Mayew, 2008). To better understand analysts'
motivations during the Q&A, we draw on and extend several key
issues within the emerging sociologically-oriented meetings-with-
management literature.
This literature can be summarised into two strands: studies of
private meetings (e.g. Barker, Hendry, Roberts, & Sanderson, 2012;
Holland, 1998a,b; Roberts, Sanderson, Barker, & Hendry, 2006;
Solomon, Solomon, Joseph, & Norton, 2013); and studies of public
meetings (e.g. Carrington & Johed, 2007; Catasus & Johed, 2007,
2012; Johed & Catasus, 2015; Uche & Atkins, 2015). Following this
tradition we take into account the social and cultural context of
investor meetings “without falling back on the common fallacies of
economic analysis” (Carrington & Johed, 2007, p. 722). We identify
a unifying theme, namely that these are social situations in which
visibility offers participants transformative opportunities, which
are sometimes used to divert the encounter away from its intended
purpose.
Within the context of meetings-with-management, there are a
number of distinct features that make the Q&A stand apart. These
are chiefly related to the increased visibility arising from the
presence of surveillance technologies and networks. First, the
meetings literature is divided into private and public, but the Q&A
comprises aspects of both. A select group are invited to attend the
presentation in-person, but the real-time online broadcast is
available to everyone with access to the internet. Because of this
profound level of visibility, the Q&A has attracted a wide range of
onlookers who can peak behind the curtains of this one-time pri-
vate encounter. As a result, this is not only an opportunity for an-
alysts to pose questions, it is also a rare moment to render
themselves visible. Second, the Q&A occurs at a critical moment in
the financial year. It is the first opportunity after the close period3 to
hear from senior company management and enter dialogue with
them, thus accentuating the visibility of the event. Third, for some
sell-side analysts, this is their annual opportunity to observe and
question senior management face-to-face. Relatedly, this is also an
opportunity for onlookers to appraise the private relationships
between analysts and management. Finally, most meetings are
attended by institutional and private investors, but the physical
Q&A audience is made up almost entirely of sell-side analysts4 who
do not have ownership interest in the company and therefore have
little (Fogarty & Rogers, 2005) or no incentive to hold management
to account (Brown et al., 2015).
Using our data from interviews and observations, we draw on
1
Similar regulation governs US results presentations (conference calls)
(Securities and Exchange Commission (2000), Selective Disclosure and Insider
Trading).
2
Barker et al. (2012) refer to a ‘value-attendance paradox’. The conflict of motives
addressed in this current article is similar, but not identical to that.
3
The close period is the period between the completion of a listed company's
financial results and the announcement of these results to the public. Commonly,
this is a two-month period at the end of Q4.
4
We have the sense, however, that this is changing. Whilst attendance lists are
not published, we have attended recent results presentations where buy-side an-
alysts and hedge fund managers have been present, as well as other professional
service providers such as lawyers and accountants.
theories of surveillance (Foucault, 1979 [panoptic]; Deleuze &
Guattari, 1988 [rhizomatic]; Mathiesen, 1997 [synoptic]; Latour,
2005; Latour & Hermant, 1998 [oligoptic]) and interaction ritual
(Goffman, 1955, 1956, 1959, 1963, 1967) as a means of structuring
our data and communicating and constructing its significance
(Ahrens & Chapman, 2006). Through this approach, we are able to
explore two inter-related sociological issues: first, how the pres-
ence of surveillance transforms the Q&A into a dramaturgical
encounter; and second, how interrogators' behaviours and strate-
gies are modified as they manage others' perceptions of their self
(Goffman, 1955, 1956, 1959, 1967). As Vollmer (2007, p. 578) puts it,
when “numbers are performed”, there is more at stake. In so doing,
we demonstrate the importance of adopting a sociological
perspective to investor-manager meetings, and specifically that the
dual notions of surveillance (e.g. Foucault, 1979) and presentation
of self (Goffman, 1959) are crucial to understanding individuals’
behaviours and strategies during these encounters.
Our analysis allows us to make a series of important contribu-
tions to the literature. We demonstrate the various incentives for
analysts to ask ‘good’ questions and why they strive to do so, but
also that there are disincentives as well as deleterious conse-
quences to ‘poor’ questioning. Following this, we propose a
descriptive framework that clarifies what happens during the Q&A
and why it happens that way, which includes an identification of
the types of questions that analysts are likely to ask, why they ask
them and, to an extent, how they will ask them. Thereby we isolate
various key factors which affect the creation, production and
dissemination of information, and the potential implications that
publicly broadcasting interrogations has for information flows.
Thus, our work can be usefully extended to similar public interro-
gation settings such as Prime Minister's Questions, Presidential
debates, academic conferences, televised judiciary hearings, and so
forth.
The remainder of this article is organised as follows. We begin
by elaborating on the theoretical foundations that we briefly out-
lined above, namely surveillance and interaction ritual. We
consider the implications of these in the context of the Q&A, before
presenting the research design and methods of data collection and
analysis. We then present our empirical data, followed by a dis-
cussion of the theoretical and practical implications, from which
our descriptive framework emerges. Finally, we provide some
concluding remarks and recommendations for further study.
2. Literature and theoretical foundations
2.1. Meetings with management
As mentioned in the Introduction, the emerging sociologically-
oriented meetings-with-management literature can be divided
into two strands: private and public. Applying a Foucauldian lens,
Roberts et al. (2006) argue that one purpose of private investor
meetings is to subject executives to a field of visibility, hold the
mirror up to their stewardship and in this way, introduce another
layer of accountability. Although not discounting the accountability
angle, Barker et al. (2012) propose that the primary reasons for
attendance at private meetings are information clarification,
assurance, behavioural bias and marketing tactics. Similarly, a
number of studies (e.g. Solomon & Darby, 2005; Solomon &
Solomon, 2006; Solomon et al., 2013, 2011) have examined pri-
vate meetings from the perspective of social and environmental
reporting engagement. A central conclusion of this work is that
private meetings are a useful place to discuss, develop and enhance
the extent of a firm's social and environmental reporting, but there
is scant evidence that this leads to substantial change in corporate
behaviour. Rather, these meetings appear to be motivated by social
 S. Abraham, M. Bamber / Accounting, Organizations and Society 58 (2017) 15e31
and political motivations.
Public meetings, such as AGMs, are generally believed to be a
vehicle for stakeholders to hold companies to account and, corre-
spondingly, for management to display their accountability (e.g.
Jeacle, 2008). Interestingly, studies suggest that participants such as
shareholder associations frequently use public meetings to subvert
the agenda according to personal or group ends (Catasus & Johed,
2007, 2012; Johed & Catasus, 2015; Uche & Atkins, 2015).
Carrington and Johed’s (2007) study of AGMs found that, on the one
hand, questions can be challenging for management, partly because
of their breadth and unpredictability, but on the other hand, the
audience participation can be beneficial and desirable for senior
company officers and their advisors because it is a mechanism to
create or maintain legitimacy. Following on from this, Johed and
Catasus (2015) demonstrate how the Swedish Shareholders' Asso-
ciation “needed the AGM and the conflict at the AGM” (p. 119) as a
means to legitimise themselves. Similarly, Uche and Atkins (2015)
describe how investor groups employ gamesmanship and publicity
stunts to gain visibility during the AGM, and how this might also be
advantageous for these groups, as well as for the meetings
themselves.
Three broad themes emerge from this literature which have
important implications for our study. First, whilst information
retrieval is important, it is not the sole motive for participation.
Second, to assume that investor meetings are ostensibly economic
encounters is an over simplification. Third, the visibility afforded to
participants resulting from the presence of surveillance technolo-
gies and networks offers the opportunity to transform the intended
purpose of the meeting. Through the analysis of extensive quali-
tative data and the rigorous application of theory, our study ex-
plores these themes, demonstrates relevant implications, and
thereby extends current knowledge. Thus we now turn to a central
theme of this paper, the subjectivation of social actors to the field of
visibility.
2.2. Participation during meetings: visibility
Haggerty and Ericson (2006) define surveillance as the “collec-
tion and analysis of information about populations in order to
govern their activities” (p. 3). Through a process of subjectivation,
the questions asked by the hand-selected,5 invited and attendant
analysts render the senior management team visible. Surveillance
of this type is a prominent feature of modern society (e.g. Caluya,
2010; Elmer, 2012; Haggerty, 2006; Lyon, 2001), and as the tech-
nologies develop further, companies and their representatives are
subject to increasingly demanding and varied forms of it (Brivot &
Gendron, 2011), whereas until recently this would not have been
the case (Boll, 2014; Latour & Hermant, 1998; Latour, 2005).
Haggerty and Ericson (2000) call this effect the surveillant assem-
blage. They explain the transformative effects of the encroachment
of mass-surveillance into modern society, namely how this has
impacted on both “the purposes of surveillance and the hierarchies
of surveillance, as well as the institution of privacy” (p. 605).
There are several theories of surveillance, but Foucault’s (1979)
metaphor of the panopticon is the cornerstone. Through this lens,
scholars have sought to understand the nature, extent and purpose
of surveillance (cf. Caluya, 2010; Elmer, 2012; Haggerty & Ericson,
2000, 2006; Lyon, 1994, 2006). The panopticon is based on
Bentham’s (1995 [1843]) proposed institutional correctional facil-
ity, which places inmates into cells in a circular building around a
watchtower housing an unobserved guard who can observe any
5
Discussions indicate that most firms generally invite all the analysts who have
written research notes about the firm within the last 12 months.
17
inmate at any time. The central premise is that inmates can never
be certain whether they are being watched, and thus they behave as
though they always are. Foucault believed that the omnipresent eye
could facilitate compliance with accepted social norms and create
docile bodies. Elements of the panopticon are evident in the Q&A,
and given the constraints imposed by the fair disclosure regulation,
it is impossible to ignore how and why the underlying dual notions
of control and discipline encourage the social actors' behaviours
during this encounter.
However, Foucault's panoptic metaphor does not fully transpose
onto modern business settings, such as the Q&A, principally
because they are more complex and ambiguous situations (cf.
Brivot & Gendron, 2011). The panopticon is based on the notion that
‘the few watch the many’, whereas surveillance during this
encounter, as with most meetings, is multi-dimensional (e.g. Barker
et al., 2012; Catasus & Johed, 2012; Gabbioneta, Greenwood,
Mazzola, & Minoja, 2013; Uche & Atkins, 2015). A key challenge
to this metaphor is that it simplifies, most notably by focusing on
one target at a time (e.g. Bogard, 2006; Boyne, 2000; Brivot &
Gendron, 2011; Eckersley, Ferry, & Zakaria, 2014; Haggerty &
Ericson, 2000, 2006; Haggerty, 2006; Lyon, 2006; Mathiesen,
1997). To this end, a synoptic lens helps to explain the behaviour
of actors when ‘the many watch the few’ (Bogard, 2006; Mathiesen,
1997). This is useful because some interactions during public in-
terrogations, such as the Q&A, can be characterised in this way.
Notably, however, the Q&A comprises the full range of alternatives:
the few watching the many; the many watching the few; occa-
sionally, the few watching the few; and, sometimes, the many
watching the many (see Fig. 1). Therefore, whilst the Q&A shares
certain common elements with the panoptic metaphor and the
synopticon, and there are some straightforward target-agent re-
lationships, analysis of our data also reveals the existence of mul-
tiple layers of lateral and hierarchical surveillance, with certain
stakeholders simultaneously monitoring and being monitored in a
system of networks. We unravel these threads below and discuss
the potential relevance of alternative post-panoptic theories of
surveillance.
The Q&A also shares attributes consistent with an oligoptic
reading (Boll, 2014; Latour & Hermant, 1998; Latour, 2005). This
theory of surveillance is more aligned with the modern, complex,
digital world we live in. An oligopticon is a system of monitoring
that relies on a target being observed by multiple actors where each
social actor sees a minuscule, but nonetheless clear, translated
version of the observed (Boll, 2014; Latour, 2005). . Also consistent
with the Q&A, the oligopticon is described as fluid, where social
actors shift between target and agent. This is in contrast to the
guard in the panopticon's watch-tower, for example, who is the
prime and principal agent, with a singular and all-encompassing
view of the targets, each of whom are unable to shift their status.
A feature of the Q&A is the existence of a complex, ambiguous
but inter-connected web of networks. This description aligns with
the notion of rhizomatic surveillance (Deleuze & Guattari, 1988).
For example, and focusing on the interrogators, they are primarily
individuals promoting themselves and their research, as such they
are aware that they are being watched by the management team,
clients, current and potential employers, as well as others (hierar-
chical surveillance networks). In addition, they learn from one
another as a result of the interrogation (lateral surveillance net-
works) (e.g. Brown et al., 2015).
Equally relevant to the discussion of surveillance metaphors are
counter surveillance and resistance strategies which provide the
space for analysts to pursue self-interest motives (Brivot &
Gendron, 2011; Doyle, 2011; Gilliom, 2006; Lyon, 2006; Marx,
2003). If participating analysts believed that the surveillance
technologies and networks were present solely to keep them under
18 S. Abraham, M. Bamber / Accounting, Organizations and Society 58 (2017) 15e31

Fig. 1. A visualization of the typical Q&A highlighting directions of surveillance.

control during the Q&A, it is likely that this would disincentivise transformation into dramaturgical encounters. Therefore, we turn
meaningful, goal-oriented participation and affect future atten- next to a discussion of interaction ritual and impression manage-
dance. However, unlike with Foucault's prison inmates, the Q&A is ment techniques.
not forced subjectivation and visibility; the analyst attends and
participates more or less voluntarily. Indeed, analysts compete to
2.3. Participation during meetings: impression management
attend (Mayew, 2008).
Returning to Foucault, in later work he (1994, p. 298) argued
Surveillance and interaction ritual emerge as salient from our
that targets of surveillance could combat their sense of power-
data. These perspectives overlap and are complementary, but they
lessness e created by the disciplinary code and compulsion to
operate from different directions, namely mandatory versus
conform e by “think[ing] differently”, using alternative perspec-
voluntary, and compulsion versus consensuality. It would be wrong
tives and approaches to challenge the hierarchies (Deleuze, 1988;
to view firms’ Q&As as one-off stand-alone situations. Instead, as
Foucault & Blanchot, 1990; Foucault, 1984). To this end, Gilliom
with all social encounters, some aspects, attributes and character-
(2006) states: “The sheer ingenuity of many of these [counter-
istics are repeated between events (Collins, 2005). For example,
surveillance] measures makes resistance an intrinsically
participants seek to overcome the sense of powerlessness associ-
appealing topic” (p. 20). If it were not possible for the interrogator
ated with being the target of surveillance (Deleuze, 1988; Flynn,
to subvert the agenda, the Q&A might become an anodyne, proc-
1985; Foucault & Blanchot, 1990; Foucault, 1984, 1994;
essual, non-informative encounter. Instead, analysts take advan-
Spellmeyer, 1989). In this instance they do so by transforming the
tage of their frontstage opportunity and “contribute to the
event into a dramaturgical encounter and engaging in impression
transformation” and “re-direction” of the encounter (Brivot and
management techniques. Reputational capital is a key commodity
Gendron, p.137). Although Brivot and Gendron (2011) were
among analysts, and how each individual is perceived by others is
considering a different surveillance setting, there are parallels here.
critical to their success. Therefore, to maximise the value of
They argue that social actors’ behaviour can have “ramifications
attendance and participation, analysts are likely to employ
[for] the very mechanisms that aimed to originally undermine their
impression management techniques.
own autonomy at work” (p.137).
Thus, to provide depth and meaning to our analysis, we draw on
In summary, the Q&A is a complex and ambiguous surveillance
interaction ritual theory, which is fundamentally a theory of situ-
context where no single theoretical construct offers a full expla-
ations (see Bell, 1992). Although all situations vary and change, they
nation.6 No model fits perfectly, but nonetheless, the field of sur-
tend to repeat (Collins, 2005). This repetition allows participants to
veillance research holds the key to some crucial insights. We are
learn and ascribe social norms and expectations to the encounter
drawn to Foucault's work despite the inherent over-simplification
(Goffman, 1956, 1959, 1967). It also encourages actors to behave in
problem because every subsequent thesis is underpinned by its
certain ways such as engaging in ‘face-work’ and face-saving
core principles, namely that surveillance compels a form of social
games. Informed by ideas from surveillance, dramaturgical per-
control that, in turn, modifies the target's behaviour to conform to
formance and situational causality, we can explain how and why
social norms, and that the threat of disciplinary mechanisms allows
“one kind of thing” might happen during this social encounter
the construction of docile bodies (Caluya, 2010). We focus on these
“rather than another” (Collins, 2005; Goffman, 1959, 1967, p. 9).
aspects of surveillance theory to better understand the behaviours
Impression management allows an individual to present a ‘front’
and strategies of analysts during the Q&A. A notable consequence
(a face) to the world (Goffman, 1955, 1959, 1967). Therefore, in the
of the introduction of surveillance to these meetings is their
same way that an auditor needs to act according to societal ex-
pectations (Pentland, 1993), an analyst-interrogator must also play
6
his or her allotted role. Goffman (1955, 1956, 1959, 1967) believes
This is not uncommon in the surveillance literature. Surveillance theory(ies) is
(are) difficult to ‘test’ empirically, which is partly why there is a proliferation of
that, for a performance to be accepted by the audience, there must
post-panoptic surveillance alternatives; new ones emerge to fit varying empirical be a consensus around a ritual code, including roles, duties, re-
settings. sponsibilities, power and acceptable behaviour. Key to this is how
 S. Abraham, M. Bamber / Accounting, Organizations and Society 58 (2017) 15e31
performers establish their line during “legitimate institutionalized”
contact and how an individual's “attributes” alongside the “con-
ventionalized nature of the encounter” open up a limited choice of
line and face (1955, 1967, p. 7). During the Q&A, observers are
“asked to believe that the character they see actually possesses the
attributes he appears to possess, that the task he performs will have
the consequences that are implicitly claimed for it, and that, in
general, matters are what they appear to be” (Goffman, 1959, p. 28).
Goffman (1955, 1967) describes how an individual strives to
“have, or be in, or maintain face” while at the same time avoiding
“be[ing] in wrong face”, “be[ing] out of face” or “losing face” (pp.
6e9). Thus, face-work is employed to “counteract incidents e that is,
events whose symbolic implications threaten face” (Goffman, 1955,
1967, p. 12 [emphasis in original]). Goffman's interaction ritual
theory describes basic forms of face-work, namely the avoidance
process, which includes protective and defensive manoeuvres, and
the corrective process. Goffman also presents a framework for the
observer to understand and interpret face-saving practices,
aggressive face-work and co-operative face-work. We employ this
framework to explore the Q&A as a social encounter and to help
explain any impression management techniques used during it.
Although silence is the most basic form of ‘avoidance face-work’
(Goffman, 1955, 1967), the presence of multiple layers of lateral and
hierarchical surveillance make it difficult to hold one's tongue. The
competitive nature of the sell-side analyst role (Bradshaw, 2015;
Brown et al., 2015) implies that, should the analysts choose to
interact, it is important that they ask questions that are perceived to
be more insightful than those of their peers. This may improve their
social value within their existing network and facilitate access to
new points of contact, compared with poor questions, which may
have negative consequences. This competitive aspect of the ana-
lysts' occupation seems to rule out any form of team work between
individuals; however, we do identify some rare incidents of inter-
rogatory ‘team performance’ (Goffman, 1959) and ‘co-operative
face-work’ (Goffman, 1955, 1967). This may be because of the
ordering of activities in the ‘regulatory drama’ (Vollmer, 2007).
Exhibitionism might be considered an alternative interrogation
strategy. This behaviour has been identified among actors in other
accounting-related encounters (e.g. Brivot & Gendron, 2011).
Exhibitionism would be recognised by surveillance theorists as a
form of resistance (Gilliom, 2006; Marx, 2003) and referring to the
use of social media technologies, Koskela (2002) argues that “to be
(more) seen is not always to be less powerful … Sometimes it is
more radical to reveal than to hide” (p. 199).
Brown et al. (2015) conjecture e although not in these terms e
that analysts are reluctant to interrogate because it might be
deemed by management as aggressive face-work; and the analy-
stemanagement relationship is as valuable, and potentially more
so, than the information that might be ascertained. An unneces-
sarily aggressive challenge to the management team's line, or a
corresponding threat to the executives' ‘face’, would put the
interrogator in breach of the ritual code (Goffman, 1955, 1967).
Showing appropriate levels of deference and demeanour (Goffman,
1956, 1967) are the “basic rules of conduct” (p. 48) governing so-
cial interaction. An interrogator not displaying these attributes is
likely to be confronted (Catasus & Johed, 2012; Gabbioneta et al.,
2013; Uche & Atkins, 2015), and given the analysts' less powerful
position in the exchange, they may suffer incrementally more
harm. Thus, participants are unwilling to disrupt the ‘dramatic
truce’ (Vollmer, 2007).
Brivot and Gendron (2011) conclude their study of another
similarly complex context: “This leads us to surmise that Goffman’s
(1959) classic notions of front stage, backstage and impression
management are undeniably pertinent today in trying to under-
stand social life, in this age where sophisticated technologies of
19
communication and surveillance abound” (p. 153). Taking up this
challenge, we explore and extend current understanding of the
creation, production and dissemination of information during a
public interrogation that is bound by regulatory and ritual codes by
using the Q&A as an important interactive event in which ac-
counting information is communicated and translated to key
stakeholders.
3. Research methods
Little is known about the Q&A (Li et al., 2014; Matsumoto et al.,
2011) and even less about the intent, nature and purpose of ana-
lysts' interrogation strategies during it. Thus, it is important to
adopt a research method that allows sufficient flexibility to explore
issues without the constraints of availability and technique.
Therefore, we adopted a qualitative approach to lift the lid on the
‘black box’ of analyst research and decision-making (as described
by Brown et al., 2015). Our study involved discussions with tradi-
tionally hard-to-reach individuals (Pettigrew, 1992; Roberts et al.,
2006) to ensure a wide range of contributions. We also undertook
a series of observations because, as Pentland (1993) notes, “obser-
vational data are essential to provide an empirical basis for an
interpretative, contextual theory” (p. 605). The interviews and
observations enabled us to engage with the social realities that
influence analysts' behaviour and hence to take part in the ‘contact
zone’ described by Hastrup (1997) and Ahrens and Chapman
(2006).
Research of this nature can be affected by self-selection bias as
well as subsequent subjectivity in the data collection and analysis.
Conversely, it has the major advantage of allowing interviewees to
express their opinions on a wide range of (often predetermined)
issues, some of which had previously not been considered (Stoner
& Holland, 2004). It also allows the interviewer to ask related
questions and seek clarification and further explanation (Barker,
1998; Barker et al., 2012; Jones & Solomon, 2010; Roberts et al.,
2006).
We began by watching the online presentations of five FTSE 100
firms for the years 2005 through 2012 and analysing the accom-
panying documentation, which included presentation slides and
transcripts. These improved our understanding of results pre-
sentations in the UK. Following conversations with informed ac-
quaintances, friends and colleagues, we then formulated a letter to
the two key stakeholder groups involved in the investor pre-
sentations: sell-side analysts and senior company management. In
our letter, we drew attention to possible gaps in the extant litera-
ture and to our preliminary findings. Gathering evidence from the
primary communities of participants serves to substantiate and
strengthen the integrity and veracity of findings (Barker et al.,
2012).
We initially conducted a small pilot study including physical
observance of three presentations, three pilot interviews with se-
nior management and three pilot interviews with analysts (drawn
from the original sample of five firms), before contacting the
remainder of the sample. The format comprised semi-structured
interviews with general questions about the importance of the
presentation event, particularly the Q&A, as a means of creating,
producing and disseminating information. Although the same basic
questions were asked at each stage of the interviewing process, in
line with Solomon, Solomon, Norton, and Joseph (2011) and Jones
and Solomon (2010), they were subsequently modified in light of
new evidence. For example, a general question about the impor-
tance of the results presentations revealed a greater interest in the
Q&A than in the presentation, and therefore in subsequent in-
terviews we placed more emphasis on the Q&A.
Using information available on companyeinvestor relations
20 S. Abraham, M. Bamber / Accounting, Organizations and Society 58 (2017) 15e31
websites, we hand-collected details of preparers and analysts from
the UK's 100 largest companies and approached them via a letter
that explained the nature of the project. Twenty-four analysts from
the major investment research firms in London responded, and all
were interviewed. When we contacted the senior officials of the
sample companies (finance director, managing director, chairman),
we were directed almost without exception towards the firm's
investor relations manager as the principal point of contact and the
person most closely involved in the event.7 Where there was more
than one principal co-ordinating person responsible for the prep-
aration of the presentation, interviews were held either sequen-
tially or jointly, depending upon the preference of the interviewee.
In total, 35 interviews were conducted with senior company
managers and 24 with analysts. It is difficult for academics to gain
interview access to sell-side analysts and senior company man-
agement, because they are reluctant to talk to researchers owing to
time pressures, personal hesitancy and confidentiality (Armitage &
Marston, 2008; Johansson, 2007; Pettigrew, 1992; Roberts et al.,
2006). Despite this, we believe that the number of interviews we
conducted provides sufficient depth and rigour. Lincoln and Guba
(1985), among others (e.g. Power & Gendron, 2015), support this
by suggesting that it is important that the researcher does not
collect so much data that he or she feels overwhelmed, but collects
enough so that the research is ‘trustworthy’: “(…) it is usual to find
that a dozen or so interviews, if properly selected, will exhaust most
available information; to include as many as twenty will surely
reach well beyond the point of redundancy” (p. 235).
Before each interview, we watched the representative firm's
prior periods' presentations (minimum 1, maximum 7). After each
interview to which we were invited, we asked whether it could be
arranged for us to be present at their 2013 Q4 presentation. Where
this request could not be granted and/or the request could not be
followed through, we watched the presentation online, where
possible in real-time. However, because the presentations are
simulcast, recorded and subsequently transcribed, as a minimum
we were able to catch up and watch the presentations for every
firm in the sample. In total, we watched, reviewed the slides and
read the transcripts for more than 300 results presentations. These
observations improved our understanding of the broad range of
Q&A sessions within which to situate, locate and frame our inter-
pretation of the interview data.
The interviews were conducted face-to-face wherever possible
(n ¼ 22), but for practical reasons some were telephone-based
(n ¼ 37).8 Studies that have used both approaches include
Graham, Harvey, and Rajgopal (2005) and Armitage and Marston
(2008). All except two of the interviews were recorded and later
transcribed. In the case of two interviews where permission was
not granted to record, detailed notes were taken. In all cases, in-
terviewees were assured of anonymity, privacy and confidentiality.
Table 1 lists the interviewees’ industry sector and the mode (face-
to-face/telephone) of the interview (see Table 1).
We used the pilot interviews, our observations from the pre-
sentations, the interview data and theory to develop codes. Our
research question was the central driver of our analysis. However,
we did not adopt a linear process, but repeatedly moved back and
forth between research problem theory and data in an iterative
7
For example, the following email from the head of investor relations of a FTSE
100 company was typical of the response we received: “Both XX (CFO) and YY
(CEO) have passed on your letter requesting help on investor presentations. As
Head of Group Investor Relations I am very closely involved in this process and
would be delighted to help you with your questions”.
8 10
Our introductory letter clearly stated a preference to visit management on-site.
However, as a second-best alternative, we offered the option to conduct the
interview by telephone or other means.
fashion to develop our understanding of the data. This abductive
approach (Ahrens & Chapman, 2006; Eisenhardt & Graebner, 2007)
enabled us to better understand analysts’ motivations in asking, or
refraining from asking, questions during the Q&A. Dialogue with
peers was also important in generating fit between problem, theory
and data (Ahrens & Chapman, 2006, p. 827).9
The transcript analysis draws on a version of the staged
approach suggested by Easterby-Smith, Thorpe, Jackson and Lowe
(2008) and used in other interview-based research (Armitage &
Marston, 2008; Jones & Solomon, 2010; Solomon et al., 2011).
First, we read all of the transcripts to familiarise ourselves with the
general themes. Then, we micro-analysed the data to reveal codes
or themes10 from the interviews (Jones & Solomon, 2010, p. 22). We
adopted a similar approach to Solomon et al. (2011) and coded the
transcripts manually as well as electronically using the NVivo
software package. We initially took a sample of ten interviews (five:
analysts; five: preparers) and, independently from each other,
manually coded them. After discussion to resolve any differences or
disagreements, we used this analysis to generate an agreed coding
system. Several common key themes emerged between the analyst
and preparer interview data, and the findings section is presented
according to these. Following this, using NVivo, one of the re-
searchers analysed all of the analyst transcripts, while the other
analysed the preparer transcripts. Before these were exchanged, a
sample of five from each category were blind re-coded by the
opposite party, and any dissimilarities were discussed. Probably
because of our diligence in the earlier phases of the analysis, no
substantive disagreements arose. Nonetheless, in line with
Krippendorff’s (2004) advice, we undertook inter-coder reliability
checks. The number of noted differences was low, and results of the
kappa coefficient tests were above k ¼ 0.90. Where differences
arose, they were investigated, and if necessary, discussed and
adjusted.
We stop short of claiming that these are generalizable, and in
line with Barker et al. (2012) we do not assert that our process has
resulted in a single and unambiguous reading of the dynamism
involved in analysts' interrogation strategies. Nonetheless, the
common and recurring views supported by our observational data
enable us to identify key aspects of analysts’ decision-making
processes in the Q&A. We hope that future researchers will sub-
ject our framework to rigorous empirical testing in this and other
public interrogation contexts and challenge, develop and extend it.
Throughout the following discussion, case quotations are used as
illustrative examples of the empirical findings.
4. Analysis
4.1. Management under surveillance
We were told that one consequence of increased surveillance is
that the analyst community is more visible during the Q&A than
any other meeting, private or public. Being the target of the gaze of
surveillance might create docility among prison inmates (Foucault,
1979), but analysts can choose whether to attend and, employing
counter-surveillance strategies (e.g. Gilliom, 2006; Marx, 2003),
9
Ahrens and Chapman (2006) provide an excellent summary of this process:
“Like other practices, the doing of qualitative research is difficult to articulate. One
can point to the golden rules, but at the heart of it lies a process of transformation.
Out of data, snippets of conversations and formal interviews, hours and days of
observation, tabulations of behaviours and other occurrences, must arise the
plausible field study” (p. 837).
For example, one of the main codes was ‘Reasons for analysts asking questions’,
and under this one of the subcodes was ‘information seeking’, which had a further
subcode titled ‘confirmation’.
 S. Abraham, M. Bamber / Accounting, Organizations and Society 58 (2017) 15e31 21

Table 1
Basic Respondent Details. Sector coverage and mode of interviews.

Number of sell-side analysts Number of senior management

Sector coverage:
Financials 5 11
Telecoms 4 2
Retail 4 4
Media 3 4
Support services 2 6
Pharmaceuticals 1 2
Industrials and basic materials 1 3
Oil and gas 1 3
Head of equity research 3
Total 24 35
Mode of interviews:
Face-to-face interviews 11 11
Telephone interviews 13 24
Total 24 35

All the face-to-face interviews were conducted in London.

Table 2
An overview of information zones and related interrogation opportunities.

Definition Types of question

Zone A Common knowledge  Normally avoided

 Exceptions:
o where there are issues that management may be trying to dissimulate
o interrogator's naivety
Zone B Band of uncertainty  Confirmation
 Clarification
 Incoherence
 Revelation of sensitive issues
Zone C Prohibited territory  Normally avoided
 Exceptions:
o abnormal circumstances, e.g. fraud, error, crisis, scandal
o interrogator's naivety

can also make sophisticated decisions about whether, as well as
how, they participate. By asking a question, they become the agent
of the surveillance rather than the target (Brivot & Gendron, 2011;
Koskela, 2002) and thereby thrust themselves frontstage (Goffman,
1959). Our study shows that the interrogators engage with the rules
and systems, but in line with theoretical expectations they devise
strategies to pursue self-interest goals (Flynn, 1985; Foucault, 1994;
Spellmeyer, 1989). This is important because the legitimacy of the
performance and the acceptance of the performers’ narrative rely
on a consensus agreement that the institutional integrity of the
ritual is being upheld (Goffman, 1959; Vollmer, 2007). First,
therefore, we establish that the two key groups e interrogators and
interrogatees e fully understand where the regulatory boundary
markers are set, and specifically what questions can be answered
without breaching the legal line. To this end, the following extracts
are representative:
“If it was price-sensitive and they were releasing it to a privi-
leged group of people then they should not be doing that.”
(Analyst 24)
“All the information that you would consider to be price-
sensitive is released in the official announcement to the stock
exchange.” (Analyst 17)
“We cannot disclose in a small gathering information that is
potentially price-sensitive and that is not generally available to
the market. So there has to be a governance process around
that.” (IR 2)
Having confirmed this, we next discuss the three key forms of
questions that analysts ask management during the Q&A: (a) to
confirm, clarify and/or determine an incoherence; (b) to allow
management the opportunity to disclose information that might
otherwise be deemed inappropriate or unacceptable during the
formal results presentation; and (c) to encourage management to
disclose information they may otherwise be trying to dissimulate.
Each of these is considered in more detail below.
4.1.1. Confirmation, clarification and incoherence
Analysts claim not to be deliberately trying to entrap manage-
ment into releasing prohibited price-sensitive information. After
all, we were told that: firstly, it would be a futile endeavour given
the levels of managerial planning and preparation which are spe-
cifically devoted to not releasing this information during this highly
visible encounter, and secondly, that this strategy risks jeopardising
a valuable asset, namely the interrogator's relationship with the
firm. Thus, surveillance operates to control information flows by
subjecting management to a field of visibility which includes a
diverse range of stakeholders, such as regulators and the mass
media.
We were told that the Q&A is a social interaction and the
product of a bilateral exchange of information. Analysts ask
22 S. Abraham, M. Bamber / Accounting, Organizations and Society 58 (2017) 15e31

questions knowing that both they and the management team are
being observed, and management respond fully aware of this vis-
ibility. It is unsurprising that the management team e who spend
the majority of their working lives backstage e feel some degree of
anxiety about being frontstage (Goffman, 1959). For example:
“We'll work in advance. We'll go through the Q&A with them
[i.e. the management team]. We'll do dry runs. We wouldn't put
somebody up on the stage at the results presentation unless
they were comfortable and they had significant exposure to
that.” (IR 15)
Analysts take advantage of management's visibility and their
relative docility imposed by the exposure to surveillance (Foucault,
1979; Lyon, 2006; Roberts et al., 2006) by asking questions that
require confirmation and/or clarification of certain issues and as-
sumptions. The complementarity between theories of surveillance
and interaction ritual is exemplified here. Surveillance transforms
the event into a dramaturgical encounter as analysts employ
counter-surveillance strategies to wrest power from the situation
and use impression management techniques to create value. To be
effective, the audience must believe both the actor and the perfor-
mance and there must be agreement between the actors and
audience about the “realness” of the situation (Goffman, 1959, p.
28). This is a results presentation Q&A and must play out as such. In
addition, the actors' performance(s) should “incorporate and
exemplify the officially accredited values of the society” (Goffman,
1959, p. 45). Therefore, provided that the questions do not breach
either the legal and regulatory boundaries or the social and political
norms and expectations, management are expected to answer
them. These norms correlate closely with Goffman’s (1956, 1967)
principles of deference and demeanour.
Confirmatory and clarificatory questions exist for several rea-
sons. For instance, the interrogator has had to draw on his or her
own subjective interpretations of the earnings statement (the press
release) and the interrogator has simply not had the time and op-
portunity to review the results information before commencement
of the presentation. For example, we were told:
“It is probably in the details somewhere, but you have just not
had a chance … because it is one hell of a rush on the day. These
things [the press release] come out at 7 o'clock, you're giving
your first view on the microphone at the morning meeting at
7.30, you are talking to clients by 8, you rush off to the analysts'
meeting from 9 o'clock and you have not necessarily had the
chance to go through the detail at that point.” (Analyst 5)
Given that there is an expectation upon management to
respond, questions that require confirmation or clarification can be
helpful. However, it is striking that as useful as these questions and
their responses can be, we were told that an inability to confirm or
clarify was sometimes more valuable. Occasionally, a question of
this type provides an opportunity to recognise an inconsistency or
an incoherence in the managerial message. Thus, our qualitative
evidence supports and extends findings from recent quantitative
results presentations work. For example, Hollander, Pronk, and
Roelofsen (2010) show that managerial silence is a signal of bad
news, and others suggest that tone and non-verbal communication
are important (Allee & Deangelis, 2015; Davis et al., 2014; Li et al.,
2014; Mayew, 2008). We were told that these performance signals
are particularly useful to an analyst who has followed a firm for a
long time and has had time to study the behaviours and strategies
of the CEO and CFO. One respondent (Analyst 9) asked questions
“when things don't quite add up, or don't make sense … I like to see
how they react.”
Under the watchful gaze of ‘the many’ (Bogard, 2006; Brivot &
Gendron, 2011; Deleuze & Guattari, 1988; Haggerty, 2006; Lyon,
2006; Mathiesen, 1997), management find themselves account-
able for any promises they might make or have made in the past.
This further contributes to the credibility and legitimacy of the
encounter (Goffman, 1959). For some participants, the Q&A pro-
vides an opportunity to ‘punch holes’ in the management's ‘story’.
“A couple of years ago, many UK banks had 15% ROE targets. The
management thought they had to say they would achieve ROE of
15% or at least above the cost of equity (c. 10%) otherwise why
invest in the sector? But you could work backwards and look for
what that implied about revenue, costs, bad debts, and punch
holes in their story.” (Analyst 10)
Despite being rehearsed, management occasionally misjudge
their performance during the Q&A. Simple clarificatory questions
can lead to information opportunities, particularly when the
managerial message is inconsistent with market information or
expectations. For example, we were told that analysts' subjective
interpretations of managerial performance during the Q&A is
important for several reasons, and investors rely on sell-side ana-
lysts’ written research to inform their opinions (cf. Brown et al.,
2015).
“… Most famously [Company X's] management in the year-end
meeting in February/March 2002 in their answers to questions
showed they had no understanding about their exposure to
corporate credit risk. This was hugely price-sensitive e very
negative for the stock, but I heard from a friend who was their
corporate advisor that management walked away from the
meeting thinking it had gone well. Analysts had to piece
together the complacency of [Company X's] management, with
rumours about the low quality [Company X] wholesale bank,
with market concerns that there might be more Enrons out
there. The management's responses just became one important
piece in the jigsaw.” (Analyst 9)
4.1.2. Information management prefers not to put it in writing
We found that it is common for participants on both sides of the
house going into a firm's results presentation to be aware of certain
sensitive issues affecting the firm, which they want to discuss
further, but feel they must deliberately avoid during the formal
presentation. This is likely to be out of courtesy, respect, deference
and demeanour (Goffman, 1956, 1959, 1963, p. 248). Our observa-
tional and interview data indicate that these sensitive issues
include litigation and regulatory problems, information concerning
peers and competition, mergers and acquisitions, and changes to
senior management teams. This is another reason why the Q&A is
important to the creation, production and dissemination of
knowledge. The Q&A is a relatively safe space because it is governed
by its own distinct ritual code. Certain disclosure decisions appear
to be prohibited during the formal presentation phase because
management feel obliged to behave in a certain way. However, they
feel permitted to make those disclosure decisions during the Q&A
because this phase is considered to be governed by its own face-
saving rules; it is a separate interaction ritual (Goffman, 1955,
1959, 1963, 1967).
“Sometimes they would rather discuss than have it written
down in a presentation. Anyone can go and read the recordings,
but you know it's slightly less harmful for them if they're saying
 S. Abraham, M. Bamber / Accounting, Organizations and Society 58 (2017) 15e31
things about their competitors in the Q&A than if they are
massively documenting that in the presentation.” (Analyst 7)
Goffman (1959) believes that when an individual performs, he
or she “typically conceals something” (p. 51). This may be because
disclosure might compromise the individual's idealised identity
(Goffman, 1955, 1967). However, when the formal presentation is
finished and the performer is asked what he or she is concealing,
then the ritual code might be changed sufficiently for the performer
to reveal it (Goffman, 1955, 1959, 1967).
“You know these questions will be asked and we rely on that as a
way of getting information out in a public forum. You wouldn't
want to necessarily do a slide on it … but sometimes you do
want to get your point across.” (IR 21)
4.1.3. Issues that companies may be trying to dissimulate
In line with expectations from theories of surveillance (Deleuze
& Guattari, 1988; Foucault, 1979; Mathiesen, 1997) and impression
management (Goffman, 1955, 1959, 1963, 1967), our data indicate
circumstances in which management are reluctant to respond
publicly to certain issues because the item being concealed creates
a “discrepancy between appearance and actual activity” and
therefore compromises the idealisation of the performer and/or the
performance (Goffman, 1959, p. 59). Throughout this public inter-
rogation, analysts encourage management to disclose information
that they may have been trying to dissimulate. They exploit the
exposure of management to surveillance technologies and net-
works, which promote conformity with expected social norms
because they operate under an externally imposed docility (e.g.
Deleuze & Guattari, 1988; Deleuze, 1992, pp. 3e7; Foucault, 1979).
Thus, and consistent with Hollander et al. (2010), we were told that,
if the firm remains silent on key issues, this can be a useful signal.
“Sometimes there are questions and you will ask investor re-
lations and they will say ‘We can't tell you that’; but if you want
to make a point that this is an important issue and that you are
just sort of being ignored by the investor relations then you do it
during the Q&A because then people say ‘Well, why is the CEO
avoiding this question?’.” (Analyst 19)
The gaze of surveillance falls on the interrogator when they pose
a question. They find themselves shifted from backstage to front-
stage (Goffman, 1959), but equally, they enact a self-transformation
from being the target of surveillance to the agent (Boll, 2014; Brivot
& Gendron, 2011; Deleuze & Guattari, 1988; Foucault, 1979; Gilliom,
2006; Latour & Hermant, 1998; Latour, 2005; Marx, 2003;
Mathiesen, 1997). This means that the interrogator will almost
certainly engage with impression management techniques
(Goffman, 1955, pp. 159, 1967), show deference and demeanour
(Goffman, 1956, 1967) and at the same time fulfil their key objective
of being seen to be asking good questions. This becomes more
difficult when analysts are asking questions that management do
not like, and a skilled analyst must get his or her strategy right.
“Analysts want to be seen like their questions are cutting
through the marketing side of the company, and they're getting
to the heart of the problems and the issues.” (IR 32)
Given the rarity of this frontstage moment for the analyst, it is
unsurprising that self-interest e often in the form of self-
promotion e is far from the surface. This is a benefit of being the
agent of surveillance as opposed to the target (Brivot & Gendron,
23
2011).
“[Analysts] ask questions because they're advertising the fact
that they're there and they're interrogating the CEO.” (IR 32)
It was striking that throughout our observational work we
recorded a small number of instances of repetitious interrogation,
where analysts ask a series of similar questions located around a
common theme, seemingly in concert with each other, probing one
after another on a specific topic. Given the competitive nature of
the sell-side analyst role (e.g. Bradshaw, 2015; Brown et al., 2015),
any co-operation or team work may appear surprising.11 However,
we were told that this was a way to share the burden of asking
difficult questions that could disrupt the relationship with man-
agement or be viewed as overly aggressive by the virtual audience
(Goffman, 1955, 1959, 1967). Vollmer (2007) argues that “cross
party co-operation between audiences and performers appears to
be particularly common in calibrating numerical displays of regu-
latory drama for inspection by third parties” (p. 590). Although he
suggests that this might be because it provides an opportunity for
all parties to become comfortable with the numbers, it might also
be a product of backstage negotiation and resource mobilisation.
Over the next few paragraphs, we present two examples to illus-
trate the employment of this co-operative face-work and the
impact it has on the creation, production and dissemination of
information.
First, in 2013, Rolls-Royce were asked to restate their financial
statements in light of an investigation by the Financial Reporting
Council (FRC) into their revenue recognition policy. In line with
International Accounting Standard 8, the company disclosed the
restatement in a press release, but analysts appeared to continue to
be concerned about it. The following is an extract from their results
presentation Q&A:
Analyst 1: Is the FRC looking into any other accounting issues, or
does this draw a line under absolutely everything? There are no
other issues we should be concerned about? Just if you can
update us on what's actually going on there in discussions with
the FRC. Thank you …
[and later in the presentation]
Analyst 2: And the second (question) goes back to [the] question
right at the beginning about the FRC talks, because we can't
really ignore it, and for you, Mark, to say they are private dis-
cussions, are you absolutely rock-solid sure that there will be no
forced changes to Totalcare accounting?
This is not just unusual because of the individualistic nature of
the analyst's role, but this pack behaviour is striking because it
would seem to challenge the line taken by management, and in
turn pose a threat to their ‘face’. Therefore, we asked respondents
about this phenomenon. First, we were told that this co-ordinated
response strategy forces management to respond ‘on the record’
(Jeacle, 2008; Roberts et al., 2006), even if that response is silence
(Hollander et al., 2010) or obfuscation. Second, this co-ordinated
interrogation strategy highlights to the audience (and manage-
ment) that the issue being focused on is an important one. For
example:
11
Although our data do not point in this direction, it is possible that this industry
is less competitive than some of the extant literature suggests (cf. Barker, 1998).
This would be a fascinating area to follow up.
24 S. Abraham, M. Bamber / Accounting, Organizations and Society 58 (2017) 15e31
“I think the people who keep asking the same question, which
management aren't really answering, then they feel that they're
making a point to the management, even if they don't get an
answer.” (IR 12)
Analysts do not often engage in this type of co-operative face-
work (Goffman, 1955, 1967). From a sociological perspective this
behaviour is interesting. Goffman (1956, 1959, 1967) describes how
situations become difficult to sustain when individuals are required
to adopt a version of self frontstage that contradicts their version of
self backstage (cf. Collins, 2005). Analysts are typically secretive
about their information, are self-serving in its gathering and
dissemination, and ultimately are competitive among each other
(e.g. Brown et al., 2015). However, during these moments of co-
ordinated interrogation, they become a “performance team”
(Goffman, 1959, p. 85). Goffman offers an explanation: “Coopera-
tion between two [or more] performers each of whom was osten-
sibly involved in presenting his own special performance could be
analysed as a type of collusion or understanding without altering
the basic frame of reference” (p. 85 [emphasis in original]). The
implication is that the management team will see this as part of the
analysts' performance, and therefore the threat will not be attrib-
uted to any one in particular because this is not their own ‘special
performance’. Through this practice, analysts are able to spread the
risks associated with asking management to disclose information
that they may have been consciously trying to dissimulate. How-
ever, we witnessed that the analysts who take part in this team
performance almost always forfeit their opportunity to ask a
question of their own.
Given that we attended several presentations, we had the op-
portunity to contrast the backstage and frontstage performances of
analysts. We observed how they came together as a community
before and after the event. Unsurprisingly, they know each other
socially as well as professionally. Invitees assemble in the lobby of
the results presentation venue up to 30 min before commencement
and, amid small talk and refreshments, discuss issues relevant to
their research. In line with Goffman's views, this backstage meeting
provides an opportunity for Machiavellianism (Collins, 2005;
Goffman, 1959, 1967), plotting and manipulation of the agenda.
Indeed, Foucault (1979) and other surveillance theorists (e.g.
Mathiesen, 1997) point towards a similar pattern of behaviour,
namely that, when under surveillance, actors are likely to devise
strategies to overcome feelings of powerlessness (Flynn, 1985;
Foucault, 1994; Spellmeyer, 1989).
Occasionally, analysts threaten the line taken by management,
but this is less harmful when done as a pack. This is important given
that this form of interrogation strategy might be translated by
watchers as an aggressive form of face-work. When this occurs, a
“corrective” process often follows (Goffman, 1955, 1967, p. 22),
namely “challenge, offering, acceptance, and thanks”. In these cir-
cumstances, it is imperative that both parties emerge without
losing face and that the social boundaries are re-introduced
without too much damage.
Our second observation example points towards this backstage
Machiavellianism, demonstrates co-operative face-work and re-
veals elements of the corrective process:
Analyst 1: Good morning, it's [Analyst 1 name] from [Firm X]. A
couple of questions on John Crane and one on Interconnect,
please …
[Group CEO thanks the sell-side analyst, makes a brief comment
and then passes over to John Crane's CEO, who is seated in the front
row. He approaches the podium, takes the microphone and re-
sponds to the questions before retaking his seat in the front row] …
Group CEO: Yes, again, second row please?
Analyst 2: Morning. It's [Analyst 2 name] from [Firm Y] and I
actually had three on John Crane please and I wondered if you
could make a comparison, but perhaps against the late 1990s,
because I wonder whether that's a more interesting comparison
to what we saw with the great financial crisis … [questions
continue]
John Crane CEO: In the late 1990s … [responds to first two
questions] … And the third question was around?
Analyst 2: Aftermarket margin?
John Crane CEO: I don't think we've ever disclosed OE and
aftermarket margins [looks at Group CEO, who nods], and we
don't do that for competitive reasons as you can imagine.
Group CEO: Thank you, [John Crane CEO name].
John Crane CEO: You are welcome. [Retakes seat in the front row]
Group CEO: Next one along in the second row. It seems to be a
hotbed of questioning there.
Analyst 3: [Analyst 3 name], [Firm Z]. Another John Crane, I am
afraid. I don't know why you're bothering sitting down, [John
Crane CEO name] [laughs; others also laugh]. So, I just wanted to
ask a couple of questions, maybe just as more clarification than
anything else. [questions continue] …
[John Crane CEO stands, responds and then retakes his seat in the
front row] …
Group CEO: Thank you. At the moment there is no sign of any
further questions, so you [speaking directly to John Crane CEO]
may be able to sit down [all laugh].
Smiths Group's 2015 Q4 results presentation Q&A was domi-
nated by questions about one subsidiary, John Crane. This extract
highlights three issues. First, the collaborating interrogators sit in
the same row,12 indicating backstage collusion. The Group CEO
notes this ‘hotbed’ of interrogation. Second, the interrogators not
only push the social boundaries, but also test the regulatory
boundaries. Indeed, the Group CEO is required to arbitrate on the
non-disclosure of the ‘aftermarket margins’ requested by the sec-
ond analyst. Finally, the third analyst uses humour as a ‘corrective’
device to soften the potential interpretation that this is aggressive
face-work and a breach of the ritual code (note: this is omitted from
the published transcript, and therefore we have transcribed it
verbatim). The CEO completes the corrective process. As Vollmer
(2007) notes, there is an unwillingness to compromise the truce
unnecessarily and therefore “a certain dishonesty … might even be
deliberately disattended” (p. 592). Later in the discourse, the CEO
mimics this tone to indicate acceptance, and at the end of the
session he thanks all of the physical and virtual audience, including
staff and analysts.
This interrogation strategy allows ‘the many’ to exert control by
operating as a team, wresting power away from ‘the few’ (i.e.
management) (Haggerty & Ericson, 2000, 2006; Lyon, 2006;
Mathiesen, 1997), and thus subverting the normal rules of this
interaction ritual (Goffman, 1955, 1959, 1967). For example, one
manager told us about the apologetic manner that is adopted to
avoid this face-threatening behaviour. Given their independence
12
This might be due to chance, but our lived experiences suggest that this is
unlikely. Having gathered into small social groups in the atrium before the event,
the audience are then led to the conference room. We noticed that they typically sit
with this same social group with whom they walked in.
 S. Abraham, M. Bamber / Accounting, Organizations and Society 58 (2017) 15e31
and self-interest motives, it would be difficult for the analysts to
sustain this ‘team performance’ for long periods. Analogously,
Goffman (1959) states that spies and counterespionage operatives
often fail because they cannot sustain false cover for extended
periods. Instead, the key to this interrogation strategy e in which
one line of questioning is relentlessly pursued by multiple in-
terrogators e is that it relies on only brief moments of co-operation.
These interrogators confound normalcy because the abnormal be-
haviours are advantageous.
“I have been in sessions where [management] may have glossed
over the issue, even when everyone knows there is an issue.
They have glossed over it on the first answer. Then you will get a
second analyst come in and say, ‘This is very important. Sorry to
go over this again but I am going to dig a bit deeper on what is
going on here’.” (IR 31)
Interestingly, we were told that firms occasionally issue press
releases after a results presentation, particularly when they have
been subject to this co-operative pack-interrogation strategy.
Therefore, this Q&A behaviour can directly affect the creation,
production and dissemination of information.
4.2. Analysts under surveillance
Whilst the Q&A displays elements consistent with the uni-
directional network surveillance frameworks e panopticon and
synopticon e the encounter also shares commonalities with the
more complex post-panoptic frameworks. Similar to the oligopti-
con (Boll, 2014; Latour & Hermant, 1998; Latour, 2005), we were
regularly told that the key protagonists are being simultaneously
watched and judged by one another and by the virtual audience.
However, with the exception of volunteering to ask a question,
shifting between target and agent is rarely at the will of the pro-
tagonist. In addition, this is an interconnected web of surveillance
networks, lateral and hierarchical, which suggests that rhizomatic
surveillance (Deleuze & Guattari, 1988) is an important factor
determining what happens during the Q&A, and why. Whilst sur-
veillance is characterised as a mechanism of control, it is human
nature to challenge and problematize it (Deleuze, 1988; Foucault &
Blanchot, 1990; Foucault, 1984), and engage in counter-surveillance
behaviours (Gilliom, 2006; Marx, 2003), such as exhibitionism
(Brivot & Gendron, 2011; Koskela, 2002). In the following para-
graphs, we outline three key groups to whom the interrogator
becomes visible: current and potential clients; company manage-
ment; and the wider market.
We were told that the Q&A is an opportunity for analysts to put
their expertise on public display, showing that they are pro-
fessionals who can provide insights into the future performance of
the company. Thus, our findings support propositions from other
work in the field. First, we were told that this is a highly competitive
profession and one in which personal reputation and exposure are
critical success factors in the career prospects (e.g. Bradshaw, 2015;
Brown et al., 2015). Furthermore, in line with Brown et al. (2015),
we were told by analysts (in formal and informal conversations)
that even a small amount of contact with senior management is a
signal of superior knowledge to their clients. Similarly, Barker et al.
(2012) cite the importance of social interaction with management
as a means for investors to foster legitimacy and credibility in the
eyes of their clients. One finance director succinctly explained:
“It is not just you on show e they are on show as well for a rare
moment.” (IR 38)
25
To ask a question during the Q&A, an audience member must
raise a hand and wait for the microphone. A striking feature of the
interrogation performance ritual is that, before the individual
questioning begins, the analyst will declare his or her name and
organisation. This exhibitionism is a form of face-work and is often
combined with impression management techniques (Goffman,
1955, 1956, 1959, 1967).
“So you will have the guy saying what his name is and where
they are from. Like, ‘So and So from XX’. And then they'll ask the
question to be sure that everyone knows who they are, where
they are from. And you think, they would probably give their
email address if they got the opportunity.” (IR 28)
This opportunity for exhibitionism is important for a sell-side
analyst. This is a consensual performance, and the management
team acknowledge analysts’ enthusiasm for exhibitionism. Vollmer
(2007) argues that, when a recognised model of translation of
financial numbers is established, such as the Q&A, then the
numbers take on a new quality. This means that there is a greater
onus on the interrogator to issue a sensible challenge while also
balancing out their keenness to be noticed.
“They are all asking sensible questions, but I mean, basically, it is
about them marketing themselves.” (IR 11)
In contrast to Brown et al. (2015) claims that silence is preferred
among analysts during the Q&A, we found that there is a tacitly
accepted value in shifting from backstage to frontstage, which an-
alysts exploit. This corresponds with the “disappearance of disap-
pearance” noted by Haggerty and Ericson (2006) in their
commentary on the modern surveillance society. We were told:
“So if one of the guys goes and asks a question … and others
don't ask a question, an investor might be thinking: ‘Ah, who's
this chap? He sounds really clever and he's asking all the right
questions, maybe I should give him a call’.” (IR 33)
We were told that fund managers seldom attend the results
presentations physically, but usually watch and review the pre-
sentations (live, recorded and/or transcribed). Therefore, the
quality of the interrogation can have consequences for the in-
terrogator's perceived social, political and economic value.
“So I mean if you've got a big client, and if you've got [name] at
RBS [Royal Bank of Scotland] standing up asking you a question
… [the client] will be listening to the call thinking: ‘Oh yeah he's
all over this!’ So when I come to buy or sell the shares the next
time, I'll go through RBS.” (IR 32)
The Q&A also provides an opportunity for analysts to make their
opinions and advice appear credible to their potential clients. This
is not only economically motivated, it is also positive face-work
(Goffman, 1955, 1956, 1959, 1967), made possible by the presence
of surveillance. It also helps to explain why the Q&A might be a
forum for the creation, production and dissemination of useful
information. In other words, the interrogators' questions can reveal
valuable information as much as can the management team's
responses.
“You get a lot of credibility from asking the right questions with
investors listening [in].” (Analyst 1)
We were told that the quality of the interrogator's performance
26 S. Abraham, M. Bamber / Accounting, Organizations and Society 58 (2017) 15e31
can enhance (or harm) their profile and their career prospects.
Asking good questions is a means to distinguish oneself from the
pack.
“If there is something that is really important that no one else
has mentioned then I would be desperate to get my question in
so that I am the one who asked the question.” (Analyst 13 [italics
indicate vocal emphasis])
The shift from backstage to frontstage presents a profile-raising
opportunity.
“If you ask questions at the analysts' meetings, you find that
your profile does rise and that can have all sorts of knock-on
effects for an analyst's career. If you're a well-known analyst in
the City, other job opportunities open up at other brokers and
other fund managers. You'll find that you're continually being
bombarded by head-hunters and other institutions trying to
poach you away. Those analysts who have a higher profile tend
to get a lot of coverage in the media, in the written media and
broadcast media as well. If you're high profile in these meetings
then it has a knock-on effect there as well.” (Analyst 24)
We observed how the “ritual order” is organised along
“accommodative lines” (Goffman, 1956, 1967, p. 42). We witnessed
numerous instances where management were able to display their
depth of understanding of issues relating to strategy, position and
performance, but interestingly we also witnessed managers dis-
playing more subtle levels of knowledge and interactional sophis-
tication. For example, an impression management technique
employed by many interrogatees is to greet their interrogator by
name before entering the standard introductory ritual. Goffman
(1955, 1967) specifically identifies greetings (and farewells) as a
recognition of a pre-existing affiliation and notes how this ritual is
an “obligation of many social relationships”; it “prevent(s) disrup-
tion” of that bond or the situation and establishes a “shared face”
(pp. 41e42). This means that “in the presence of third parties an
improper act on the part of one member becomes a source of acute
embarrassment to the other members” (pp. 41e42). These small
acts of deference and demeanour not only reveal that an analy-
stemanager relationship exists prior to the Q&A, but they also avert
one member from acting “shamelessly” and “destroying the others’
face” (Goffman, 1955, 1956, 1967, p. 42).
“Partly [asking questions] is a form of advertising. You put up
your hand and say, ‘Why have you raised your dividend by 5%
not 10%?’ The company then say, ‘Ah, [your name], great to see
you, how are you doing? Thanks for the question; it was a really
good one.’ Your clients are going to hear that and think ‘Oh, he's
all over it!’” (Analyst 9)
Relatedly, we were told that analysts have a great deal of
reputational capital invested in their prior recommendations, and
so, if the general tone of the results presentation is contrary to that
view, the questions are an opportunity to ‘save face’ or ‘be in face’
(Goffman, 1955, 1967). Although dissimilar to the pack behaviour
described above, this is nonetheless an example of co-operative
face-work (Goffman, 1955, 1967, pp. 27e31). Management told us
that they prepare extensively for these events and that one part of
that is anticipating questions. This involves carefully reviewing
analysts' recent research to understand what angle each might
take. Management attach importance to providing measured re-
sponses to interrogators' questions that do not threaten their face.
“You have got to read what they have written in the past … and
respond accordingly … Because what they don't like is being
proven wrong.” (IR20)
An assumption in the conference call literature is that man-
agement are required to respond to questions spontaneously (e.g.
Larcker & Zakolyukina, 2012; Mayew et al., 2012) and, corre-
spondingly, that analysts unrestrainedly interrogate management.
However, if this were an accurate representation, then we might
expect the Q&A to be an explosive moment of public social and
political tension, whereas this is seldom the case. Instead, we
observed a consensual awareness of social and political boundaries
that correspond to those described by theories of interaction ritual
(Goffman, 1955, 1956, 1967) and specifically ‘performance ritual’
(Goffman, 1959). The Q&A is almost always nuanced, sometimes
critical, but in our experience is rarely intentionally confrontational
or combative.
Thus, face-work is important (Goffman, 1955, 1956, 1967), and
we were frequently told that the Q&A provides an opportunity for
analysts to accrue reputational capital and credibility with the se-
nior management team. As Goffman (1955, 1967, p. 10) notes,
accepting everyone else's line is the “kind of mutual acceptance”
that “seems to be a basic structural feature of interaction, especially
the interaction of face-to-face talk”. The key benefit of a non-
threatening exchange is that it allows participants to build re-
lationships under the public gaze and therefore enhance the par-
ticipants' perceived social value.
The literature shows that analysts' relationships with manage-
ment are important (Barker et al., 2012; Chen & Matsumoto, 2006;
Francis & Philbrick, 1993; Libby, Hunton, Tan, & Seybert, 2008), and
our interviewees spoke of this as well. Goffman (1955, 1956, 1967, p.
55) suggests that there are certain “rules of conduct” that “tend to
be organised into codes which guarantee that everyone acts
appropriately and receives his due”. We observed how analysts’
questions are always professional in articulation, tone and
construct. “Appearance” and “manner” are important during this
social encounter because “the performers can stop giving expres-
sions but cannot stop giving them off” (Goffman, 1959, p. 111).
Relationships are built and developed over time and do not
emerge as the result of a single interaction or one good (poor)
question. Thus, we were told that the analyst's question(s) must be
perceived as ‘serious’ and this corresponds to the view manage-
ment take of them. Interrogation strategy and behaviour has im-
plications for the analystemanager relationship and, by extension,
for their credibility and reputation in the eyes of the wider in-
vestment community (Barker et al., 2012; Brown et al., 2015).
“There are times when you need them [i.e. management] to feel
that you are serious about coming to the stock, that you are
someone that they ought to be talking to and that [i.e. asking
questions] helps … if you are one of the analysts that they really
respect they will give you that call because they want to make
sure that you are on top of things. If they think you don't really
matter, you are going to be down their list.” (Analyst 24)
We observed and were told about rare exceptions where ana-
lysts breach the established social norms by asking questions that
are deliberately provocative, aggressive, confrontational and/or
naive. These disrupt the performance ritual by challenging the
management line and often require immediate counteractive face-
work (Goffman, 1955, 1956, 1959, 1967). We return to this later to
highlight the deleterious consequences this can have for the indi-
vidual analyst and how poorly conceived interrogation strategies
and behaviours can impair important relationships and damage
 S. Abraham, M. Bamber / Accounting, Organizations and Society 58 (2017) 15e31
reputational capital and credibility.
4.3. Visibility and the disincentives to questioning during the Q&A
Although we found a strong willingness among analysts to ask
questions, we also found support for Brown et al. (2015) claim that
questions are sometimes deliberately suppressed. We extend their
work by exploring the causes of this behaviour. To this end,
although we observed and were told about exhibitionism as an
impression management strategy, we also heard about the benefits
of ‘avoidance’ (Goffman, 1955, 1959, 1967) and secrecy (Brivot &
Gendron, 2011). Surveillance theorists believe that secrecy is a
common and effective counter-surveillance strategy (e.g. Gilliom,
2006; Marx, 2003), and Goffman (1955, 1967, p. 15) believes that
“avoidance” is “the surest way for a person to prevent threats to his
face”. It is crucial for actors to “have, or be in, or maintain face”
during this social encounter (Goffman, 1955, 1967, p. 5), as opposed
to being caught “in wrong face” or “be out of face” (p. 7 [emphasis in
original]).
We were told by both management and analysts that questions
are withheld because of the presence of surveillance technologies
and networks. Specifically, analysts are concerned about: (i) giving
away their private information; (ii) suffering social humiliation;
and (iii) damaging their relationship with management. We discuss
each of these below.
4.3.1. Giving away insights to competitors
We were told, in line with Brown et al. (2015), that analysts’
questions are likely to reveal private insights to their direct com-
petitors; and this tension is ratcheted because of the high levels of
visibility. The controlling effects of this lateral surveillance network
add weight to Mayew et al.'s (2012) proposition that analysts are
unwilling to pass information on to their peers. Under the gaze of
surveillance, analysts sometimes choose to remain silent (Gilliom,
2006; Marx, 2003). This basic form of face-work is a “protective
maneuver” (pp. 15e16). Goffman (1955, 1967) argues that it is
common for individuals to employ “discretion” (p. 16), which may
include leaving certain facts unstated. Furthermore, it is likely that
the performer will “employ circumlocutions and deceptions” (p. 17)
to avoid losing face, and/or in this case, their information advantage
as well. We were told that questions that reveal innovative or novel
insights allow others to capitalise on those insights by either
reproduction or refutation.
“I would say a strange behavioural thing about the Q&A, if you
do share your thoughts and your questions, and they are
genuine questions, and they're genuinely your competitors in
the room, then why are you sharing your thoughts when your
key asset is your intellectual property?” (Analyst 22)
However, another analyst saw this moment of visibility as a
social, political and economic ‘trade-off’:
“You have got a trade-off between how clever you think the
question will make you sound and also giving away your entire
argument to all your competitors.” (Analyst 14)
4.3.2. Fear of social humiliation
We had the opportunity to witness certain interrogators' anxi-
ety first-hand and talk to them about it. We were often told about
the risk of ‘losing face’ that could result from poor interrogation
strategies or behaviours. Goffman (1955, 1967) predicts that, under
these circumstances, individuals might employ defensive mea-
sures, for example keeping “off topics and away from activities that
27
would lead to the expression of information that is inconsistent
with the line he is maintaining” (p. 16).
“There are people who don't perhaps want to stick their hand up
and ask a question that they might find embarrassing in front of
their peer group because they didn't understand something.” (IR
9)
Until now, we have discussed the consensual nature of the
performance and how face-work can be employed effectively. In
doing this, we have attempted to demonstrate that there exists a
“state where everyone temporarily accepts everyone else's line is
established” (1955, 1956, 1959, 1967, p. 11). This mutual acceptance
has a profound effect on the encounter whereby actors establish a
temporary working acceptance of each other's line (Goffman, 1955,
1967), which allows participants to respond and act accordingly.
However, there is always the possibility that this co-operation will
fail or falter, and the established line becomes threatened. There is
an information asymmetry gap between management and audi-
ence, and therefore secrets and rumours abound. Goffman (1959, p.
141 [emphasis in original]) highlights the complexity of individuals'
secrecy during most social encounters and specifically how some
information is withheld: “There are usually facts which, if attention
is drawn to them during the performance, would discredit, disrupt,
or make useless the impression that the performance fosters. These
facts may be said to provide destructive information.” Occasionally,
these come to light during the Q&A.
Goffman (1955, 1967) continues: “It is no wonder that trouble is
caused by a person who cannot be relied upon to play the face-
saving game” (p. 31). This is a simple social code that has only
two rules: self-respect and considerateness (Goffman, 1955, 1967, p.
11). Each participant is “expected to go to certain lengths to save the
feelings and the face of others present … he is disinclined to wit-
ness the defacement of others”, and if the participants do, they are
considered “heartless”. If they participate actively, they are
considered “shameless” (pp. 10e11). However, reminiscent of
stories of the bogeyman, several respondents told nightmarish tales
of others being publicly shamed by management responses to
questions. These stories may be allegorical, but they were also often
specific. We heard, for example, of the ferocious approach of Fred
Goodwin (CEO of RBS between 2001 and 2009) to responding to
questioning. We learned that any question deemed irrelevant,
inane or contrary to the management message may be ridiculed
and, consequently, the questioner may be socially humiliated.
“Sometimes companies go out of their way to disprove you.”
(Analyst 16)
“It's embarrassing for them when they ask something stupid …
and sometimes [the CEO] will enjoy letting them know it's
stupid.” (IR 29)
One analyst referred to this assertive response strategy as both
‘defensive’ and ‘aggressive’ face-work. Nonetheless, the analyst
believed this to be potentially useful to their decision-making
process:
“I definitely think that the Q&A session is extremely useful in its
own right. And often you can get quite a lot out of it … but you'll
find that some companies and some management are quite
defensive in the Q&A session e I'd even go so far as to say that
some are quite aggressive and will often turn it back on the
analyst and try and make the analyst look slightly foolish if they
ask a question that they don't particularly like.” (Analyst 17)
28 S. Abraham, M. Bamber / Accounting, Organizations and Society 58 (2017) 15e31
4.3.3. Relationship with company management
The relationship between analysts and managers is important
for social, political and economic reasons. Furthermore, this is a
two-way exchange. A good relationship has benefits for both
parties, whereas a failed or failing relationship can have deleterious
consequences. We were told that analysts aim to build strong re-
lationships with company management to garner occupational
credibility and capital (e.g. Bradshaw, 2015; Chen & Matsumoto,
2006; Francis & Philbrick, 1993; Libby et al., 2008), but an analyst
who threatens the face of senior company officials in this public
forum can themselves end up losing face. This is because, when the
interrogatee's face is threatened, so is his or her “pride, and his [or
her] honour” (Goffman, 1955, 1967, p. 32). Therefore, according to
the ‘ritual code’, face-work must be done. It is unusual that either
party would want a complete breakdown in relations, and there-
fore a corrective face-saving process could be operationalised. The
breach of the tacitly agreed ritual code might be mutually ratified as
just an unfortunate “incident” (Goffman, 1955, 1967, p. 12).
Participants agreed that they would interpret an aggressive
interrogation strategy and associated behaviours as a departure
from the ritual code. In these cases, management might feel
compelled to “resort to” what Goffman (1955, 1967) calls “tactless,
violent retaliation, destroying either themselves or the person who
had refused to heed their warning” (p. 23). This might be an im-
mediate harsh response, or it could be a delayed effect.
“There are some companies that will not let you ask a question if
they think you're going to be nasty. I was edited out of the
transcript of one company that particularly did not like my
questions. If you have a really negative view or are very
aggressive in terms of questions, they are probably not going to
give their banking business to you … The other thing is that a
very large proportion of pay from a lot of buy-side companies is
corporate access, so you get paid for the quality of your research,
you get paid for your ability to set up meetings. So you know,
guess what, if you ask a lot of really, really nasty questions you
don't get a lot of corporate access.” (Analyst 21)
There are potentially lasting damaging effects for the analyst
who misjudges his or her interrogation strategy or has a lack of
awareness of the ritual code:
“Companies do complain to heads of research about analysts
sometimes. There is one in my sector who got into a lot of
trouble actually. I won't go into details, but they [the firm]
complained and I don't know, cause and effect, but he [the an-
alyst in question] wasn't working there much longer.” (Analyst
9)
5. Discussion
The interpretation of the Q&A as a dramaturgical encounter
brought about by the presence of surveillance technologies and
networks provides a richer and deeper understanding of the social
and political motivations behind analysts’ interrogation strategies
and behaviours. As Vollmer (2007, p. 578) notes, there is more “at
stake when numbers are performed” than when they are simply
exercises in calculation. Therefore, our analysis section demon-
strates how and why the Q&A can be useful as a mechanism to
create, produce and disseminate information; but at the same time
we also identify factors that constrain this. Analysis of our data
shows how the presence of surveillance has introduced risks and
rewards, costs and benefits, stemming from how participants
problematize the sense of powerlessness associated with being the
target of surveillance.
We have focused on analysts and demonstrated how they can
benefit (or lose) during this exchange, but our findings have far
broader applicability. A combination of visibility and ritual bring
social meaning to the Q&A. The extant conference call literature
conceptualises these meetings as economic encounters, whose
principal objective is information retrieval. As a result of our
methodological approach and adoption of a sociological lens, we
are able to identify other concerns. In so doing, we arrive at a
similar conclusion to Pentland (1993, p. 619) in his analysis of audit
engagements, namely that researchers could undertake any
amount of rationalistic analysis, but that this alone will never fully
explain social and political motives, in this case interrogation
strategies and behaviours.
The remainder of this section examines the Q&A as a public
interrogation. We provide a descriptive framework (Fig. 2) based on
our findings that is useful both for academics and for practitioners.
In the traditions of interaction ritual (e.g. Collins, 2005; Goffman,
1955, 1959, 1967), it explains not simply what happens, but why
it happens that way. We identify and discuss the discrete social,
political and regulatory boundaries that govern the Q&A and pro-
vide alongside this a reasoned set of expectations about the ques-
tions that can and will be asked.
Zone A demonstrates the sub-set of information that is previ-
ously known to all those who participate in the public interroga-
tion. Before commencement of the Q&A, some information will
have been established, for example the firm's financial results and
some strategy decisions. We were told that, by the time the inter-
rogation starts, analysts will have received the earnings statement
and heard the formal presentations, and some will have already
released a brief note and verbally advised their sales team(s).
Therefore, a ‘good’ interrogator will avoid questions that fall within
Zone A, because the chance of hearing anything new and useful is
low. These questions are unnecessary and would do little to raise
the positive social value a person effectively claims for himself or
herself (Goffman, 1955, 1967). Additionally, zone A-related ques-
tions could have negative consequences for the interrogator
because the physical and virtual audience may judge them to have a
lower social value. In some cases where the question is deemed
“stupid” (IR 29), a hostile management team might become
Fig. 2. Discussion boundaries and information interrogation zones. Notes: Zone A
represents the information that is publicly available and known prior to the interro-
gation. Therefore, boundary wall B1 demarcates that which is known from that which
either requires clarification or is unknown. Boundary wall B2 represents the re-
strictions imposed by the governing regulatory code. Hence, there is a divide between
the information that can be lawfully disclosed (Zone B) and that which cannot (Zone
C). Zone B represents the band of uncertain knowledge, and Zone C represents that
which is unknown and prohibited from being revealed.
 S. Abraham, M. Bamber / Accounting, Organizations and Society 58 (2017) 15e31
impatient, dismissive and/or aggressive in their response.
Gabbioneta et al. (2013) suggest likewise, albeit in a different
setting. However, this reaction risks unnecessarily jeopardising the
tacit truce that governs the social situation (Vollmer, 2007). This is
important because being ‘in face’ (Goffman, 1955, 1967) is critical to
an analyst's success in this fiercely competitive industry (Table 2).
B2 demarcates the legal and regulatory frontier. We were told
that the management team will be well prepared for the Q&A and
equally aware of the implications of providing information that
breach strict legal and financial governance (Financial Conduct
Authority, 2014; Financial Services Authority, 2000, 1996). There-
fore, asking questions that threaten this boundary would almost
certainly be futile and worthless. However, there is more at stake
for the interrogator who chooses to pursue this strategy. Surveil-
lance theory states that being watched has a socially conforming
effect (Deleuze & Guattari, 1988; Foucault, 1979; Latour, 2005;
Latour & Hermant, 1998; Mathiesen, 1997). An analyst who choo-
ses to challenge this line would be considered to be simultaneously
a threat to the face of management (Goffman, 1955, 1956, 1959,
1963, 1967; Collins, 2005) and in breach of the established code,
and thus at the mercy of a disciplinary system (e.g. Deleuze &
Guattari, 1988; Foucault, 1979; Haggerty & Ericson, 2000, 2006;
Lyon, 2001; Mathiesen, 1997). We were told that the (unwritten)
disciplinary implications of these questions have social, political
and career dimensions. Therefore, the majority of questions fall
within Zone B.
These Zone B questions are dominated by points of clarification
and confirmation, and some will be related to issues that man-
agement prefers not to put in writing and may be attempting to
dissimulate. In this zone, there can be varying levels of questions,
some close to B1 and some close to B2, motivated by information
retrieval and visibility. However, the response could also reveal
useful information and provide new insights. In addition, analysts
are not only absorbing the manifest content, they are watching out
for a lack of clarity or an inability to confirm information, because
this can reveal incoherence in the management message. Analysts
claimed that this can be beneficial as it allows them an avenue to
interpret subjectively the responses, and from this deduce infor-
mation beyond B2. Silence can be beneficial to analysts because it
offers them an opportunity to display their expertise to clients
through their interpretation and fill in an information gap, while at
the same time respecting management's face.13 Occasionally, an
important issue is publicly known among participants but not
discussed during the formal presentations, and this will drive an-
alysts to follow up with questions that might be deemed to be
threatening the face of management (Goffman, 1955, 1967). These
questions are a rare exception because they are likely to fall outside
Zone B, namely either everyone already knows and they simply
want to see the management respond, or there is nothing more
than a rumour that appears to have some validity. At this stage, the
analyst must weigh up the trade-off between how critical this is to
his or her investment decision, the cost of losing face and the
damage that might arise from threatening the management line. In
these circumstances, interrogators sometimes come together as a
‘performance team’ and engage in ‘co-operative face-work’
(Goffman, 1959; Vollmer, 2007).
6. Conclusion
Libby et al. (2008) draw attention to the need for further
research on conference call (hence also results presentation)
participation: “Given that the answers to conference call questions
13
We are grateful to an anonymous reviewer for highlighting this point.
29
are now immediately made public, the exact nature of the benefits
from conference call participation is not clear” (p. 194). Accordingly,
the present article improves our understanding of the (dis)in-
centives that motivate analyst participation in the Q&A, how and
why the Q&A operates in the way it does, and whether this explains
its perceived value.
Interviews, although not without limitations, allow us the op-
portunity to ask direct questions rather than inferring intent from
statistical associations. This advances our understanding of the
complex and dynamic nature of the Q&A (Bradshaw, 2015), and in
particular the fluidity of the transition from backstage to frontstage
and the visibility brought about by the surveillance technology. To
further mitigate these limitations, we conducted extensive obser-
vations. Drawing on theories of surveillance (Deleuze & Guattari,
1988; Foucault, 1979; Latour & Hermant, 1998; Mathiesen, 1997; )
and interaction ritual (Goffman, 1955, 1956, 1959, 1963, 1967), we
have formulated a descriptive framework that demonstrates how
social actors modify their behaviour during public interrogations.
The Q&A is a dramaturgical encounter that is necessarily interactive
during which accounting information is both communicated and
translated. However, the usefulness of this observational work and
our descriptive framework is not limited to accounting research. It
can also be applied to similar public interrogations in other fields.
By viewing the process through a sociological lens, we throw
light on an understudied and poorly understood situation. Prior
studies tell us that notions of self and self-promotion are important
to meetings (Carrington & Johed, 2007; Catasus & Johed, 2007,
2012; Johed & Catasus, 2015; Uche & Atkins, 2015), and we
extend this literature in several ways. We explore a relatively highly
visible encounter, made so by the presence of surveillance tech-
nologies, where stakeholder demand is accentuated because this is
a key moment in a firm's financial calendar. We explain the types of
questions that analysts will ask (or not ask), and we provide in-
sights into the way that information is created, produced and
disseminated in a semi-public setting. Our evidence supports
conclusions from prior literature (Matsumoto et al., 2011; Mayew
et al., 2012) by showing that analysts play an important role
alongside management in co-constructing corporate discourse and
public disclosure.
Analysis of our data shows that analysts strive to pose ‘good’
questions, but it also highlights the potential consequences of
asking ‘poor’ ones. We highlight self-promotion as a major moti-
vating factor in the choice of questions posed. Our paper demon-
strates how self-promotion can both assist and supress information
available to the wider public. We therefore reaffirm Brivot and
Gendron’s (2011) assertion that “control mechanisms in real-life
settings do not operate deterministically but with complex,
lateral and unanticipated effects” (p. 152).).14 One manifestation of
this complexity can be seen in the way that analysts compete
against each other by trying to showcase their expertise but at the
same time are reticent about giving away their private information.
We strongly encourage further empirical investigation of the
Q&A, particularly observational fieldwork. This is a fascinating area,
and our work has barely scratched the surface regarding the
revelations that could emerge. Given the public, face-to-face nature
of the meeting, and its importance as a communication event, it
seems ripe for interdisciplinary work. However, understanding of
the process of emergence is limited,15 and therefore addressing this
should be a priority before any further assumptions arise that are
14
We are grateful to a reviewer for highlighting this point.
15
A notable exception is Li et al. (2014) review of conference calls, although there
are significant differences between US and UK events, some of which we have
outlined above.
30 S. Abraham, M. Bamber / Accounting, Organizations and Society 58 (2017) 15e31
not wholly appropriate or applicable. The Q&A has become an in-
tegral interactive opportunity. This may be because it is a forum in
which accounting information is communicated and translated.
More needs to be known about how the market for information has
changed and what (if anything) this implies for accounting
research.
The face-to-face aspect of UK-based results presentations is an
interesting feature, as is their quasi-public nature. It would be
useful to know whether invited participants learn more than those
who watch online, and whether live face-to-face meetings offer an
informational advantage over conference calls. Therefore, inter-
jurisdictional co-operation between researchers is a possible way
forward. Moreover, little is known about the relationship between
sell-side analysts and others in the investment community, such as
fund managers. This moment of visibility for sell-side analysts was
described as important, but several other communication and
interaction encounters between these groups need to be consid-
ered in more detail. Although we explore the ‘public’ discussions
between sell-side analysts and management on results day, we are
aware that there is an off-camera opportunity for these groups to
interact,16 including informal ‘post-Q&A, Q&As’. Obtaining access
to (details of) these conversations is difficult, but would provide
rich and interesting insights into verbal and written research
strategies and outcomes.
From a theoretical perspective, the prevalence of surveillance,
alongside the complexity, ambiguity and reach of the modern
surveillant assemblage, is crucial both to and beyond the Q&A. This
article presents a descriptive framework for understanding public
interrogations. Other areas where it might apply are diverse and
eclectic. Just a few examples include: (i) legal contexts such as
criminal cases where suspects are interrogated under oath; (ii)
political contexts such as televised leadership debates, Prime
Minister's Questions and international round-table events; and (iii)
academic contexts such as conference presentations.
Acknowledgements
We are very grateful to Professors Richard Barker and Ian
Thomson for their untiring advice, support and comments on our
work. We would also like to acknowledge the many helpful com-
ments from Professors Mark Clatworthy, Stuart Cooper, Richard
Taffler, Michael Jones, John Roberts, Joel Amernic and workshop
participants at various conferences including, Canadian Academic
Accounting Association Conference (Toronto, 2015), Interdisci-
plinary Perspectives on Accounting Conference (Stockholm, 2015),
European Accounting Association Conference (Glasgow, 2015),
Financial Reporting and Business Communication Conference
(Bristol, 2013, 2014), South-West Accounting Group (Gloucester,
2013; Cardiff, 2014). Finally, we offer our thanks to those in-
dividuals who gave up their valuable time to be interviewed as part
of this project.
References
Ahrens, T., & Chapman, C. S. (2006). Doing qualitative field research in management
16
When the camera stops rolling, management typically stay behind in the pre-
sentation room. They can remain there for up to an hour afterwards. Analysts,
management and other firm representatives form part of these discussions. This is a
private meeting, and therefore, as with other meetings of this type, the dissemi-
nation of price-sensitive information is constrained by fair disclosure regulation.
Following this, it is usual for the management team to break off into sub-groups to
answer questions from the media, analysts and investors for the remainder of the
day. Then the investor relations team and some of the senior management will take
to the road for a series of private meetings that can last for several weeks.
accounting: Positioning data to contribute to theory. Accounting, Organizations
and Society, 31(8), 819e841.
Allee, K. D., & Deangelis, M. D. (2015). The structure of voluntary disclosure nar-
ratives: Evidence from tone dispersion. Journal of Accounting Research, 53(2),
241e274.
Armitage, S., & Marston, C. (2008). Corporate disclosure, cost of capital and repu-
tation: Evidence from finance directors. The British Accounting Review, 40(4),
314e336.
Barker, R. G. (1998). The market for informationdevidence from finance directors,
analysts and fund managers. Accounting and Business Research, 29(1), 3e20.
Barker, R., Hendry, J., Roberts, J., & Sanderson, P. (2012). Can company-fund manager
meetings convey informational benefits? Exploring the rationalisation of equity
investment decision making by UK fund managers. Accounting, Organizations
and Society, 37(4), 207e222.
Bell, C. (1992). Ritual theory, ritual practice. Oxford University Press.
Bentham, J. (1995). Originally published in 1843. In M. Bozovic (Ed.), The panopticon
writings. London: Verso.
Bogard, D. (2006). Welcome to the society of control: The simulation of surveillance
revisited. In The new politics of surveillance and visibility (pp. 55e78). Toronto:
University of Toronto Press.
Boll, K. (2014). Shady car dealings and taxing work practices: An ethnography of a
tax audit process. Accounting, Organizations and Society, 39(1), 1e19.
Boyne, R. (2000). Post-panopticism. Economy and Society, 29(2), 285e307.
Bradshaw, M. (2015). Analysts' forecasts: What do we know after decades of work?.
Working paper. Boston College.
Brivot, M., & Gendron, Y. (2011). Beyond panopticism: On the ramifications of
surveillance in a contemporary professional setting. Accounting, Organizations
and Society, 36(3), 135e155.
Brown, L. D., Call, A. C., Clement, M. B., & Sharp, N. Y. (2015). Inside the “black box”
of sell-side financial analysts. Journal of Accounting Research, 53(1), 1e47.
Caluya, G. (2010). The post-panoptic society? Reassessing Foucault in surveillance
studies. Social Identities, 16(5), 621e633.
Carrington, T., & Johed, G. (2007). The construction of top management as a good
steward: A study of Swedish annual general meetings. Accounting, Auditing &
Accountability Journal, 20(5), 702e728.
Cascino, S., Clatworthy, M., Osma, B. G., Gassen, J., Imam, S., & Jeanjean, T. (2013). The
use of information by capital providers. Academic literature review. The Institute
of Chartered Accountants of Scotland and the European Financial Reporting
Advisory Group.
Catasus, B., & Johed, G. (2012). Behind open doors: Auditor’s face-work at the
annual general meeting. Working paper presented at interdisciplinary confer-
ence held at Cardiff in UK.
Catasus, B., & Johed, G. (2007). Annual general meetingsdrituals of closure or ideal
speech situations? A dual analysis. Scandinavian Journal of Management, 23(2),
168e190.
Chen, S., & Matsumoto, D. A. (2006). Favorable versus unfavorable recommenda-
tions: The impact on analyst access to management-provided information.
Journal of Accounting Research, 44(4), 657e689.
Clatworthy, M. (2005). Transnational equity analysis. John Wiley & Sons.
Collins, R. (2005). Interaction ritual chains. Princeton University Press.
Davis, A. K., Ge, W., Matsumoto, D., & Zhang, J. L. (2014). The effect of manager-
specific optimism on the tone of earnings conference calls. Review of Account-
ing Studies, 1e35. Ahead of Print.
Deleuze, G. (1988). Foucault. Ed. And trans. Sean hand. University of Minnesota Press.
Deleuze, G. (1992). Postscript on the societies of control. October, 59.
Deleuze, G., & Guattari, F. (1988). A thousand plateaus: Capitalism and schizophrenia.
Bloomsbury Publishing.
Doyle, A. (2011). Revisiting the synopticon: Reconsidering Mathiesen's ‘the viewer
society’ in the age of web 2.0. Theoretical Criminology, 15(3), 283e299.
Easterby-Smith, M., Thorpe, R., Jackson, P., & Lowe, A. (2008). Management research:
Theory and practice. London, UK: Sage Publications Ltd.
Eckersley, P., Ferry, L., & Zakaria, Z. (2014). A ‘panoptical’ or ‘synoptical’ approach to
monitoring performance? Local public services in England and the widening
accountability gap. Critical Perspectives on Accounting, 25(6), 529e538.
Eisenhardt, K. M., & Graebner, M. E. (2007). Theory building from cases: Opportu-
nities and challenges. Academy of Management Journal, 50(1), 25.
Elmer, G. (2012). Panopticon e discipline e control. In D. Lyon, K. Ball, &
K. D. Haggerty (Eds.), Routledge handbook of surveillance studies. London:
Routledge.
Fama, E. (1970). Efficient capital markets: A review of theory and empirical work.
Journal of Finance, 25(2), 383e417.
Financial Conduct Authority. (2014). UKLA Technical Note 521.2-Assessing and
handling inside information. Retrieved from http://www.fca.org.uk/your-fca/
documents/technical-note-521-2.
Financial Services Authority. (1996). Guidance on the dissemination of price-sensitive
information (London).
Financial Services Authority. (2000). Financial services and market act. Retrieved
from http://www.legislation.gov.uk/ukpga/2000/8/contents.
Flynn, T. R. (1985). Truth and subjectivation in the later Foucault. The Journal of
Philosophy, 82(10), 531e540.
Fogarty, T. J., & Rogers, R. K. (2005). Financial analysts' reports: An extended
institutional theory evaluation. Accounting, Organizations and Society, 30(4),
331e356.
Foucault, M. (1979). Discipline and punish: The birth of the prison. Knopf Doubleday
Publishing Group.
 S. Abraham, M. Bamber / Accounting, Organizations and Society 58 (2017) 15e31
Foucault, M. (1984). The use of pleasure: The history of sexuality, volume 2 (Original
L'Usage des Plaisirs, English translation 1985. London: Penguin Books.
Foucault, M. (1994). The ethics of the concern of the self as a practice of freedom. In
Michel Foucault: Ethics, essential works of Foucault 1954e1984 (pp. 281e301).
London: Penguin Books.
Foucault, M., & Blanchot, M. (1990). Maurice Blanchot: The thought from outside (Vol.
13). New York: Zone Books.
Francis, J., & Philbrick, D. (1993). Analysts' decisions as products of a multi-task
environment. Journal of Accounting Research, 31(2), 216e230.
Frankel, R., Johnson, M., & Skinner, D. J. (1999). An empirical examination of con-
ference calls as voluntary disclosure medium. Journal of Accounting Research,
37(1), 133e150.
Gabbioneta, C., Greenwood, R., Mazzola, P., & Minoja, M. (2013). The influence of the
institutional context on corporate illegality. Accounting, Organizations and So-
ciety, 38(6), 484e504.
Gilliom, J. (2006). Struggling with surveillance: Resistance, consciousness, and
identity. In K. Haggerty, & R. Ericson (Eds.), The new politics of surveillance and
visibility (pp. 111e129). Toronto: University of Toronto Press.
Goffman, E. (1955). On face-work: An analysis of ritual elements in social interac-
tion. Psychiatry, 18(3), 213e231.
Goffman, E. (1956). Embarrassment and social organization. American Journal of
Sociology, 264e271.
Goffman, E. (1959). The presentation of everyday life. New York: Anchor Books.
Goffman, E. (1963). Behavior in public places: Notes on the social organization of
gatherings (p. 248). New York: Free Press.
Goffman, E. (1967). On face-work. Interaction Ritual, 5e45.
Graham, J. R., Harvey, C. R., & Rajgopal, S. (2005). The economic implications of
corporate financial reporting. Journal of Accounting and Economics, 40(1e3),
3e73.
Haggerty, K. D. (2006). Tear down the walls: On demolishing the panopticon. In
D. Lyon (Ed.), Theorizing surveillance: The panopticon and beyond (pp. 23e45).
Devon: Willan Publishing.
Haggerty, K. D., & Ericson, R. V. (2000). The surveillant assemblage. The British
Journal of Sociology, 51(4), 605e622.
Haggerty, K. D., & Ericson, R. V. (2006). The new politics of surveillance and visibility.
Toronto: University of Toronto Press.
Hastrup, K. (1997). The dynamics of anthropological theory. Cultural Dynamics, 9(3),
351e371.
Holland, J. (1998a). Private voluntary disclosure, financial intermediation and
market efficiency. Journal of Business Finance & Accounting, 25(1&2), 29e68.
Holland, J. B. (1998b). Private disclosure and financial reporting. Accounting and
Business Research, 28(4), 255e269.
Hollander, S., Pronk, M., & Roelofsen, E. (2010). Does silence speak? An empirical
analysis of disclosure choices during conference calls. Journal of Accounting
Research, 48(3), 531e563.
Jeacle, I. (2008). Beyond the boring grey: The construction of the colourful ac-
countant. Critical Perspectives on Accounting, 19(8), 1296e1320.
Johansson, J. (2007). Sell-side analysts' creation of valueekey roles and relational
capital. Journal of Human Resource Costing & Accounting, 11(1), 30e52.
Johed, G., & Catasus, B. (2015). Institutional contradictions at and around the annual
general meeting: How institutional logics influence shareholder activism. Ac-
counting, Auditing & Accountability Journal, 28(1), 102e127.
Jones, M. J., & Solomon, J. F. (2010). Social and environmental report assurance:
Some interview evidence. Accounting Forum, 34(1), 20e31.
Koskela, H. (2002). Webcams, TV shows and mobile phones: Empowering exhibi-
tionism. Surveillance & Society, 2(2/3), 199e215.
Kothari, S. P. (2001). Capital market research in accounting. Journal of Accounting
and Economics, 31(1), 105e232.
Krippendorff, K. (2004). Content analysis: An introduction to its methodology.
Thousand Oaks, CA: SAGE Publications Inc.
Larcker, D. F., & Zakolyukina, A. A. (2012). Detecting deceptive discussions in con-
ference calls. Journal of Accounting Research, 50(2), 495e540.
Latour, B. (2005). Reassembling the social. An introduction to actor-network theory.
Oxford: Oxford University Press.
Latour, B., & Hermant, E. (1998). Paris ville invisible. Translated by Carey-Libbrecht, L.
31
and amended by Pihet, V. (2006).
Libby, R., Hunton, J. E., Tan, H.-T., & Seybert, N. (2008). Relationship incentives and
the optimistic/pessimistic pattern in analysts' forecasts. Journal of Accounting
Research, 46(1), 173e198.
Li, F., Minnis, M., Nagar, V., & Rajan, M. (2014). Knowledge, compensation, and firm
value: An empirical analysis of firm communication. Journal of Accounting and
Economics, 58(1), 96e116.
Lincoln, Y. S., & Guba, E. G. (1985). Naturalistic inquiry. Newbury Park, CA: Sage
Publications.
Lyon, D. (1994). Electronic eye: The rise of surveillance society. Minnesota: University
of Minnesota Press.
Lyon, D. (2001). Surveillance society: Monitoring everyday life. UK: McGraw-Hill
Education.
Lyon, D. (2006). 9/11, synopticon and scopohilia: Watching and being watched. In
K. Ericson, & R. Haggerty (Eds.), The new politics of surveillance and visibility (pp.
35e54). Toronto: University of Toronto Press.
Marx, G. (2003). A tack in the shoe: Neutralizing and resisting the new surveillance.
Journal of Social Issues, 59(2), 369e391.
Mathiesen, T. (1997). The viewer society: Michel Foucault's {\curr `}panopticon'
revisited. Theoretical Criminology, 1(2), 215e234.
Matsumoto, D., Pronk, M., & Roelofsen, E. (2011). What makes conference calls
useful? The information content of managers' presentations and analysts' dis-
cussion sessions. The Accounting Review, 86(4), 1383e1414.
Mayew, W. J. (2008). Evidence of management discrimination among analysts
during earnings conference calls. Journal of Accounting Research, 46(3),
627e659.
Mayew, W. J., Sharp, N. Y., & Venkatachalam, M. (2012). Using earnings conference
calls to identify analysts with superior private information. Review of Accounting
Studies, 18(2), 386e413.
Pentland, B. T. (1993). Getting comfortable with the numbers: Auditing and the
micro-production of macro-order. Accounting, Organizations and Society, 18(7),
605e620.
Pettigrew, A. (1992). On studying managerial elites. Strategic Management Journal,
13(S2), 163e182.
Power, M. K., & Gendron, Y. (2015). Qualitative research in auditing: A methodo-
logical roadmap. Auditing: A Journal of Practice & Theory, 34(2), 147e165.
Roberts, J., Sanderson, P., Barker, R., & Hendry, J. (2006). In the mirror of the market:
The disciplinary effects of company/fund manager meetings. Accounting, Or-
ganizations and Society, 31(3), 277e294.
Securities and Exchange Commission. (2000). Selective disclosure and insider trading,
17 CFR Parts 240, 243, and 249. Release Nos. 33-7881, 34-43154, IC-24599, File
No. S7-31-99, RIN 3235-AH82.
Solomon, J. F., & Darby, L. (2005). Is private social, ethical and environmental
reporting mythicizing or demythologizing reality? Accounting Forum, 29(1),
27e47.
Solomon, J. F., & Solomon, A. (2006). Private social, ethical and environmental
disclosure. Accounting, Auditing & Accountability Journal, 19(4), 564e591.
Solomon, J. F., Solomon, A., Joseph, N. L., & Norton, S. D. (2013). Impression man-
agement, myth creation and fabrication in private social and environmental
reporting: Insights from Erving Goffman. Accounting, Organizations and Society,
38(3), 195e213.
Solomon, J. F., Solomon, A., Norton, S. D., & Joseph, N. L. (2011). Private climate
change reporting: An emerging discourse of risk and opportunity? Accounting,
Auditing & Accountability Journal, 24(8), 1119e1148.
Spellmeyer, K. (1989). Foucault and the freshman writer: Considering the self in
discourse. College English, 51(7), 715e729.
Stoner, G., & Holland, J. (2004). Using case studies in finance research. In
Lee Humphrey (Ed.), The real life guide to accounting research: A behind-the-
scenes view of using qualitative research methods. Amsterdam: Elsevier.
Uche, C. O., & Atkins, J. F. (2015). Accounting for rituals and ritualization: The case of
shareholders' associations. Accounting Forum, 39(1), 34e50. March.
Vollmer, H. (2007). How to do more with numbers: Elementary stakes, framing,
keying, and the three-dimensional character of numerical signs. Accounting,
Organizations and Society, 32(6), 577e600.
 Accounting, Organizations and Society 58 (2017) 32e49

Contents lists available at ScienceDirect

Accounting, Organizations and Society

journal homepage: www.elsevier.com/locate/aos

Goffman's theory of frames and situated meaning-making in

performance reviews. The case of a category management approach in
the French retail sector
Philippe Lorino a, Damien Mourey b, *, Ge

raldine Schmidt b
a
ESSEC Business School, Paris, Singapore
b
IAE Paris, Universit

e Paris I Panth

eon-Sorbonne (Sorbonne Business School)

a r t i c l e i n f o a b s t r a c t

Article history: Following the scholarly view of accounting as a social practice, this research explores the role of ac-
Received 23 April 2015 counting in managers' ongoing efforts to effectively understand and influence organizational change in
Received in revised form the course of a series of review meetings. To study the role of accounting in meaning-making processes -
17 March 2017
i.e. collective processes of ascribing meanings to experienced social situations - we develop a processual
Accepted 19 March 2017
Available online 18 April 2017
and semiotic view of Goffman's theory of frames. Central to the paper's argument is the concept of
framing, defined as an ongoing social process of context production in an unfolding situation. In this
perspective, accounting numbers are viewed as signs mediating situated interactions. This theoretical
Keywords:
Accounting numbers
framework is applied to complex situations of negotiation between a retailer and sixteen suppliers,
Frames under a category management approach. The crucial finding of our research, illustrated empirically, is the
Framing plurality of competing frames and the occurrence of frame-shifting episodes, the process by which one
Meaning-making frame is suddenly replaced with another frame that has a deeply different way of narrating the situation:
Semiotic mediation this process is triggered by a specific event, and the frame shift has potentially significant effects on
Keying practices. The case study highlights the mediating role of frames, and the plasticity and vulnerability of
Shifting framing processes. It exemplifies the dual nature of accounting numbers in situated meaning-making:
they can be viewed simultaneously as generic models - parts of social frames - and singular events -
parts of the current situation. Due to this dual nature, they can act as mediators between a singular
situation and socially-constructed, generic classes of meaning.
© 2017 Published by Elsevier Ltd.

“A conversation has a life of its own and makes demands on its
own behalf. It is a little social system with its own boundary-
tendencies; it is a little patch of commitment and loyalty with
its own heroes and its own villains”.
Erving Goffman, Interactional ritual, 1967
Following Boland's suggestion that “the processes through
which participants frame and reframe a situation is an important
part of understanding the making of meanings” (Boland, 1989, p.
602), this research explores the role of accounting in managers'
ongoing efforts to effectively understand and influence organiza-
tional change in the course of review meetings. The focus is placed
* Corresponding author.
E-mail addresses: lorino@essec.edu (P. Lorino), mourey.iae@univ-paris1.fr
(D. Mourey), schmidt.iae@univ-paris1.fr (G. Schmidt).
on accounting as a social practice and process. This research seeks
to analyze how the situated utilizations of accounting numbers are
shaped, what impact they have on managerial practices, and how
they relate to meaning-making processes, i.e. collective social
processes of ascribing meanings to experienced situations. More
specifically, the study aims at shedding light on what accounting
numbers “say” to managers and what managers “do with” these
numbers when they meet in review meetings.
We choose to refer to Goffman's micro-sociological concept of
frames (Goffman, 1974) to explain the production of meanings
through managerial utilizations of accounting during review
meetings. Goffman's theory of frames is appropriate to examine the
social construction of narrative meanings in situations that are felt,
sensed and responded to, and to study the role of past experience -
of which accounting numbers are a key component - in that situ-
ated social construction. We elaborate a processual and semiotic/
mediating interpretation of Goffman's theory: central to our

http://dx.doi.org/10.1016/j.aos.2017.03.004
0361-3682/© 2017 Published by Elsevier Ltd.
 P. Lorino et al. / Accounting, Organizations and Society 58 (2017) 32e49
argument is the concept of framing, defined as an ongoing social
process of context production in an unfolding situation (Scheff,
2005; Vollmer, 2013). In this perspective, accounting numbers are
viewed as signs mediating situated interpretations: they both refer
to generic meaning frames and are engaged in malleable local
framing and (re)framing processes.
We use a case-based field study to consider our theoretical
proposition. An inter-organizational setting seems particularly
likely to incorporate the pluralistic meaning-making and different
co-existing - and often conflicting - world views found in experi-
enced situations, especially in the retail sector (Frances & Garnsey,
1996; Free, 2007, 2008; Hingley, 2005). Our empirical focus is on a
large French retailer, which introduced a category management
approach with sixteen major international suppliers, considering
them officially as “partners” rather than opponents or tough ne-
gotiators. Through a fourteen-month participant observation, we
developed a thick description of the history and changing institu-
tional context of these inter-organizational relationships, and ac-
counts of forty-eight official category reviews that took place
between the retailer and each of its suppliers. The crucial finding of
our research, illustrated empirically, is the plurality of competing
frames and the occurrence of frame-shifting episodes, the process
by which one frame is suddenly replaced with another that has a
deeply different way of narrating the situation: this is triggered by a
specific event or sign and has potentially significant effects on
practices. These findings contribute to a processual and semiotic
reading of Goffman's concepts of frames and (re-)framing,
emphasizing the mediating role of frames and the plasticity and
vulnerability of framing processes. They also exemplify the dual
nature of accounting numbers - as generic accounting models, and
as specific numerical values that can be viewed as situated events.
These two dimensions are rolled together in their situated utiliza-
tion. Thanks to this dual nature, numbers can mediate between the
singular situation and socially-constructed classes of meaning.
The paper is organized as follows. The first section surveys the
literature about accounting numbers in meaning-making pro-
cesses. The second section applies a processual and semiotic
perspective to Goffman's theory of frames. It characterizes the
relationship between accounting numbers and managerial prac-
tices as a dynamic meaning-making, framing and (re)framing
process. We then present our abductive, qualitative research
method in the third section. The fourth section narrates complex
inter-organizational negotiation situations in the retail sector and
examines the role of accounting numbers in the course of review
meetings. Thereafter, the discussion revolves around the role of
accounting in local framing and (re)-framing processes. The
conclusion summarizes the main contributions of this study and
suggests some possible avenues for future research.
1. Accounting numbers in meaning-making processes
1.1. Accounting as a resource for managing interactions
Mouritsen and Kreiner (2016) develop a radical critique of the
means-ends paradigm: “For both entrepreneurs and project man-
agers, there is no simple means-ends relation. Instead they effec-
tuate relations by engaging the surprises they meet. Many of these
surprises are produced in relation to accounting” (Mouritsen &
Kreiner, 2016, p. 28). This complements the claim that the role
played by accounting information goes far beyond that of providing
input for decision-making and that accounting is a major resource
€nsson, 1998;
for managerial action (e.g. Ahrens & Chapman, 2007; Jo
Swieringa & Weick, 1987; Hall, 2010; to name but a few). By shifting
the focus away from decision-making towards what happens to the
decision made, Mouritsen and Kreiner (2016) also introduce a
33
temporal dimension to the study of the role of accounting in
organizational life. Temporality should not be viewed as a
conception/implementation sequence, but rather as a continuous
movement full of surprises and potential disruptions in which ac-
counting plays a significant role, along with substantial managerial
efforts: It should thus be studied in relation to the development and
maintenance over time of a “context of knowledge and meaning for
unknown future actions” (Hall, 2010, p. 303). Accounting numbers
are thus fundamentally engaged in action. The relationship be-
tween accounting and managerial action should not be viewed as
straightforwardly deterministic. The meaning-making process is
part of managerial action, and the meaning accounting numbers
acquire in managerial situations is manifold, shifting, situated and
requires substantial managerial effort to be negotiated and estab-
lished, even if only temporarily. Mouritsen and Kreiner stress that
accounting is a resource for dealing with the necessary “in-
vestments and adjustments that have to be developed” (Mouritsen
& Kreiner, 2016, p. 21) once a decision has been made, considering
the unforeseeable surprises faced by managers in making sense of
and making do with the unexpected. Boland (1993) stresses that
organizational actors both actively contribute to the ongoing pro-
duction of the context in which accounting is interpreted, and
creatively generate new meanings symbolically using managerial
numbers “beyond their apparent literal use-in-context” (Boland,
1993, p. 127) in “an interpretive act” (Boland, 1993). Jo € nsson
(1998) argues that accounting is utilized to nurture a social prac-
tice of understanding each other and the other participants’ social
institutions through conversations. Shedding light on the multiple,
intertwined and complex meanings accounting numbers acquire
for participants in meetings, Pentland (1993) also demonstrates
that meanings are produced through cognition and behavior in
social situations.
Accounting numbers therefore take on manifold meanings in
concrete interactions, and do not conform to unequivocal pre-
determination. Admittedly, a long-established tradition of inter-
pretive research has already stressed the meaning-making
dimension of action, but the two most common ways of consid-
ering meaning-making involve some thorny issues.
Firstly, subjectivist and inter-subjective views specifically favor
analyzing interaction situations via the individual subjects and
their motivations, affects, and personal interpretations and repre-
sentations. The reasoning goes that accounting numbers are
unavoidably exposed to subjective and individual interpretations,
as well as mental representations and emotional responses. As a
result, they can endorse diverse meanings through the subjects'
psychological interpretations. But the over-emphasis on an inter-
subjectivist view to explain the plurality of numbers’ meanings is
criticized by some accounting scholars (e.g. Boland, 1989; Jo €nsson,
1998; Justesen & Mouritsen, 2011) for failing to give due recogni-
tion to the social construction of meanings in the day-to-day
experience of organizational life. Jo €nsson (1998) emphasizes the
social elements underlying any situated interaction: “Subjectivity is
virtually eliminated by the fact that the registered “text” is a con-
versation between competent persons (Atkinson, 1988) in their
area of competence and in a situation where their purpose is to
reach a solution that works” (Jo € nsson, 1998, p. 432). In other words,
the interpretation of what is going on in a situated interaction re-
quires a new grasp of “contextual sensibility” (Jo €nsson, 1998, p.
415), suggesting that a situation is not limited to its immediate,
observable elements but requires a broader understanding that
includes its context. ANT-inspired researchers are also fierce critics
of a strict inter-subjectivist perspective: Justesen and Mouritsen
(2011) suggest that accounting should not be viewed as “a matter
between an accounting report and an inquisitive mind” (p.180)
since the situated production of meanings of and with accounting is
34 P. Lorino et al. / Accounting, Organizations and Society 58 (2017) 32e49
a social, recursive, and ongoing process. Accounting numbers do
not just help describe situations but, theorized as actants, they also
perform them. As a result, the inter-subjective view entails not only
methodological difficulties, as mental representations are not easily
observed (Jo €nsson, 1998), but also theoretical difficulties. This
makes it hard to understand how the psychological inter-subjective
conversation and the broader social context are linked in coherent,
continuous and meaningful ways.
Secondly, in order to address the question of how a social
perspective can be constructed from an (inter)subjective view,
some scholars have proposed combining social and cultural struc-
tures (a “global” or “macro-level” of analysis) with subjective be-
haviors and situated interactions between actors (a “local” or
“micro-level” of analysis) (e.g. Ho, Wu, & Wu, 2014; Scapens &
Macintosh, 1996). To explain the micro-foundations of the macro-
social level, this vertical “macro-micro” conceptualization often
embraces a mimetic or a sharedness/commonality link. For
instance, the “person-organization fit” emphasizes the psycholog-
ical mechanisms of identification between individual and organi-
zational goals (Ho et al., 2014). Alternatively, the “sharedness” view
claims that macro-structures can emerge, exist and last since actors
share cognitive representations, interpretive schemes, values or
habits of action (Scapens & Macintosh, 1996). This leads us to “see
institutionalized, monolithic sets of meanings being imposed by
[management accounting systems]” and thus ignore the “actor's
creative, open possibilities for making meaning” (Boland, 1993, p.
140). To some extent, the sharedness or commonality frameworks
neglect the dynamic dimension of situated interactions, although
many empirical studies (Ahrens & Chapman, 2007; Boland &
Pondy, 1986; Jo €nsson, 1998; Nahapiet, 1988; Pentland, 1993) sug-
gest that dialogical, creative interactions between different,
pluralistic world views play a role in the shaping of collective ac-
tivity that is at least as important as common views.
The continuous multiplicity, uncertainty and controversies that
are potential practical effects of the utilizations of accounting, and
the fact that organizational actors must constantly make do with
such utilizations, are not always highlighted by accounting
scholars. Swieringa and Weick (1987), for example, analyze the role
played by the imperfect and sometimes conceptually inconsistent
ROI accounting model as a shared tool for collective sensemaking,
and view accounting utilization as a unifying process leading to
managerial actions that are not really contested or a source of
controversy. In their practice-based view, Ahrens and Chapman
(2007) implicitly suggest a process of convergence and
consistency-strengthening to explain how social practices emerge
from situated activities, rather than disagreement and conflict: they
quote Shatzki, for whom “practices are thus constituted by “tangle
[s] of samenesses and similarities'” (Schatzki, 2001, p. 42), and they
argue themselves that “practices depend on the intended, mean-
ingful relatedness between activities with respect to outcomes,
clients, practitioners, techniques, resources, strategies, institutions,
etc” (Ahrens & Chapman, 2007, p. 23). Admittedly, their theoretical
perspective leaves room for limited adaptations to local contexts,
but only within defined meaning guidelines elaborated at a higher
level of the organization. Even if only loosely defined, such guide-
lines frame local contextual adaptations: “Social order arises from
actors' ongoing efforts at developing their actions as part of certain
fields of practices, that is to say, with reference to the fields’ un-
derstandings, rules and engagements” (Ahrens & Chapman, 2007,
p. 23). Eventually, drawing on other theoretical perspectives that
fundamentally recognize heterogeneity and pluralism of views,
other scholars more or less explicitly suggest a view of meaning-
making which to a certain extent underestimates the potential
for multiple, simultaneous, competing and sometimes conflict-
ridden explanations of a given situation. Accounting is then
involved in a process of convergence or linking up between
different worlds. For instance, considering accounting as a
“boundary object”, Briers and Chua (2001) show how accounting -
in their case an ABC system - contributes to overall consistency in a
context of professional or cultural diversity (Gerdin, Messner &
Mouritsen, 2014), and do not focus on challenging dissonances.
Justesen and Mouritsen (2011) highlight that, although most ANT-
inspired accounting research starts by acknowledging pluralistic
views of the world to analyze the role of accounting, this pluralism
seems to fade away over time, and accounting gradually stabilizes
into fabricated entities. To sum up, taking the view that accounting
is a resource for continuous meaning-making concerning poten-
tially surprising, divergent and precarious situated interactions
offers a way to better relate accounting and its performativity to
ongoing managerial work.
1.2. Accounting and the concept of mediation
Several streams of research have tried to theorize the relation-
ship between accounting numbers and managerial practices as a
process in which “structures” are permanently engaged in singular
situations, not in a “vertical” perspective as transcending cate-
gories, but as semiotic mediations of meaning-making and action. A
range of terms are used to label these mediations, including “signs”
(Graham, 2008), “calculative inscriptions” (Vinnari & Skærbæk,
2014), “intervening media” (Cooper & Ezzamel, 2013), “non-hu-
man actors” (Mouritsen & Thrane, 2006), “differentiated univer-
sals” (Gallhofer, Haslam, & Yonekura, 2015) and “boundary objects”
(Dechow & Mouritsen, 2005; Quattrone & Hopper, 2005). In their
diversity, these contributions have gradually developed another,
more dynamic, view of the interplay between situated action, cal-
culative devices and social schemes of meaning, taking into account
the recursive nature of meaning-making and action processes. To
understand meaning-making and the role of accounting in orga-
nizing processes, accounting researchers have to go the extra mile
and “understand existence as variable” (Justesen & Mouritsen,
2011, p.184) and filled with constant pluralistic world views that
interact, are transformed, undermined, or dominated, and some-
times openly clash in unforeseeable ways with unexpected timing.
Recent accounting research has highlighted the need to better
reflect pluralistic and sometimes competing world views in ac-
counting theorizing (e.g. Vinnari & Skærbæk, 2014; Yang & Modell,
2015). Some authors have also made this the starting point of their
research, as in Chenhall, Hall, and Smith (2013) who demonstrate
how “accounts are involved in ongoing contests and struggles as
various groups advance particular interests” (Chenhall et al., 2013,
p.269). They draw on Stark’s (2009) concepts of “heterarchy” and
“organizing dissonance” to highlight the friction between different
world views. Following in their footsteps, we suggest embracing a
“horizontal” approach, elaborating a theoretical framework that
takes into consideration the role of accounting in meaning-making
processes characterized by pluralistic situations. In this view, so-
called “structures” (e.g. accounting models) are semiotic media-
tions that mediate the ongoing meaning-making process in specific
situations and interactions, in the same way as language mediates
specific conversational interactions and situated discursive
practices.
Accounting numbers have two dimensions that are two sides of
the same coin. First, accounting is expressed in the form of specific
numerical values participating in a collectively-experienced, sin-
gular situation, which they can disrupt by showing surprising re-
sults that may challenge the accepted meaning of the ongoing
activity. Second, accounting numbers are simultaneously part of
social classes of meaning (such as accounting concepts and rules,
and accounting models adopted by organizations to conceptualize
 P. Lorino et al. / Accounting, Organizations and Society 58 (2017) 32e49
their activity in an economic syntax), which, visibly or otherwise,
convey hypotheses about the organization of activity and economic
behavior. Thanks to this dual nature, accounting numbers are signs
that mediate situated activity.
What do we mean by “mediating signs” applied to accounting?
Meaning-making processes require social communication, the
memory of past experience, and projection into the future(s). Signs
are double-faced artefacts that are simultaneously participants in
situated interactions, and elements of generic, social schemes of
meaning. They can dynamically relate the singular situation to
social and historical experience: semiotic mediations extend the
boundaries of a situation from immediate interactions to a broader
social, cultural and historical experience. Signs instantiate distant
places and actors, past and future, into the present and local action-
in-progress (Lorino & Mourey, 2013). Signs are not purely mental
moves by individual psychological subjects; it is hard to see how
such moves could serve as socializing devices. Based on collective
experience, signs convey shared dispositions to act in a certain way
in a certain type of situation, without determining how the actors
will in fact act. Rather than being the determining sources of action,
mediating signs are resources of situated action that leave room for
the evolving, pluralistic and debatable understanding of the situ-
ation. Their meaning, i.e. the way they relate the “here-and-now”
situation to social classes of meaning, requires ongoing effort to
discuss and understand the situation collectively in the light of
social experience.
Mediating artefacts, particularly accounting numbers, are
neither entirely malleable by actors’ present interactions (since
they are constrained by history and sociality), nor immutable
structures imposed on situated interactions (since their meaning
can be transformed when involved in a situation). They are
involved in an ongoing social process of meaning-making in situ-
ations of plurality that may be destabilized at any moment in time.
Accounting is not an isolable entity, but a malleable phenomenon, a
multiple being that is always made up of the multiple connotations
attached to it by actors in a situation. A unifying narrative is
required to confer meaning on it (Mouritsen, Larsen, & Bukh, 2001).
That is the role played by social frames as studied by Goffman.
2. A processual and semiotic view of Goffman's theory of
framing
Goffman (1981b) argues that humans do not construct their
reality from scratch at every encounter, but engage frames to
organize and ascribe meaning to their ongoing experience in a
social situation.
“Given their understanding of what it is that is going on, in-
dividuals fit their actions to this understanding and ordinarily
find that the on-going world supports this fitting. These orga-
nizational premises e sustained both in the mind and in activity
e I call the frame of the activity.” (Goffman, 1974, p.247)
Goffman defines a social situation as “an environment of mutual
monitoring possibilities, anywhere within which an individual will
find himself accessible to the naked senses of all others who are
‘present’, and similarly find them accessible to him” (Goffman,
1964, p. 135). This concept is very appropriate to study how man-
agers make meaning of and with accounting in physical meetings.
However, Goffman himself worked not only with situations
“bounded by physical places”, but also “with a less explicit concept
of situation”, that “does not limit the theorist to the interaction of
people who are literally face to face” (Riggins, 1990, p. 4). Goffman's
main research interest is in the practical necessity for any partici-
pant in a social situation to continuously define “what exists, what
35
happens and what matters” (Gitlin, 1980, p. 6). This is precisely
where Goffman's theory of frames - on the narrow crest between
structuralist and interactionist views (Cefaï & Gardella, 2012) - in-
tertwines with and paves the way for a processual and semiotic/
mediating turn. The framing perspective focuses on the ongoing,
mutual process of meaning-making and ordering of social situa-
tions (Czarniawska, 2006; Vollmer, 2013) and sheds light on how
frames, viewed as narrative mediations, enable people to make
meaning of experienced situations collectively, beyond the imme-
diate present interaction and its visible, explicit and literal com-
ponents alone. When the mediation introduced by frames is
categorized as narrative, we refer to Ricœur's concept (1990) of
“narrative”, i.e. a specific treatment of time, meaning and the close
connection between the two. Narrative “emplotment” transforms
linear flowing temporality into a non-linear “logical (cause-effect)/
chronological (precedent-subsequent)” temporality. It defines
specific events and acts (e.g. managerial acts) as simultaneously
temporal and logical punctuations (“that's what usually happens
then”) as a situation develops and progresses from a beginning to
an end.
Frames are neither structural determinations of the meaning of
situations, nor are they individual schemes of cognition or stabi-
lized routines. Instead, they offer meaning perspectives whose
content is still malleable and can evolve in the course of social
situations. Frames convey a shared disposition to understand a
situation in a certain way, but they do not predict how participants
will actually act. Situated meaning-making is creative. Therefore,
the course of a situation is always indeterminate and subject to
unexpected developments. Participants interact with one another
based on assumptions they are making about events and others'
perceptions, states of mind and doings. It is paramount to bear in
mind at all times the fluidity, plasticity and vulnerability with
which social situations are ordered and remember that the contexts
to our doings are constantly being produced by the participants’
practical ability to adapt in real time to what they feel is going on
“diffusively and affectively” (Vollmer, 2013, p. 34). A frame is
therefore the context within which participants understand what is
going on in a social situation, and framing is an ongoing social
process of context production in an unfolding situation (Scheff,
2005; Vollmer, 2013).
“Not frames as ready-made products or contexts for actual oc-
casions but processes of framing as producing social order within
a collective of participants through producing contexts for ac-
tivities and experience are at stake.” (Vollmer, 2013, p. 28)
Through framing, actors actively extend the immediate, face-to-
face situation to its assumed or remembered pasts, predicted or
projected futures, and other socially distant but connected settings,
and they attach it to a narrative which is always in-the-making. In
the framing perspective, the relationship between social/institu-
tional “structures” and situated actions is not a “container/content”
inclusion relationship: situated actions are not particular occur-
rences of an institutional structure. It is rather a semiotic and dy-
namic meaning-making relationship: meanings involved in
experienced events make some practical transformation of the
situation achievable. This makes managers’ work conversational, as
they cooperate in producing a context.
Framing events and activities and the emerging production of
context for participants' activities are two aspects of the same
process (Scheff, 2005; Vollmer, 2013). Frames are reconceived at
the same time as they are brought into play in the immediate
experience of social interaction. Frames are permanently subject to
experience-led transformations. Framing has the double meaning
of framing the situation by involving a social frame - applying frames
36 P. Lorino et al. / Accounting, Organizations and Society 58 (2017) 32e49
- and adapting/transforming/building frames - making frames, in an
interconnected and inseparable move. Goffman's concept of frame
is a recursive model of mutual reflexive awareness, through the
process of keying (Scheff, 2005). Goffman defines keying as “the set
of conventions by which a given activity, one already meaningful in
terms of some primary framework,1 is transformed into something
patterned on this activity but seen by the participants to be quite
something else” (Goffman, 1974, pp. 43-44). Keying therefore refers
to the reflexive re-examination of a given frame that is adapted to
new circumstances without altering its basic form and emplot-
ment. It transforms the frame's epistemic status, from a mediation
of the present situation to the object of specific reflection, such that
it can be redesigned and adapted. It is the unique, situational
lamination of frames - the never-ending outcome of a process of
framing that resembles candyfloss2 as new frames are attached to
the sticky substance in the making - that produces the context of
the interaction and explains the existence of different meanings
with respect to ongoing activities. This, of course, does not preclude
mishaps, as in the Goffmanian perspective the variability of expe-
rience always lurks at every corner of interactions, preventing any
structuralist determinism (Manning, 2008; Que
re

, 1989). There is
the distinct possibility of disruptions such as “ambiguities and er-
rors” in framing, “breaking frame” and “the manufacture of the
negative experience” that are among the vulnerabilities of human
experience (Manning, 1977, p. 1363). Certainly, framing does not
reveal the intentions of the social situation's participants, who may
exploit (re)framing strategies for ulterior motives.
This perspective - the mediating role of frames and the
malleability of local framing processes - is relevant to the role of
accounting in meaning-making. Re-conceptualized as signs medi-
ating situated interpretations, accounting numbers are simulta-
neously the expression of a socially-constructed class of meaning
(and therefore part of a generic narrative frame taking the form of
an accounting model) and an element of the here-and-now situa-
tion (an experienced event in a concrete framing process, taking the
form of singular numerical values). The managers concerned,
finding themselves experiencing contradictions and facing an un-
certain future, in practice enact systems of quantification to
construct the meaning of action-in-progress.
2.1. Goffman's theory of framing in the accounting literature
Organization (Cornelissen & Werner, 2014) and accounting
studies have already drawn on Goffman's theory of frames
(Solomon, Solomon, Joseph, & Norton, 2013), notably via Callon's
concepts of ”framing” and “overflowing” applied to markets and
economic figures (e.g. Callon, 1998; Kastberg, 2014; Roberts &
Jones, 2009; Skærbæk & Tryggestad, 2010; Vinnari & Skærbæk,
2014) and Vollmer's notions of “reproductive” and “consumptive”
frames (Vollmer, 2007). Yang and Modell (2015) have also drawn on
the theory of framing as re-interpreted by social movement theo-
rists (Bendford & Snow, 2000). In Callon's perspective, framing
establishes boundaries, and boundary-overflowing triggers and
1
In this paper we do not draw on Goffman's definition of social and natural
primary frameworks. The distinction between primary and secondary frameworks is
not meaningful for our demonstration, which focuses on framing and keying pro-
cesses. However, Goffman suggests that primary frameworks are central elements
of the culture of a social group: “I say primary because application of such a
framework or perspective is seen by those who apply it as not depending on or
harking back to some prior or “original” interpretation; indeed a primary frame-
work is one that is seen as rendering what would otherwise be a meaningless
aspect of the scene into something that is meaningful” (Goffman, 1974, p. 21).
2
The telling candyfloss metaphor was used by Goffman in another context in
Stigma, 1963, p.57.
stimulates change. This highlights the performativity of accounting
numbers: accounting numbers not only describe situations, they
also perform them. Accounting is thus viewed both as a “framing
device” and “sources of overflow, causing destabilization” of orga-
nizational relationships (Kastberg, 2014, p. 743). As Yang and
Modell (2015) stress, in the Callon-inspired ANT perspective,
frames appear “as a temporary and fragile state which may be
destabilized at any point in time” (Yang & Modell, 2015, p. 2). The
question is: why is the frame destabilized, with the meaning
challenged and reconstructed, at a particular point in time and
space? Why here and now? In a semiotic perspective, any media-
tion inherently involves overflow, since mediation segments and
categorizes the unstructured, shapeless continuum of experience
by selecting and bounding areas of meaning corresponding to a
specific system of signs. Therefore, real experience continuously
and unavoidably overflows, but this is accepted as an inherent part
of the meaning-making process. Explaining change by overflow
alone is not sufficient.
For their part, Yang and Modell (2015) favor a “social move-
ment” reading of the concept of frames, putting the emphasis on
“framing” as the translation of general social ideologies into classes
of situations. This three-level view of “framing” (ideology, frame,
situated interaction) clearly stresses the mediating/translating role
of frames as simultaneously referring to social categories and their
active involvement in managerial situations. One difference from
Goffman's approach may be in the relative de-emphasizing of sit-
uated managerial interactions. Consistent with their long (10-year)
macro-perspective, Yang and Modell's research method draws
mostly on interviews and document analysis, with only a minor
role for participant observation of interactional situations.
Ultimately, Vollmer's use of Goffman's theory of framing
(Vollmer, 2007, 2013) converges with our research interest, as it
offers a social perspective on the dynamic process of meaning-
making for accounting numbers within the confines of social situ-
ations. Vollmer focuses particularly on the role of accounting in the
ordering process of social situations and, more specifically, theo-
rizes the three qualities (symptomatic, calculative and existential) of
numbers that make them active elements in the framing process:
“symptomatic qualities relate numbers to realities, calculative
qualities relate them to other numbers, existential qualities relate
them to participants of social situations” (Vollmer, 2007, p. 593).
This suggests a semiotic reading of Goffman's theory of frames,
where accounting numbers are viewed as systems of signs and can
relate singular situations to mathematized models. The keying
moves triggered by the utilization of accounting numbers between
“reproductive frames” - referring to the calculative properties of
numbers organized by arithmetic calculation rules - and
“consumptive frames” - highlighting the symptomatic properties of
numbers realized “through specifying their significance as carriers
of information” (Vollmer, 2007, p. 588) illustrate the ongoing
transformation process of meanings and action through the
handling of those signs. However, Vollmer's strictly theoretical
approach favors an abstract view of frames (reproductive and
consumptive frames are not materialized in concrete, situated
schemes of managerial action) and may be difficult to operation-
alize in the field study of managerial interactions.
In our research, we adopt a specific reading of Goffman's theory
of frames that is semiotic (frames are viewed as semiotic media-
tions instantiating social and historical experience in singular sit-
uations) and processual (it emphasizes framing as an ongoing
process of ordering social situations and collective meaning-
making, rather than frames as structures). This involves a narra-
tive view of temporality, leading us to apply Goffman's methodo-
logical approach to the practical study and theorizing of (re)framing
processes in review meetings that were experienced first-hand by
 P. Lorino et al. / Accounting, Organizations and Society 58 (2017) 32e49
one of the authors (Strong, 1983). Our research question can now be
specified into: how exactly do the situated utilizations of accounting
numbers connect to framing and re-framing processes? This question
was explored through a case study of an inter-organizational
negotiation in the retail sector.
3. Research method: an abductive form of participant
observation
The case studied concerns the business relationships between a
large French retailer and several international suppliers of con-
sumer goods such as Nestle
, Unilever and Danone. In February
2006, the head of the retail company's trade marketing depart-
ment, Nicolas,3 launched a category management approach with
sixteen international suppliers. This collaborative turn was offi-
cially for the purpose of enhancing customer value. Nicolas
declared that he wanted to “upgrade” relationships with suppliers,
which had previously focused on the rather adversarial annual
round of commercial negotiations. He explained that confronta-
tional relationships were no longer good for business and that the
time had come to adopt a more cooperative stance and treat sup-
pliers as “strategic partners”. This initiative was taken just as the
dramatic effects of a new law introduced by the French Ministry of
Trade were first beginning to show in some performance indicator
trends. The buyer-retailer’s long-established business model was
destabilized. Novel, more collaborative inter-organizational prac-
tices were explored to bring about a yet-to-be invented business
model of as yet unproven sustainability.
The ambiguity of supplier-retailer relationships in the retail
sector has been clearly shown by Frances and Garnsey (1996), along
with the controversial aspects of category management (Free, 2007,
2008). Studying a similar approach in the UK, Free suggests that the
management accounting techniques engaged are not collaborative
by design but may be permeated by a coercive or enabling orien-
tation (Free, 2007). He also implies that some organizational actors
have the power to impose the context of interpretation of ac-
counting during the category review process - viewed as the
“dominant meaning system for action and intervention” (Free,
2008, p. 649) - by using trust and collaboration rhetoric as a
discursive resource to preserve the veneer of joint collaboration. As
a result, the way the concept of category management and its
associated accounting techniques will be enacted in practice and
play out in a specific buyer-supplier relationship is a matter for
empirical investigation. This calls for close scrutiny of situations of
accounting utilization, within the confines of category reviews,
bearing in mind that every participant remains constantly con-
nected to a nexus of interactions they cannot be entirely detached
from: this may be a source of disruptions or surprises during
category reviews, leading organizational actors to attempt to
creatively adapt to them - successfully or otherwise. The inter-
organizational negotiation situations studied in this research offer
an edifying example of complexity and situated meaning-making,
in a relational situation of strategic negotiation between two or
more groups of people referring to a plurality of social frames. The
3
We refer to the head of the retailer's trade marketing department as Nicolas
throughout the paper. He is a member of the Executive Committee of the French
hypermarket firm concerned, and initiated the strategic category management
approach. Nicolas, the researcher doing fieldwork, and the coordinator of the
category management project were the only three individuals who attended all
top-to-top meetings and most preparatory category reviews. They enjoyed a global
view of the evolving category management process across all sixteen inter-
organizational relationships. Nicolas played a crucial role in promoting the “small
gains” achieved across the retail company to keep skeptics (buyers) at bay and
attract support from sponsors.
37
category management approach studied is underpinned by a
standardized accounting and information infrastructure to which
the researcher had permanent access. It is described in the table
below: (see Table 1).
3.1. About the researcher's positioning in the field
This research belongs to the “naturalistic” tradition of man-
agement accounting research, an alternative stream of research
that has yielded insightful but fragmented studies of management
accounting practices (Baxter & Chua, 2003). Fieldwork and the
case study method are considered to have the potential to enrich
our understanding of accounting in action (Ahrens & Chapman,
2006; Ahrens & Dent, 1998; Baxter & Chua, 2003; Otley & Berry,
1994), management accounting as practice (Ahrens & Chapman,
2007) and accounting change (Andon, Baxter, & Chua, 2007). A
rich description of everyday organizational life is not sufficient to
yield a significant research contribution. It is also necessary to
clarify the theoretical intention, to link the writing-up of the case
with the theoretical framework, and to enable readers to assess
the trustworthiness of the accounts and the plausibility of the
claims. In our case, the fieldwork methodology used was inspired
by Goffman's research practices, which he informally commented
on during a conference in 1974 (Goffman, 1989). It entails
everyday “insider” observation of the activity of actors involved in
the category management project. One of the authors of this
research conducted a fourteen-month participant observation,
indicating an enduring relationship with the actors on site. Our
methodology also requires the inter-organizational events wit-
nessed first-hand to be related to the broader institutional, his-
torical, economic and interpersonal context in which they are
embedded. Theorizing is thus viewed as “a lived, social process of
sense-making” (Chua, 1986, p. 585) and is deeply grounded in the
logics of the field's specific modus operandi (Ahrens & Chapman,
2006). Such exploratory research involving long-lasting immer-
sion is not easy to sustain over time (Ahrens & Chapman, 2006;
Suddaby, 2006). The researcher must expect to be questioned by
the people he intends to study.
In our case, access to the field of research was made possible by
the researcher's credentials as a management accountant familiar
with supplier-retailer relationships. Before joining academia, he
worked for Unilever, then for the case study retailer for ten years. To
consolidate long-term access to field actors, he agreed to be in
charge of producing monthly inter-organizational category-specific
scorecards and monitoring category budgets without any financial
compensation. During the sixteen kick-off meetings held at the
beginning of 2006, the researcher was introduced to all suppliers as
“a researcher getting his hands dirty” and “the guarantor of
impartiality and reliability in the numbers”. He took great care to
strictly abide by the rule of not taking sides with any team. He made
sure reporting deadlines were met, and handled coordination and
standardization of all flows of information coming from the
different departments involved in the category management proj-
ect (hypermarkets, logistics, trade marketing and buying), as well
as from the suppliers' teams. This operational positioning allowed
him to benefit from “deep familiarity” (Goffman, 1989): the
researcher, through her/his operational contribution, is assigned “a
place” (Favret-Saada, 1977) in the social system under study and, to
use a metaphor from the natural world, becomes a kind of
chameleon.
3.2. The empirical material obtained and used by the researcher
The following table presents the empirical material used in the
paper and how it was obtained: (see Table 2).
38 P. Lorino et al. / Accounting, Organizations and Society 58 (2017) 32e49

Table 1
Accounting and information infrastructure for the category management process.

Strategic planning ➢ To become “category captain” a supplier must submit its strategic vision (in quantitative and qualitative terms) for the category market at
domestic level, for its relationship with this retailer and for the main joint growth levers at hand.
➢ Every category manager in the retail company prepares a strategic assessment of their category (including a SWOT analysis, the forecasts for
the retailer's brand, premium and entry-level products and an evaluation of the supplier best aligned with this plan).
Joint performance Monthly scorecard with category-specific KPIs covering:
measurement ➢ The financial performance by the category and the category captain (sales revenues and profit)
➢ Market shares (per category segment, retailer and supplier)
➢ Supply chain information (stock level at distribution centre, stock-out rates, etc.).
➢ Buying data (value of purchases made by the retailer)
➢ Pricing indexes (base 100 - per retailer and supplier)
➢ Detailed category profitability for the retailer
Joint business ➢ Comparison of the supplier's and retailer's respective evaluations of the past year's performance (in quantitative and qualitative terms)
planning covering all aspects of the business relationships (pricing, promotion, stock-out, merchandising, commercial negotiations, etc.).
➢ Financial performance targets set for the year 2006 for the category and the “category captain” in terms of sales revenues and profit,
identifying the additional growth provided by the category management projects.
➢ Detailed joint action plan at category level, including measurement of the financial impact of each lever.
Category review ➢ Monthly category review involving inter-organizational actors involved at the category level (buyer, trade marketing, category manager).
➢ Quarterly top-to-top meeting involving the highest-ranking officials of both companies: their respective CEOs, the retailer's head of buying,
trade marketing, supply chain and category manager, and the category captain's counterparts.
➢ A standardized framework detailing the content of management reporting and key points of category statements was assigned to every top-
to-top meeting, discussed in preparatory meetings and sent out to the meeting's participants a few days in advance.

The technical difficulty of standardizing all data from unrelated,
dispersed information systems gave the researcher an opportunity
to meet all the relevant actors one-to-one in their offices, in the
retail company's functional departments and in suppliers' teams.
On average, the participant-observer spent three days per week in
the field for fourteen months, attending inter-organizational
meetings as often as requested. He had a desk in the trade mar-
keting department's open plan office, which meant he could have
on-the-spot unstructured interviews with key organizational ac-
tors about events witnessed incidentally. He also attended more
than a hundred official “kick-off”, “preparatory”, “top-to-top” and
“backstage” inter- and intra-organizational meetings. He never
took a proactive part in these meetings, but answered technical
questions regarding the production of figures. Considering the
sensitivity of the discussions, no recording of inter-organizational
meetings was possible. However, the researcher was sent the
final slide presentations for every category review a few days in
advance. This made note-taking during meetings easier, since
quotes, remarks and comments could be chronologically related to
the slide on display when they were made. It also strengthened the
researcher's ability to focus on gestures, embarrassment, expres-
sions given off, loss of poise, breach of normalcy, last-minute
changes of slides whenever they happened, and to engage in the
kind of analytical work recommended by Goffman when doing
fieldwork (Goffman, 1989; Hacking, 2004). Short memos were
written to report on the questions raised by certain specific scenes,
events or turn-taking observed. This research was conducted under
the supervision of two senior researchers, providing a constant link
to the academic world. In particular, the decision was made to apply
the comparison principle and to observe the trajectories of all
sixteen retailer-supplier relationships rather than focusing on a
selected few. We also abided by Goffman's insistence that re-
searchers should not “get uptight about one single data-source”
(Strong, 1983, p. 351) and should cross-reference what people said
with internal and external secondary data. We centered our anal-
ysis on managerial activities that took place in specific spatial and
material arrangements and considered accounting information, not
in isolation (Hall, 2010), but when it was utilized by managers as
part of a wider range of information through speeches, gestures,
slides, and graphics. There was no formal coding of the empirical
material, as such a gigantic task would have seriously impaired our
ability to do the fieldwork, witness unanticipated events and
conduct our operational assignment satisfactorily.
3.3. An exploratory and abductive inquiry
One aim of the research was to identify the emergence of
coherent narratives from the actors' behaviors during category
review meetings (Czarniawska, 1998). Therefore, through his
participant role the researcher sought to identify the emergence of
different frames, viewed as narrative regularities and defined
through their opposition and the systematic comparison of pat-
terns of action across the category reviews concerning all sixteen
relationships over fourteen months. The research was thus
abductive, i.e. it was an exploratory inquiry that focused on the
emergence of new hypotheses, shaping a testable, plausible
narrative to account for the puzzling situations experienced by
researchers. It combined intuition and reasoning, first-hand expe-
rience of puzzling events and attempts to bring out new social
patterns, discussions with field actors and scholars and cross-
referencing analyses (Lorino, Tricard, & Clot, 2011). More particu-
larly, the few moments of disruption, embarrassment and loss of
meaning that occurred during the more than one hundred, formal
and informal category reviews offered critical insights into the
regularity of the framing process that “ultimately has to be the
outcome of a practical training of participants” (Vollmer, 2013, p.
30). Incidentally observed moments of disruption such as frame-
breaking or frame-shifting were methodologically interesting, not
because they were extraordinary per se, but rather because they
helped to understand the ordinary by contrasting it with the
extraordinary: the way category review participants mostly coop-
erated “in maintaining expectations” (Vollmer, 2013, p. 28) and
produced order through their skill at local framings. These scenes,
mishaps and moments of hesitation made it possible to identify
“narrative cues” - examples include a dramatic scene when a buyer
slammed a competitor's leaflet down onto the table, to send a
message to a supplier in the room; a deadly silence signaling the
difficulty of carrying on from the point at which the discussion had
just ended; a murderous look; a breach of social conventions; a
feigned indifference; a hasty mouse click; the dramatization of a
specific number; a speaker suddenly having breathing difficulties;
new slides added at the last minute. Such events, through their
departures from usual proceedings, revealed the narrative regu-
larities that were disturbed by the situation. At the least they
suggested plausible hypotheses. Such abductive clues were not
always easily readable and it took a long time to assign a frame to
some of them. Discussions with the two senior research supervisors
 P. Lorino et al. / Accounting, Organizations and Society 58 (2017) 32e49 39

Table 2
The empirical material used and how it was obtained.

Direct observation of category
reviews and preparatory
meetings
Direct participation in meetings
about the researcher's
operational mission
Main internal data sources
External secondary sources
➢ Unlimited access to 3 rounds of 16 quarterly top-to-top category review meetings - 48 meetings held in AprileMay 2006,
JuneeJuly 2006 and SeptembereOctober 2006. Each meeting lasted an average 2 h and most took place at the retailer's head
office. Some were also held in hypermarket stores or at suppliers' production sites.
➢ Unlimited access to inter- and intra-organizational preparatory meetings for top-to-top reviews. Observation of at least one
preparatory meeting for every top-to-top meeting.
➢ Meetings to prepare the yearly category budgets with all category managers in FebruaryeMarch 2006.
➢ Meetings to solve long-lasting technical problems with data collection for the production of inter-organizational scorecards.
➢ Category review presentations sent out before each top-to-top meeting.
➢ Access to the financial and reporting databases of the retailer's hypermarkets and buying departments.
➢ Access to the trade marketing department IT platform where all internal memoranda were uploaded.
➢
, 2000).
Two French parliamentary reports detailing coercive business practices in the French retail sector (Charie
➢ New legal framework introduced by the French Ministry of Trade, overhauling and potentially destabilizing existing
negotiation and pricing practices in France's retail sector.
➢ Press releases commenting on the new regulations and the buyer-supplier relationships.
➢ Documentation from ECR (Efficient Customer Response), a joint trade and industry body promoting greater responsiveness to
consumer demand in the retail sector. Observation of two meetings at ECR's Paris headquarters about standard-setting for
“shelf-ready packaging” in France's retail sector.

to make sense of them took place in parallel to the participant completely aligned their views on how best to deal with suppliers.
observation, in order to decipher what these episodes told us about Buyers had to be “on board with” the category management proj-
the emergence of types of situations, corresponding to narrative ect, as the French retailer's CEO had given it his backing and took
archetypes, faced by these organizational actors. These discussions part in top-to-top meetings with his counterparts from the sup-
led recursively to new hypotheses, re-orienting the theorizing of pliers. But they were ready for ambush, and freely voiced their
the case and the participant observer's inquiry in the field. opposition to any sharing of information with suppliers that would
The following section presents up-close analysis of two con- supposedly undermine their leverage as negotiators.
trasting episodes involving utilizations of accounting numbers.
4.2. An even balance of power between organizations
4. Case study
The launch of a category management approach was by no
means an either-or choice which simply replaced confrontation
4.1. The strategic plan to transform the model of retailer-supplier
with cooperation. The managers of these relationships were always
relationships
skating on thin ice, being required to facilitate taut, no-holds-
barred yearly commercial negotiations while also constructing
Moving from a confrontational to a “partnership” approach to
and preserving a common ground where more synergistic,
supplier management is no mean feat in the French retail business.
medium-term business practices would still thrive. The balance of
Business relationships in France have a long history of regular
power between them was fairly even. For example, the studied
showdowns between buyers and suppliers' negotiators, revolving
retailer accounts for 25% of Nestle
’s sales in France and Nestle
ac-
around their yearly negotiations. All suppliers' negotiators can
counts for 10% of the retailer's total grocery sales. The selected
recall episodes of arm-twisting by buyers to pressurize them: a
category captains were international consumer goods companies
refusal to list a new product, unilateral product delisting, sabotage
and knew how to play the long game with retailers. After all, at
of promotional activities, a refusal to let sales representatives visit
category level the top two international suppliers for the 30 most
stores etc. In this confrontational, short-term mode of management
concentrated markets have a combined market share of 71% in
that has so far been dominant in inter-organizational relationships,
France (Charie
, 2000). But both the retailer and the supplier shared
buyers play a key role and will do whatever it takes to maximize the
the same uncertainty as to which way the wind was blowing: they
short-term rate of non-transparent, off-invoice discounts by which
could not afford to stray off course into conflict too long, and they
their own performance is measured. In France - unlike the UK (Free,
also had to deal with their turbulent day-to-day business lives.
2007, 2008) - buyer-supplier negotiations focus not on price and
What made this novel cooperation initiative different from the
price reductions, but on these confidential, layered discounts that
participating organizations' past experiences was the involvement
determine the total margin ultimately passed on to retailers once
of senior management executives from both sides, the high sensi-
the sale to shoppers is made. Negotiators on both sides are used to
tivity of the information exchanged, and the concept of the product
seeing themselves as opponents engaged in a confrontational
category as the basic unit of analysis for their cooperation. These
process that requires toughness, a cult of secrecy, drama, and
specific features partly shaped the nature of the entirely new sit-
standard institutionalized calculative practices. It is all about
uation of inter-organizational cooperation. Yet they were only of
hammering out an agreement each party can live with. As a result,
minor importance compared to the managers' experiences of past
the buyers' cultural mindset and vested interests clashed with the
cooperative dealings, and the balance of interaction was often en-
new official position of a more cooperative, open category man-
dangered by extraneous matters, sometimes completely unrelated
agement approach, supported by the head of the merchandise
to the substantive discussions at hand.
department and the retailer's CEO. Yearly commercial negotiations
on the old model were still taking place in parallel to the new
category management process. It was not unusual for negotiators to 4.3. A major shift in French laws on retail price floors affects the
meet in the morning for an eyeball-to-eyeball confrontation, and buyer-retailer's business model
then participate in a category review in the afternoon. The
merchandise and buying departments were still two separate en- In France, the margin made by a retailer on the sale of a good has
tities of the retail company, and their respective leaders had not two components: a front margin and a back margin. The front
40 P. Lorino et al. / Accounting, Organizations and Society 58 (2017) 32e49
margin is the difference between the selling price to the shopper4
and the net invoiced price paid to the supplier to purchase the
good. The back margin is the amount of off-invoice, non-trans-
parent discounts negotiated through yearly rounds of negotiations,
expressed as a percentage of the net invoiced purchasing price.
These discounts are given to the retailer once the retail sale is made,
in the form of fees paid for “separate services” provided by the
retailer to the supplier (such as data sharing, promotions, listing
new products etc.). A major shift in legislation took place at the
beginning of 2006. Until that date, it was illegal in France to sell
goods at a price below the net invoiced price. The following chart
recaps how the retailer's margin was made under the pre-2006
laws (see Fig. 1).
The crucial implication of this schema was that the back margin
rate was ring-fenced for the retailer. It was forbidden by the law to
pass it on as retail price reductions. When this rate was negotiated
for an annual period, it applied to the amount of purchases. Un-
surprisingly, for years retailers had sought to maximize the nego-
tiated back margin, and this explains why the rate more than
doubled in the decade preceding our study and why commercial
negotiations had become so tense. Retailers also used to have a very
low front margin rate for shopper-sensitive products, to attract
customers. There was basically no price competition between re-
tailers on these products, as the net invoiced purchasing price set
by a supplier was the same for all retailers, and they all made a zero
front margin on them. Meanwhile, suppliers sought to contain rises
in the yearly back margin through general price increases on their
products that could not be refused by retailers and applied to all
retailers, before ultimately being passed on to shoppers.
In 2006, the French Ministry of Trade decided to change the
definition of the retail price floor. It was reduced to well below the
net invoiced purchasing price, such that the law now allowed re-
tailers to pass on part - and for some product categories, all - of the
back margin as a price reduction. Retailers could now differentiate
themselves from competitors through lower prices to shoppers, at
the cost of a lower overall commercial margin. It became legally
possible to have a negative front margin should a retailer choose to
price aggressively. The following chart recaps the effects of the new
2006 law on the retailer's margin (see Fig. 2).
Suppliers were now confronted with different pricing strategies
by different retailers, and also had to face retailers’ fury on finding
out that a competitor was selling a certain product at low prices
that they were unable to match.
4.4. Utterances of numbers are ultimately addressed to other
participants
In our case, new calculative practices saw the light of day under
the category management approach. For instance, inter-
organizational category scorecards now covered all aspects of the
business relationship including the stock-out rate at shelf level, or
the category market share indexes. This supported the narrative of
a “strategic partnership”, as these indicators extended the business
relationship beyond the narrow field of commercial negotiations
focusing solely on the back margin rate. The new “front margin”
indicator was a matter of particular concern during top-to-top
meetings. As part of an engineered accounting model, this indica-
tor made visible the institutional change that was dramatically
affecting the existing business model. It was no longer enough for
the buyer to negotiate an increase in the back margin rate, which
4
VAT is excluded from all the calculative practices presented here, which have
been simplified as there is no point in detailing the complexity of the layered
financial conditions.
had ceased to be ring-fenced. However, the trend displayed over
time by the front margin indicator worried the participants in
category reviews, as it moved from -3% to -8% for some categories,
suggesting significant intensification in the price war triggered by
the retailer in the case. Nicolas took great care to display these
numbers, without advance notification, at the beginning of each
top-to-top meeting. These numerical values disturbed the “here-
and-now” situation, since absorbing such a margin squeeze over
time was a matter of concern. The suppliers remained silent when
these numbers were being displayed, although they were skeptical
of the ongoing “strategic partnership” narrative, wondering “how
exactly are we category partners when such an aggressive pricing
decision is being made unilaterally”? There were also suspicions as
to the real motivation behind what was presented as a cooperative
approach: “is category management just a retailer's trick to pres-
surize suppliers into cleaning up the mess their unsustainable price
war has made”?
Through this presentation of numbers, Nicolas was in fact
making a veiled allusion to matters that could not be discussed
openly and directly, which illustrates the “existential” quality of
numbers, i.e. the way numbers are utilized to relate to participants
in a social situation (Vollmer, 2007). He knew that suppliers could
not openly discuss the retailer's pricing policy, as that would be an
infringement to fair competition practices and put them legally at
risk. However, they could not afford to be too detached from the
aggressive pricing tactic adopted by most French retailers. The
change in the back margin rate was computed to estimate how
much had been trimmed off the retailer's margin, and how far
profitability on the category had been dented. The suppliers were
well aware that a category generating little profit for a retailer is
under threat, since the retailer is likely to reallocate shelf space to
more profitable products. They therefore understood these
numbers as an indication of the urgent need for new action. Nicolas
was also indirectly and tactfully addressing the retailer's buyers in
the meeting room, presenting the numbers as an objective account
of the pricing policy decided by the retailer's Executive Committee.
By emphasizing the negative front margin, Nicolas was suggesting
that the retail company's profitability could no longer rely exclu-
sively on negotiation of off-invoice discounts. Displaying these
numbers implied that the world view revolving around the yearly
commercial negotiations might be nearing its end, along with the
powerful position of buyers who used to orchestrate inter-
organizational relationships.
Different futures thus collided in the present utterance of these
numbers. Would the suppliers accept them, consider Nicolas
trustworthy enough to justify discussions for a new world and go
down an unknown path toward a yet-to-be-invented business
model, even though they knew there would be unavoidable dif-
ferences of opinions to resolve? Or would they instead consider
Nicolas to be in a weak position in the retail company and expect a
backlash from the buyers, so that for the time being they would
simply pretend to be engaged in this discussion while using it as an
opportunity to glean information on the retailer's strategy? Would
they try to promote their own interests over those of the category
as a whole?
4.5. Framing and re-framing category reviews by involving
accounting numbers
We now turn to two episodes of local framing involving utili-
zations of managerial numbers. As said earlier, only a few examples
of frame-breaking and frame-shifting were witnessed, when the
strategic partnership discourse could no longer be credibly sus-
tained, as happened in the second episode presented below. Yet by
deviating from the agreed path and disrupting cooperative
 P. Lorino et al. / Accounting, Organizations and Society 58 (2017) 32e49
Old legal retail price floor
Back Margin > 0
Off-invoice, non-
transparent discounts
granted to retailers
Fig. 1. Breakdown of the retailer's total margin under the pre-2006 laws.
Back Margin > 0
New legal retail price Off-invoice, non-
floor transparent discounts
granted to retailers
Fig. 2. Breakdown of the retailer's total margin under the new law of 2006.
discussions, these incidents offer critical insights as they reveal the
actors’ skill at framing, resulting from their long history of shared
ups and downs and past dealings. The following table presents
some specifics of these two meetings: (see Table 3).
4.6. The “dance” in number-based talk to sustain the strategic
partnership project
The detergent supplier's team decided to present their latest
product launch, which was not initially on the meeting's agenda.
They wanted to take advantage of the presence of high-ranking
officials to push for full support from the retail company for a
launch at store level. They showed flattering graphs of the
impressive sales growth rate observed in the United States for
liquid detergents, making powder detergents look outdated. They
argued that the same trend was inevitable in France. Their
spokesperson presented a cost-savings estimation for the retailer
based on an expected long-lasting shift in consumer behavior. The
calculation was straightforward, since the estimated cost reduction
was founded on an objective technical fact: with the new compact
packaging format, liquid detergent was now one-third the previous
standard size for the same number of washes. This would “increase
41
Retailer’s actual selling
price to shoppers
Front Margin
>0
Net invoiced price paid to
suppliers to purchase the
goods sold
Net invoiced price paid to
suppliers to purchase the
goods sold
Front Margin < 0
Retailer’s actual selling
price to shoppers
the number of bottles per pallet by 150%, resulting in a 60%
reduction in pallets and cases transported”, and cutting the per-unit
transportation cost by 60%. Other benefits for the retailer's stores
were also claimed: the new bottle meant less shelf space was
needed, and all else being equal, this would enable the retailer to
increase the number of products on display and halve stock-outs.
The sales revenues per linear shelf meter would thus increase. On
top of all this, a new slide was displayed showing a prototype shelf-
ready packaging solution for the new product, which would entail
further productivity gains for the retailer, since 10 consumer units
could now be shelved in a single action instead of 10.
Nicolas' very lukewarm response to these good-looking
numbers was perhaps a disappointment for the supplier's team,
especially his comment that “all launches of concentrated laundry
detergents in France have always been a fiasco up to now”. The
retailer's team had mixed feelings about the numbers presented,
and hesitated between skepticism and giving the supplier the
benefit of the doubt. This particular situation made a key condition
of the category management approach visible: category manage-
ment should hybridize supplier expertise about future consumer
needs for a specific product category with retailer expertise in
current shopper behavior. It would therefore be tactless to criticize
42 P. Lorino et al. / Accounting, Organizations and Society 58 (2017) 32e49

Table 3
Some specifics of two category reviews.

Detergent category review meeting Cooked meat category review meeting

Date September 26th, 2006 October 11th, 2006

Duration
Participants
A standardized framework
of the category's KPIs and
an ongoing action plan
sent to all participants
a few days before
the meeting
Other information that
was added during
the meeting
2h 2h
The retailer's team The retailer's team
 Head of merchandise department  French CEO
 Senior manager of merchandise department  Head of merchandise department
in charge of Home and Personal Care products  Senior manager of merchandise department in
 Category manager charge of Fresh and Frozen goods.
 Head of buying department  Category manager
 Buyer dealing with this supplier  Head of buying department
 Supply chain manager in charge of Home  Buyer dealing with this supplier
and Personal Care products  Supply chain manager in charge of Fresh and Frozen goods
 Senior manager in charge of hypermarkets  Senior manager in charge of hypermarkets in the Paris area
in the Paris area  Category management project coordinator
 Category management project coordinator The supplier's team
The supplier's team  French CEO
 Head of supply chain operations  Head of supply chain operations
 Category manager  Another manager from logistics
 Head of supplier's negotiators  Category manager
 Supplier's negotiator with this retailer  Supplier's negotiator
 Head of sales force  Head of sales force
The researcher doing fieldwork The researcher doing fieldwork
 The category's financial objectives and results  The category's financial objectives and results
 Developments in the detergent market in France  Developments in the cooked meat market in
(per segment, and supplier's year-to-date for 2006) France (per segment, and supplier's year-to-date for 2006)
 Performance of this supplier in the French market  Performance of this supplier in the French market
with this specific retailer h this specific retailer
 Progress report on the ongoing category action plan  Progress report on the ongoing category action plan
 Vision and strategy for 2007  Vision and strategy for 2007
 25 slides  32 slides
 A presentation of the launch of a major product  A presentation of the retailer's negative assessment
innovation by the supplier seeking the retailer's of this supplier's logistics performance.
cooperation, with number-backed claims.  A promotional leaflet issued by one of the retailer's competitors

a top supplier for claims about future consumption trends in its
own line of business, especially when both teams have vowed to
deepen their cooperation and explore new business ideas together
with open minds. This explains why Nicolas, perhaps conscious of
the deep unease that his previous remark had triggered, immedi-
ately added: “we will do everything we can to support this launch
in our stores”. Everybody on the retailer's team was convinced that
it would be hard to turn concentrated liquid detergent into a major
market trend. A previous slide showed that concentrated powder
and liquid detergents only currently accounted for respectively 1%
and 3% of detergent sales.
However, the word “fiasco” could relate to two radically
different attitudes. Would the retailer side with the supplier's team
for the product launch, considering these numbers in a “construc-
tive”, future-driven way, as a valuable representation of a future at
hand or at least as a discussion-worthy representation of the future
that needed to be given more credibility through joint work? Or did
the word “fiasco” indicate that the retailer's team could simply not
accept these numbers, because their view was over-determined by
past negative experiences? Nicolas and most participants at this
meeting knew that ten years earlier, the same supplier had seen a
similar launch of a new concentrated powder detergent collapse
totally in one of the category's biggest failed launches ever. At the
time, the supplier's main direct competitors took every opportunity
to claim that in certain circumstances concentrated powder was so
efficient it actually made holes in garments. Nicolas' response
might well have been a veiled allusion to this debacle. If so, the
numbers presented by the supplier would presumably be received
by the retailer's team with polite indifference as pure wishful
thinking or, worse, as all smoke and mirrors. They would not even
seriously discuss them, since they did not believe them in the first
place, but they would respond to the supplier's team tactfully in
order to maintain the ongoing cooperative spirit.
The important things to observe following Nicolas' comments
were the gestures and looks between the supplier's team members
in the few seconds before a hasty mouse click brought up a new
slide full of new promises and numbers. This slide presented pro-
jected market shares for the different laundry detergent segments
between 1995 and 2015, with a focus on the expected high market
share for concentrated liquid detergents by 2015. The supplier's
team is in an uncomfortable hot-seat position, as its members could
not tell which way the wind was blowing: would both parties
eventually engage tacitly in a joint inquiry nurtured by discussions
of the numbers, or would they keep going down their separate
paths, each sticking to their position and belief, numbers against
numbers, but carrying on an unproductive dialog under the mask of
apparent cooperation?
What actually happened at the meeting was that this mishap
that could have caused serious trouble was rapidly ignored.
Everyone silently adopted a “let's-agree-to-disagree” stance to
preserve the cooperative mood, and engaged tactfully in repair
work. More specifically, there was a timely shift of focus toward
more constructive discussions revolving around the joint evalua-
tion of other business actions that were already under way (such as
cross-merchandising and shelf-ready packaging solutions). No one
mentioned or even alluded to the launch of the supplier's new
product again during this meeting.
4.7. Back to a good old arm-twisting relationship
This section concerns an episode when the plan to turn the
retailer-supplier relationship into a strategic partnership
completely collapsed. A slideshow presentation by the supplier
started as usual, with the main Key Performance Indicators for the
cooked meat product category. The speaker then made a comment
on the significant deterioration in the logistics service over the past
 P. Lorino et al. / Accounting, Organizations and Society 58 (2017) 32e49
weeks. She said that the problems experienced with production
and distribution of their promotional products were almost solved,
and that the performance level was back on track even though they
had fallen short of the initial target jointly set with the retailer's
team. She backed up her words with figures: the logistics service
level had risen from 96.7% for May to 98.1% for September 2006.
The supplier expected these numbers would demonstrate that the
situation was now under control and that the quality of service had
significantly improved between May and September 2006. The
retailer's team could have pretended nothing had happened,
thanked this supplier for their efforts to put things right, and
moved on to another topic of discussion. But this is not what
happened. A retailer official from the supply chain department
decided to bring in different numbers, a different time frame and a
different basis of comparison to discredit the supplier and show
that behind good-looking numbers, attempts had been made to
conceal poor service quality that might even trigger financial
penalties. He raised his hand and started contradicting the
speaker's “rather optimistic” statement. He wanted to stress “the
perception gap [he felt] about how the situation is currently
evolving”. He insisted that his own slides, which had not been part
of the initial slideshow sent out ahead of the meeting, be displayed.
With visuals including downward curves and angry emoticons, he
claimed that the service level was still poor and that the supplier's
team was still failing in its duties. His graphs showed that the
supplier had an average logistics service rate of 97.2% in 2006, down
by 0.5% from the previous year, whereas the whole category was at
98.5%. But the coup de gra ^ce was undoubtedly a photograph taken
at a logistics platform, showing the reception of a messy, unstable
pallet so heavily stuffed with products that some units were liter-
ally leaking from their cardboard cases. This photograph was
brought to the fore to illustrate the poor service level measured by
the “case fill rate” indicator. An angry red emoticon was displayed
next to the picture. The obvious iconic value of the photograph
echoed the function of numbers used as “truth tellers” and inten-
ded as an iconic representation of the logistics performance. They
reinforced one another in a snowball effect - the eloquent silence
that spread through the meeting room. All the numbers discussed
during this episode were trustworthy metrics, but they were taken
from two different performance measurement systems. However,
the selection of the time frame and the comparison basis, as well as
the way they were dramatized, signaled the involvement of
different and competing frames in the meeting. By choosing to
torpedo the cooperative mood shown by the supplier, the retailer
official brought everyone back to a more confrontational type of
relationship, placing the supplier on the back foot and making them
lose face, as they were unable to respond to this blow.
Unfortunately, misery loves company, and this painful episode
for the supplier's team was nothing compared to what came next.
The head of the buying department suddenly raised another issue
completely unrelated to the category management approach. He
aggressively brandished a promotional leaflet put out that week by
a competitor - a hard-discounter - featuring a premium product
from the same supplier at an “incredibly” low price. How could a
seemingly harmless, transient competitor's price tag cause havoc in
an inter-organizational meeting dedicated to exploration of novel
long-term cooperative practices? The fact was that the retailer's
buyers were not passively responding to this price tag as infor-
mation about a competitor's pricing policy for this product cate-
gory. The price tag was considered part of a more complex
language. It triggered quick mental calculations on the part of the
buyers, actively recombining it with other numbers as part of an
inquiry into their own performance. At this time, the recent
introduction of a new law in France allowed retailers for the first
time to pass on a limited part of the non-transparent, off-invoice
43
discount rates negotiated with a supplier to their customers
through a lower selling price. Buyers could therefore use compet-
itors' leaflets to indirectly compare their performance as negotia-
tors with competitors. In this particular case, the buyers applied the
existing contractual agreement with the concerned supplier to
determine the minimum price at which they could legally sell this
product. These calculations were done before the meeting and
never made explicit, but everybody understood what the head of
the buying department meant when he brandished the competi-
tor's leaflet like a smoking gun: the fact that his company could not
match this price tag implied that the supplier in the room was
giving better off-invoice rates to their competitor. The buyers
wanted revenge and were ready to apply financial penalties. In fact,
two possible futures collided in this episode. If the retailer team's
primary aim had been to build a more cooperative future, they
would have handled the competitor's leaflet differently, informing
the supplier's team before the meeting that this was a problem that
needed to be addressed - although to keep the cooperative spirit
alive, it would have been necessary not to mention this matter
during the meeting. Alternatively, if - as actually happened - they
wanted to place the supplier in an awkward position, then the
decision to dramatize the leaflet's price tag during the meeting was
a logical move. By choosing to utilize this number at this point in
the meeting, the head of the buying department not only redefined
the ongoing situation; he also put an end to any immediate pros-
pects of jointly building a partnership.
5. Discussion
5.1. The dual face of accounting numbers: the generic model and
situated numerical value
Accounting numbers are continuously and simultaneously used
to tell the past story of the situation and anticipate its potential
futures. But the notion of “accounting numbers” actually points to
two distinct epistemological and ontological classes of objects. On
one hand, accounting numbers are models which have some iconic
- mimetic - type of relationship with the situation. They are engi-
neered by experts and are supposed to represent collective activity
and situations, not necessarily in true and accurate ways, but in
ways that prove meaningful for collective activity. In our case study,
the costing model used to calculate the total margin for the re-
tailer's product category is supposed to give a meaningful account
of the category's economic behavior. While the accounting model is
not supposed to guarantee scientific accuracy, it should nonethe-
less provide a sufficient level of social intelligibility and credibility
to be accepted by actors and serve as an instrument for commu-
nication and coordination. On the other hand, when the head of the
buying department brandishes a competitor's promotional leaflet
featuring a premium product at an “incredibly” low price, the ac-
counting number is presented not as a model, but as a specific
numerical value, with meaning rooted in a precise place at a precise
moment, an event in the situation. It is interpreted by everybody as
an exceptional discount granted by the supplier to the retailer's
competitor in recent weeks. But to reach such a conclusion, actors
must use the number as a clue and follow a reasoning process
involving calculation, however “quick and dirty”. Specific numeri-
cal values in given situations are singular indexical, and sometimes
surprising, signs that trigger an inquiry to understand what they
actually mean. The iconic model is based on an ex ante under-
standing of the situation: its meaning lies in its form; the indexical
numerical value leads to an ex post understanding of the specific
situation that generated it, through an inquiry. While the ac-
counting model participates in a narrative frame, the specific nu-
merical value appears as an event and is a key element of the here-
44 P. Lorino et al. / Accounting, Organizations and Society 58 (2017) 32e49
and-now situation: the accounting model “utters” the situated
number. The two natures are connected: as a model, accounting
numbers are engineered and are part of frames; as situated events,
accounting numbers occur, with the potential to surprise actors,
and are singular and situated. This is the very definition of semiotic
mediation: the mediating sign is simultaneously the expression of a
socially-constructed class of meaning, and a component of the
here-and-now situation. Accounting numbers are thus linking pins:
they link the singular situation - this precise numerical value, ex-
pected or surprising - with socially-constructed classes of meaning
- the accounting model.
5.2. Frames as a concrete and holistic approach to practice
Situations play a major role in the specific meaning of ac-
counting numbers for managers. Frames are not frozen stories: they
are interpretive resources submitted to experience and involved in
the “flow” of meaning-making, with an ever-extending narrated
past and an ever-nearer imagined future (Ricœur, 1990). Frames,
with their accounting components, are narrative patterns, main-
tained or modified by actors' permanent situated framing action.
Vollmer (2007) has drawn on Goffman's theory of frames to discuss
the dynamic uses of numbers viewed as numerical signs in social
situations. He argues that framing selectively emphasizes certain
qualities of numbers while others remain tacit or latent, or dis-
tributes roles to actors linked with the distinct qualities of numbers.
He identifies the framing dynamic with the joint involvement of
two abstract frames - the “reproductive frame” and the
“consumptive frame”. His perspective is therefore anchored in an
abstract view of frames that is not grounded in the study of situated
activity (Strong, 1983), and departs from Goffman's naturalistic
methodological approach. Vollmer's reading of Goffman's frames
tends to analytically, rather than empirically, separate the steps of
design (“reproductive”) and use (“consumptive”). He stresses the
importance of up- and downkeying moves, which dynamically
transform accounting models from models for use into models for
redesign, and vice versa. Keying towards a reproductive utilization
of numbers is viewed as “upkeying”; keying towards a consumptive
utilization of numbers is viewed as “downkeying”, i.e. an attempt
by participants to establish and negotiate the meaning of a set of
numbers: “the basic difference between upkeying and downkeying
the use of numbers therefore is not that upkeying is always stable
while downkeying is fundamentally insecure, but that one process
focuses the utilization of numbers on establishing their meaning,
the other on producing arithmetic results” (Vollmer, 2007, p. 588).
Admittedly this should not be viewed as a temporal sequence, but
even so, from a strictly analytical perspective, the reproductive/
consumptive distinction is problematic. In our case, the consump-
tive and reproductive dimensions cannot be separated in the
practical utilization of numbers. To make meaning, the number is
related to other numbers, and compared and/or combined with
them. Thus it “reproductively” generates new numerical signs. But
this reproductive treatment only takes place to “consumptively”
ascribe a situated meaning with regard to a referent object in the
situation, for example a product or an activity. In typical organi-
zational situations involving accounting uses, such as budget con-
trol or new product pricing, consumptive utilizations need to be
reproductive to provide meaningful clues, and reproductive utili-
zations are guided by consumptive logics. The object of study is
neither the reproductive nor the consumptive move, but the rela-
tionship between the two dimensions, and the elementary unit of
analysis should therefore be relational. For instance, when the price
tag of a product in a competitor's leaflet is brought to the fore by the
retailer's buyer, it is “reproductively” computed through tacit
costing/pricing formulas - but this reproductive calculation takes
place intrinsically as a consumptive demonstration of the supplier's
unfair practice. The reproductive and the consumptive are two
general dimensions of any practical utilization of numbers, rather
than specific meaning-making social frames which, in Goffman's
perspective, should correspond to typical social situations, and the
relationship between these two dimensions of the same action is a
key issue.
5.3. Two competing frames in situated meaning-making
As a semiotic mediation, a frame broadens the definition of a
situation by instantiating elements that are apparently absent from
the situation but will orient meaning-making in the as yet unde-
termined course of these business relationships. Meetings are
haunted by the past of the relationship, its projection into a desired
but uncertain future, and the explicit or implicit incorporation of
distant actors and places into the present situation of action and
interaction. When meeting face-to-face, senior managers bring
with them not only their own bodies and reputations, but also a
personal history of prior dealings with the other participants and a
socialized experience based on their professional skills, organiza-
tional attachments and strategic images of the future. They are also
involved in a nexus of interactions that is constantly being recon-
figured according to their purposes, interests and intents. For
example, CEOs and key negotiators from suppliers and retailers had
lobbied the Ministry of Trade - in very different ways - to influence
the new legal framework. Some participants meet at ECR5 to work
together on setting standards for the industry. However, the frame
at work in a meeting, which reflects this broader social experience,
does not determine the way things will actually turn out. The
course of the discussion is ultimately unpredictable, since unex-
pected social and material circumstances disrupt it, and nothing
should be taken for granted. Different temporalities and issues
sometimes run parallel, sometimes briefly surface and overlap, and
sometimes clash during meetings.
In our case, managerial meaning-making oscillated between
different frames. What we seek to understand is how these man-
agers, by their doings and physical commingling, engage in
ordering the unfolding interaction while at the same time experi-
encing the variability of social situations, the flexibility of ar-
rangements and the crumbling foundations of their new “category
management” mantra. In the course of the meetings, there was a
permanent competition between two opposing frames: the “stra-
tegic partnership” frame and the “adversarial commercial negoti-
ation” frame. These two frames have been specified empirically
through the emergence of narrative regularities in the actors’ direct
experience across sixteen buyer-retailer relationships. The
following table specifies their respective characteristics (see
Table 4).
The “adversarial commercial negotiation” frame and the “stra-
tegic partnership” frame are fundamentally different in status. The
“adversarial” frame is well-known, normalized, stabilized, ritual-
ized, made up of habits and institutionalized calculative practices
and models, whereas the “strategic partnership” frame is still
partially indeterminate and uncertain, and explores potential joint
actions. The meaning of accounting numbers involved in this frame
is not yet stabilized. The “strategic partnership” frame refers to a
new type of narrative, as yet untold, inspired and revitalized
through the connections to the rest of the world that participants
creatively contribute to the narrating situation. This connectedness
is also a threat, since the narrating situation's shift to the
5
Efficient Customer Response (ECR) is a joint trade and industry body promoting
more responsiveness to consumer demand in the retail sector.
 P. Lorino et al. / Accounting, Organizations and Society 58 (2017) 32e49
“adversarial commercial negotiation” frame is made easier by the
fact that old habits die hard. Special interests may prevail insidi-
ously. When the framing of the “strategic partnership” narrative is
disrupted or collapses, “[participants] have done more than rede-
fined a situation; [they] have destroyed a world” (Gonos, 1977, p.
861). What is lost in such circumstances is the jointly-endorsed
hope of a world where novel types of action could be co-devised,
debated and tested.
Another major difference between the two frames lies in the
roles and different temporalities of numbers, which at times clash.
The selection of numbers to be utilized in social situations is part of
local framing processes, and through the practical effects the
numbers are supposed to perform or actually perform, they help to
sustain/challenge a narrative frame. In the “strategic partnership”
frame, numbers are mainly viewed as clues triggering and
nurturing inquiries into as yet unknown innovative joint action.
Numbers are viewed here primarily through their “indexical”
properties, as clues to something, traces of some object still to
discover, just as suspicious footprints near the door to a crime scene
can trigger an identification search. In most cases, numbers do not
convey ready-made meanings. They signal “here lies something”
that is not clearly understood, for example the commercial pros-
pects for liquid detergent or a deterioration in the front margin. The
numbers displayed call for dialogical interactions and a joint in-
quiry to construct a new hypothesis about what is happening and
hopefully devise novel joint business practices. What really matters
for the managers is keeping the momentum towards novelty: like
gold-diggers searching for a profitable new seam, they are not
really interested in what is or what should be, but rather in what
may be (Follett, 1924; Tsoukas, 2009; Tsoukas & Chia, 2002).
The roles played by numbers in the “adversarial commercial
negotiation” frame are quite different. In this case, numbers are
often viewed as objective, accurate representations of their un-
derlying objects, be it the calculation of a “supplier's sales revenue”
or a “net invoice price”. They also rely on social conventions that
standardize their production and normalize their interpretation. In
this perspective, the compliance of numbers with institutionally
established norms is as important as their representational accu-
racy, because they ensure that each player respects the rules of the
game. Even when numbers are used as a clue, a trace of some
invisible fact, their meaning is considered obvious and does not
justify a joint inquiry. They are used to put the other negotiators in
difficulty and weaken their position: buyers and suppliers pit their
numbers against each other, in monological accounts of their own
side of the story. This is best illustrated by the head of the buying
department interpreting the price tag for a supplier's product in a
competitor's leaflet as justification for confrontational showdowns
in negotiation cubicles.
5.4. Up- and downkeying moves: the reflexive adaptation of an
existent frame
As already noted, Vollmer stresses the importance of up- and
downkeying moves, which transform accounting models from a
model for (re)design into episodes of consumptive uses of numbers.
In that sense, framing processes can be described as a succession of
up- and downkeying moves, in which the uniqueness of the frame
is not questioned, but reflexive steps constantly add or remove
layers of meaning. Similarly, Callon's connection of the comple-
mentary concepts of ‘framing’ and ‘overflowing’ (Callon, 1998, p.
259) resembles a kind of downkeying process as described by
Vollmer: “Once the overflows, source agents and target agents have
all been correctly identified and described, and once measuring
instruments for quantifying and comparing them have been set up”
(this is close to a reproductive action), “it becomes possible to
45
reframe the interactions” (this second step is close to a consump-
tive action). The logical “bounding/overflowing” dyad brings a
fragile, yet coherent and sustained meaning perspective. It does not
consider the potential plurality of competing ‘frames’, but rather
the process of continually adapting an existent frame. When
overflowing occurs, the uniqueness of the frame is not questioned,
but its boundaries are destabilized, and now require reflexive
critical reappraisal and redesign. In both Vollmer's and Callon's
approaches, the focus of the up- and downkeying moves is not on
the plurality of competing frames at a given moment, but rather on
the diachronic transformation of a given frame. Goffman
acknowledged this shortcoming in the first version of his theory of
frames: “there is insufficient attention paid to the shifting from one
of these frames to another” (Goffman, 1981b, p. 67).
5.5. Frame plurality and the horizontal move of frame-shifting
In our case study, frame plurality and frame-shifting are seen to
be a key dimension of managerial meaning-making, particularly in
situations of radical organizational change in which there are at
least two frames, one extant and one emergent. There might be a
dominant frame in the situation - in the case study, the “strategic
partnership” frame - but it does not eliminate other frames: pre-
vious frames that still remain as latent options in the background,
or new frames that are sidelined for the time being, but could re-
turn to the forefront in the right circumstances. Meaning-making
can then take the form of a shift from one frame to another:
events mediated by specific signs can lead to the sudden replace-
ment of one frame with another, entailing a major change in
practices. For instance, when Nicolas presents the negative trend in
the front margin rate in his preliminary speech at the beginning of
every category review, he instantiates the “strategic partnership”
frame and brushes aside the “adversarial frame” with its focus on
the back margin, which is best not discussed at these cooperative
meetings. Or, when the head of the buying department brandishes
a competitor's leaflet vaunting a low price for a specific supplier's
product, he is bringing the “adversarial commercial negotiation”
frame back into play. These are shifts from one frame to another,
not keying moves, as they do not involve reflexive adaptation of a
frame, that is, keeping its basic form but completely questioning
the prevailing primary frame in order to promote an alternative. An
example of keying is when actors adapt the definition of category
management by exploring cooperation areas that had not been
originally identified.
This pluralistic, shifting view has been developed by Vinnari and
Skærbæk (2014) as a key complement to Callon's concept of
“overflowing”, and by Yang and Modell (2015), who draw on the
social movement literature about framing and frames to analyze a
situation in which “an emerging, shareholder-focused frame
interacted with the extant work unit frame” (Yang & Modell, p.1).
While in their view, all-encompassing ideologies (Maoism or
Confucianism) pre-formed the frames involved in the managerial
use of accounting, in our case study, the two frames involved in the
negotiations do not draw their legitimacy and general orientations
from external overhanging ideologies. Goffman emphasized the
autonomy of frames mobilized and re-engineered in situated in-
teractions: (for Goffman) “the interaction order constitutes a field
of inquiry in its own right: it is not the support or the product of
functions that would be external to it. It emerges within its proper
laws”; “the interaction order does not obey the laws which rule the
social system, but it has its own self-maintenance, self-configura-
tion and self-distantiation rules” (Cefaï & Gardella, 2012, p. 241, our
translation). In our semiotic view of Goffman's theory, frame-
shifting is a horizontal move, from frame to frame, rather than
the vertical deployment of macro-social orientations.
46 P. Lorino et al. / Accounting, Organizations and Society 58 (2017) 32e49

Table 4
The social-temporal horizons of two competing frames in situated meaning-making.

Adversarial commercial negotiation frame Strategic partnership frame

Type of narrative
Unit of analysis of inter-
organizational relationships
Key accounting models,
frameworks and indicators
Legal context of inter-
organizational
calculative practices
Key actors
Time frame
Modalities of inter-
organizational meetings
Modes of conflict resolution
Values
Dominant role of
accounting numbers
Non-transparent, adversarial commercial
negotiations of the back margin
Supplier's contract
Back margin rate
Contractual and legally binding
Buyers and supplier's negotiators (opponents)
Yearly rounds of negotiations, short-term view
Confrontation, drama, tit-for-tat measures,
tight lips, guilty till proven innocent,
pressure tactics
Domination, compromise (zero-sum game)
Toughness, fighting spirit, no holds barred
Numbers are expected to be objective,
relevant representations of situations
Category management: cooperative construction of a joint
category strategy embracing all aspects of their relationship
Product category
Front margin rate and back margin rate, total amount of margin
Joint category scorecards
Joint category budgets
Joint category strategic plan
Key performance indicators: Category sales revenue,
category profit, market shares; shelf availability rates,
stock level
Non-legally binding
Corporate leaders supported by all actors involved
in the management of the relationship (partners)
Medium-term (2e3 years)
Discussion, experimentation, no taboos, risk-taking,
joint business planning
Search for a creative “third way” (positive sum game)
through dialogical exchanges
Openness, non-judgmental attitude, deep involvement, tact
Numbers are expected to provide clues to trigger
and nurture inter-organizational inquiries

Frame plurality echoes the notion of “dissonance” developed by
Stark (2009) in the sociology of conventions. Each frame corre-
sponds to a certain way of evaluating value(s) and thus to a certain
meaning of numbers and a certain “order of worth”. Shifting is the
movement from one way of evaluating situations to another: “or-
ders of worth are the very fabric of calculation, of rationality, of
value” (Stark, 2009, p. 11) and there is “uncertainty about which
order is operative in a given situation” (Stark, 2009, p. 15), leading
to an “open-ended, unpredictable conversation” (Stark, 2009, p. 3).
The “frictions” between “actors with different evaluative princi-
ples” then become a major issue (Chenhall et al., 2013, p. 283). The
plurality of competing frames and the participants’ ongoing con-
nections to the external world are the main causes of vulnerability
in complex situations of negotiation, when managers try to orient
organizational change and different possible futures may collide.
5.6. Sustaining the strategic partnership frame requires
dramaturgical cooperation
The participants in category review meetings must consider the
interests of their respective companies and departments, which are
partially divergent. Detractors, skeptics and supporters of the
category management approach are found in all organizational
teams. The slightest thing could threaten the category management
mantra. As a result, engaging a category review aligned with the
narrative of trust, collaboration and value creation in the name of
the customer requires considerable managerial effort to frame and
maintain expectations.
The framing, keying and shifting processes can be viewed as
strategic attempts by participants to manipulate the stakes of the
ongoing social situation. Under the surface of an apparent socia-
bility and conformance to the ongoing narrative, participants may
in fact be turning a falsely cordial manner to their own advantage,
aiming to engage a re-framing of the category review or to discredit
a number-based claim while remaining above suspicion if their
maneuver backfires. In Goffman's view, a frame is associated with
normative expectations related to the participants' social conduct
in a social situation. Such expectations differ significantly in the
two framing episodes described earlier. One major difference lies in
the necessity, when operating in the “strategic partnership”
narrative frame, of saving the face of the other participants, and
employing dramaturgical cooperation practices to that end
(Goffman, 1959, 1967; Patriotta & Spedale, 2009) - this is certainly
not a necessity in the “adversarial commercial negotiation” frame.
These “face-work” practices permeate managers' uses of numbers,
as no one really draws attention to facts that would affect the
cooperative mood. For example, suppliers pay lip service to Nicolas'
claim that “a true profit” per category has been calculated and
presented for once. This is a huge exaggeration to say the least, as
many categories and supplier-specific supply chain costs could not
be factored in to measure a category bottom line. Suppliers would
never be so accepting of weak numbers if the “adversarial com-
mercial negotiation” frame were dominant: such estimates could
have both financial and contractual consequences in the short term,
and they would certainly not miss the chance to pounce on this
weakness and put the buyer on the defensive.
Our case study thus offers a concrete illustration of how Goff-
man's theory of framing and his earlier work on impression man-
agement may intertwine when numbers are selected and
dramatized at specific moments in social situations. As already
suggested by Solomon et al (2013), “framing can be seen as
selecting aspects of a ‘perceived reality’ and making them more
salient in a communicating text (Entman, 1993)” (Solomon et al.,
2013, p. 198). More specifically, organizational actors dramatize
specific signs - accounting numbers, smileys, photographs, ges-
tures, silences, etc. - and roll them together into simultaneous
chains of interpretation in the very moment of their utilization, to
work as “narrative cues”. These deliberate expressive effects -
which may succeed, fail or have unintended consequences - explain
why numbers acquire multiple meanings beyond their mere cal-
culative and symptomatic qualities. Accounting numbers can be
viewed as part of an array of dramaturgical accessories and tricks
that are skillfully taken up and combined by a collective of actors, to
put on a show that in the end will be accepted - or rejected - by the
other participants. The dynamics of interactions are played out not
only between minds or organizational actors, but between frames:
actors are “tenants” of the places they occupy within their own
professional social system or, rather, within the parliament of roles
 P. Lorino et al. / Accounting, Organizations and Society 58 (2017) 32e49
and attitudes they are normatively expected to play and display.
These normative social constraints confine their free will and limit
what they can do, say and allow others to do.
Therefore, numbers act in these social situations, but their
performativity is not a given: it has to be accomplished jointly by
participants, through practices that require skill. Further, our case
suggests that any “felicitous”6 (Goffman, 1983) utilization of
numbers requires their ongoing acceptance by the other people
present. In our case, almost every number-based claim made under
the “strategic partnership” frame could easily have been challenged
or discredited (such as the retailer's measure of a category profit).
Dramaturgical cooperation between and across teams was needed
to sustain the impression that the numbers presented were what
the speaker hoped the members of the audience would take them
to be: sound, objective, indisputable proof backing his claim. And
indeed most of the numbers were tactfully accepted. Even when a
number-based claim was challenged - for example in the new
detergent episode - numbers were drawn into the present situation
not for their calculative or symptomatic qualities, though everyone
acted as if that were the case, but because it was a socially appro-
priate way to question the supplier while saving the speaker's face.
Indeed, in the “strategic partnership” frame, the managers pri-
marily utilized numbers to perform in the interest of order
(Manning, 2008) and could let major omissions pass in order to
keep up appearances.
Framing is therefore a vulnerable process that requires mana-
gerial efforts and ongoing acceptance by participants. These
meeting situations unfold in a “drifting” process, full of twists and
turns, punctuated by the unforeseeable expressive effects of
dramaturgical tricks involving numbers - as “the box that conver-
sation stuffs us into is Pandora's” (Goffman, 1981a, p. 74).
6. Conclusion
In this paper, we suggested that Goffman's theory of frames
could provide a sound basis for analyzing the role of accounting
numbers in meaning-making processes, as long as it is interpreted
as a framework that is processual and semiotic (based on the
concept of mediation). Focusing on the framing/keying/shifting
process, we analyzed a specific type of situation, an inter-
organizational negotiation in the retail sector, characterized by its
exploratory and uncertain future direction. It remains to be
explored whether this conceptual framework would be applicable
to “colder” situations, characterized by more extensive use of
routines and more habitual types of activity. Nonetheless, our case
study brings to the fore the following contributions.
Exploring a doubtful situation to make meaning does not
involve the application of given, stable frames, but rather a process
of framing. It simultaneously requires social re-engineering of
frames, as generic systems of meanings (“framing” as making
frames), and the involvement of those frames in situated action
(“framing” as giving form to a situation). Frames are modified in the
course of their practical involvement. When using accounting
6
Goffman expands Austin's view (Austin, 1967) on the “Felicity's Conditions”
generally studied in conversation, i.e. verbal communication, as “a question of who
can say what to whom, in what circumstances, with what preamble, in what surface
form, and, given available readings, will not be thought mindless in doing so. A
question of what we can say and still satisfies Felicity's Condition” (Goffman, 1983,
p. 48). Goffman further expands the Felicity's Condition to non-verbal acts: “Nor (it
should now be plain) was it right to define Felicity's Condition restrictively in terms
of verbal acts. Speech need not figure even in a reduced way for Felicity's Condition
to apply: the general constraint that an utterance must satisfy, namely, that it
connect acceptably with what recipient has in, or can bring to, mind, applies in a
manner to non-linguistic acts in wordless contexts“ (Goffman, 1983, p. 50).
47
numbers, actors are constantly inquiring into the situated numer-
ical value and the model simultaneously. If the numerical value is
surprising, does that reveal an unsuspected aspect of situated ac-
tivity, or an unsuspected aspect of the accounting model? Should it
lead to a change in the course of action, or the model, or both? The
meaning-making process can be simultaneously defined as the
process of framing the situation and the process of situating the
frames. In a continuous recursion, frames enable framing, framing
enables adaptation of frames or invention of new frames, and so
forth.
We also observed in our case study that the meaning-making/
framing process is an ongoing conversational development,
driven by the ever-evolving diversity of views and ways of
narrating the situation. There might be a prevailing frame - such as
the “strategic partnership” frame in the case study - but it does not
eliminate other frames, that are virtualities ready to turn into actual
mediations if new situations disrupt the present order. The
complexity of situations in fact results precisely from the plurality
of potential frames. Our empirical findings illustrate the plurality of
competing frames and the importance of shifting episodes, when
one frame is suddenly replaced with another, leading to major
transformations in the situation. The (re-)framing processes thus
appear as both adaptive and reconstructive through keying and
shifting moves.
Who/what are the direct participants in the conversational
framing process? At first glance it could appear to be individual
subjects - in the case study, the managers participating in the
meetings, or alternatively it could be organizations, characterized
as collective subjects - in the case study, the firms involved in the
strategic partnership discussion. In concordance with the proces-
sual and semiotic approach adopted here, the dialog does not
involve organizations as such: the case shows that defenders of the
two contending frames could be found on both sides of the inter-
organizational meeting. For example, within the retail organiza-
tion, buyers and strategic leaders tend to have divergent views of
situations, and are involved in a debate internal to their firm. Also,
the dialog does not primarily involve individual subjects; we
observed that executives, for example the head of the retail com-
pany's trade marketing department, Nicolas, can shift their narra-
tive meaning-making frame in the course of a meeting, or hesitate
between different frames in a single specific episode. The bound-
aries between alternative narrative views extend across organiza-
tions and even subjective minds. The dialogical process, with its
keying and shifting moves, takes place primarily between frames. It
is a form of inter-textual, inter-narrative rather than inter-
subjective or inter-organizational exchange. The narrative nature
of sociality was emphasized by Goffman: “For Goffman, even the
individual is a ‘team’, because it consists of a performer and an
audience (although united in one person) and the unity of society is
simply fictional” (Riggins, 1990, p. 3). The organizing dialog is an
“inter-frame” process, which suggests new avenues for future
research, focused on the framing and inter-framing dynamics.
Ultimately, our case study paves the way for seeing the per-
formativity of numbers as a skilled performance within the con-
fines of social situations: a narrative view of performativity.
Calculative devices delineate virtualities, meaning that they
establish boundaries between meaningful and meaningless clues,
feasible and unfeasible action, and thus shut off certain paths. Yet
they also open up new potential for action, by instantiating in the
situation meanings and possibilities based on collective, cultural
and historical experience, beyond the subjective experience of ac-
tors directly involved in the action in progress. The actors must thus
turn some of the infinite possibilities conveyed by the device into
concrete action. This actualization is a skilled organizational prac-
tice - and even sometimes an “ ‘artful’ practice” (Jo € nsson, 1998, p.
48 P. Lorino et al. / Accounting, Organizations and Society 58 (2017) 32e49
432) - through which the participants in a social situation develop a
practical understanding of each other, including their allusions and
what they really mean. Admittedly, without the calculative device,
such virtualities would not exist, but their actualization is a
managerial effort, socially informed by a frame and by collective
experience and skills. This sheds light on the symbolic significance
of any dramatized utilization of numbers in a social situation that
should be viewed as performing a “number act” - comparable to a
“speech act” - in a “regulatory drama” (Vollmer, 2007). Performing
a “number act” requires consideration of the other participants'
capacity to read the “evidence for our feelings, thoughts and in-
tents” (Goffman, 1983, p. 51) in the selection and dramatization of
numbers. Yet “it also allows us to bring to bear all of the world to
which the other can catch allusion” (Goffman, 1983, p. 51). The
effects performed by numbers are thus not predetermined, and
depend on the situation and the actors' contributions, in particular
managers' trial-and-error explorations.
References
Ahrens, T., & Chapman, C. S. (2006). ing qualitative field research in management
accounting: Positioning data to contribute to theory. Accounting, Organizations
and Society, 31(8), 819e841.
Ahrens, T., & Chapman, C. S. (2007). Management accounting as practice. Ac-
counting, Organizations and Society, 32(1e2), 1e27.
Ahrens, T., & Dent, J. F. (1998). Accounting and Organizations: Realizing the richness
of field research. Journal of Management Accounting Research, 10, 1e39.
Andon, P., Baxter, J., & Chua, W. F. (2007). Accounting change as relational drifting: A
field study of experiments with performance measurement. Management Ac-
counting Research, 18(2), 273e308.
Atkinson, P. (1988). Ethnomethodology: A critical review. Annual Review of Sociol-
ogy, 14, 441e465.
Austin, J. L. (1967). How to do things with words. Cambridge, Mass: Harvard Uni-
versity Press.
Baxter, J., & Chua, W. F. (2003). Alternative management accounting research -
whence and whither. Accounting, Organizations and Society, 28(2/3), 97e126.
Bendford, R. D., & Snow, D. A. (2000). Framing processes and social movements: An
overview and assessment. Annual Review of Sociology, 26, 611e639.
Boland, R. J. (1989). Beyond the objectivist and the Subjectivist: Learning to read
accounting as text. Accounting, Organizations and Society, 14(5/6), 591e604.
Boland, R. J. (1993). Accounting and the interpretive act. Accounting, Organizations
and Society, 18(2/3), 125e146.
Boland, R. J., & Pondy, L. R. (1986). The micro dynamics of a budget-cutting process:
Modes, models and structure. Accounting, Organizations and Society, 11(4/5),
403e422.
Briers, M., & Chua, W. F. (2001). The role of actor-networks and boundary objects in
management accounting change: A field study of an implementation of
activity-based costing. Accounting, Organizations and Society, 26, 237e269.
Callon, M. (1998). An essay on framing and overflowing: Economic externalities
revisited by sociology. In M. Callon (Ed.), The laws of the markets (pp. 244e269).
Oxford: Blackwell.
Cefaï, D., & Gardella, E. (2012). Comment analyser une situation selon le dernier
Goffman ? De frame analysis a  forms of talk. In D. Cefaï, & L. Perreau (Eds.),
Erving Goffman et l’ordre de l’interaction (pp. 233e266). Paris: CURAPP-ESS/
CEMS-IMM.
Charie
, J. P. (2000). Rapport d’information de l’Assembl
ee Nationale portant sur
l’

evolution de la distribution enregistr
e le 11 janvier 2000.
Chenhall, R. H., Hall, M., & Smith, D. (2013). Performance measurement, modes of
evaluation and the development of compromising accounts. Accounting, Orga-
nizations and Society, 38, 268e287.
Chua, W. F. (1986). Theoretical constructions of and by the real. Accounting, Orga-
nizations and Society, 11(6), 583e598.
Cooper, D. J., & Ezzamel, M. (2013). Globalization discourses and performance
measurement systems in a multinational firm. Accounting, Organizations and
Society, 38, 288e313.
Cornelissen, J. P., & Werner, M. D. (2014). Putting framing in perspective: A review of
framing and frame analysis across the management and organizational litera-
ture. The Academy of Management Annals, 8(1), 181e235.
Czarniawska, B. (1998). A narrative approach to organization studies. Sage
Publications.
Czarniawska, B. (2006). A golden braid: Allport, Goffman, Weick. Organizations
Studies, 27(11), 1661e1674.
Dechow, N., & Mouritsen, J. (2005). Enterprise resource planning systems, man-
agement control and the quest for integration. Accounting, Organizations and
Society, 30, 691e733.
Entman, R. M. (1993). Framing: Toward clarification of a fractured paradigm. Journal
of Communication, 43(4), 51e58.
Favret-Saada, J. (1977). Les mots, la mort, les sorts. Paris: Gallimard.
Follett, M. P. (1924). Creative experience. New York: Longmans Green.
Frances, J., & Garnsey, E. (1996). Supermarkets and suppliers in the United
Kingdom: System integration, information and control. Accounting, Organiza-
tions and Society, 21(6), 591e610.
Free, C. (2007). Supply chain accounting practices in the UK retail sector: Enabling
or coercing collaboration. Contemporary Accounting Research, 24(3), 897e933.
Free, C. (2008). Walking the talk? Supply chain accounting and trust among UK
supermarkets and suppliers. Accounting, Organizations and Society, 33(6),
629e662.
Gallhofer, S., Haslam, J., & Yonekura, A. (2015). Accounting as differentiated uni-
versal for emancipatory praxis. Accounting, Auditing & Accountability Journal,
8(5), 846e874.
Gerdin, J., Messner, M., & Mouritsen, J. (2014). On the significance of accounting for
managerial work. Scandinavian Journal of Management, 30, 389e394.
Gitlin, T. (1980). The whole world is Watching: Mass media in the making & unmaking
of the new left. Bekerley: University of California Press.
Goffman, E. (1959). The presentation of self in everyday life. Anchor Books.
Goffman, E. (1963). Stigma: Notes on the management of spoiled identities. New York:
Simon & Schuster Inc.
Goffman, E. (1964). The neglected situation. American Anthropologist, 66(6),
133e136.
Goffman, E. (1967). Interaction ritual: Essays in face-to-face behavior. Chicago: Aldine.
Goffman, E. (1974). Frame analysis. An essay on the organization of experience. Bos-
ton: Northeastern University Press.
Goffman, E. (1981a). Forms of talk. Philadelphia: University of Pennsylvania Press.
Goffman, E. (1981b). A reply to Denzin and Keller. Contemporary Sociology, 10(1),
60e68.
Goffman, E. (1983). Felicity's condition. The American Journal of Sociology, 89(1),
1e53.
Goffman, E. (1989). On fieldwork. Journal of Contemporary Ethnography, 18(2),
123e132.
Gonos, G. (1977). “Situation” versus “frame”: The “interactionist” and the “struc-
turalist” analyses of everyday life. American Sociological Review, 42(6), 854e867.
Graham, C. (2008). Fearful asymmetry: The consumption of accounting signs in the
Algoma Steel pension bailout. Accounting, Organizations and Society, 33(7/8),
756e782.
Hacking, I. (2004). Between Michel Foucault and Erving Goffman: Between
discourse in the abstract and face-to-face interaction. Economy and Society,
33(3), 277e302.
Hall, M. (2010). Accounting information and managerial work. Accounting, Organi-
zations and Society, 35(3), 301e315.
Hingley, M. K. (2005). Power imbalanced relationships: Cases from UK fresh food
supply. International Journal of Retail & Distribution Management, 33(8),
551e569.
Ho, J. L. Y., Wu, A., & Wu, S. Y. C. (2014). Performance measures, consensus on
strategy implementation, and performance: Evidence from the operational-
level of organizations. Accounting, Organizations and Society, 39, 38e58.
€nsson, S. (1998). Relate management accounting research to managerial work!
Jo
Accounting, Organizations and Society, 23(4), 411e434.
Justesen, L., & Mouristen, J. (2011). Effects of actor-network theory in accounting
research. Accounting, Auditing & Accountability Journal, 24(2), 161e193.
Kastberg, G. (2014). Framing shared services: Accounting, control and overflows.
Critical Perspectives on Accounting, 25(8), 743e756.
Lorino, P., & Mourey, D. (2013). The experience of time in the inter-organizing in-
quiry: A present thickened by dialog and situation. Scandinavian Journal of
Management, 29(1), 48e62.
Lorino, P., Tricard, B., & Clot, Y. (2011). Research methods for non-representational
approaches to organizational complexity: The dialogical mediated inquiry. Or-
ganization Studies, 32(6), 769e801.
Manning, P. K. (1977). Frame analysis. By Erving Goffman. American Journal of So-
ciology, 82(6), 1361e1364.
Manning, P. K. (2008). Goffman on organizations. Organization Studies, 29(5),
677e699.
Mouritsen, J., & Kreiner, K. (2016). Accounting, decisions and promises. Accounting,
Organizations and Society, 49, 21e31.
Mouritsen, J., Larsen, H. T., & Bukh, P. N. D. (2001). Intellectual capital and the
‘capable firm’: Narrating, visualising and numbering for managing knowledge.
Accounting, Organizations and Society, 26(7), 735e762.
Mouritsen, J., & Thrane, S. (2006). Accounting, network complementarities and the
development of inter-organisational relations. Accounting, Organizations and
Society, 31, 241e275.
Nahapiet, J. (1988). The rhetoric and reality of an accounting change: A study of
resource allocation. Accounting, Organizations and Society, 13(4), 333e358.
Otley, D. T., & Berry, A. J. (1994). Case study research in management accounting and
control. Management Accounting Research, 5(1), 45e65.
Patriotta, G., & Spedale, S. (2009). Making sense through Face: Identity and social
interaction in a consultancy task force. Organization Studies, 30(11), 1227e1248.
Pentland, B. T. (1993). Getting comfortable with the numbers: Auditing and the
micro-production of macro-order. Accounting, Organizations and Society, 18(7),
605e620.
Quattrone, P., & Hopper, T. (2005). A ‘time-space odyssey’: Management control
systems in two multinational organisations. Accounting, Organizations and So-
ciety, 30(7), 735e764.
Que
re
, L. (1989). « La vie sociale est une sce ne ». Goffman revu et corrige
par
Garfinkel. In In Le parler frais d’Erving Goffman (pp. 47e82). Paris: Les Editions
de minuit.
 P. Lorino et al. / Accounting, Organizations and Society 58 (2017) 32e49
Ricœur, P. (1990). Time and narrative (Vol. 1). Chicago: Chicago University Press.
Riggins, S. H. (1990). Beyond Goffman: Studies on communication, institution, and
social interaction. Berlin and New York: Mouton de Gruyter.
Roberts, J., & Jones, M. (2009). Accounting for self interest in the credit crisis. Ac-
counting, Organizations and Society, 34(6), 856e867.
Scapens, R. W., & Macintosh, N. B. (1996). Structure and agency in management
accounting research: A response to Boland's interpretive act. Accounting, Or-
ganizations and Society, 21(7), 675e690.
Schatzki, T. R. (2001). Practice mind-ed orders. In T. R. Schatzki, K. Knorr Cetina, &
E. von Savigny (Eds.), The practice turn in contemporary theory (pp. 42e55).
London and New York: Routledge.
Scheff, T. J. (2005). The structure of Context: Deciphering frame analysis. Sociological
Theory, 23(4), 368e385.
Skærbæk, P., & Tryggestad, K. (2010). The role of accounting devices in performing
corporate strategy. Accounting, Organizations and Society, 35, 108e124.
Solomon, J. F., Solomon, A., Joseph, N. L., & Norton, S. D. (2013). Impression man-
agement, myth creation and fabrication in private social and environmental
reporting: Insights from Erving Goffman. Accounting, Organizations and Society,
38(3), 195e213.
Stark, D. (2009). The sense of dissonance. Accounts of worth in economic life. Princeton
and Oxford: Princeton University Press.
49
Strong, P. M. (1983). Review essay. The importance of being Erving - Erving Goff-
man, 1922-1982. Sociology of Health & Illness, 5(3), 345e355.
Suddaby, R. (2006). From the editors: What grounded theory is not. Academy of
Management Journal, 49(4), 633e642.
Swieringa, R. J., & Weick, K. E. (1987). Management accounting and action. Ac-
counting, Organizations and Society, 12(3), 293e308.
Tsoukas, H. (2009). A dialogical approach to the creation of new knowledge in or-
ganizations. Organization Science, 20(6), 941e957.
Tsoukas, H., & Chia, R. (2002). On organizational becoming: Rethinking organiza-
tional change. Organization Science, 13(5), 567e582.
Vinnari, E., & Skærbæk, P. (2014). The uncertainties of risk management. Accounting,
Auditing & Accountability Journal, 27(3), 489e526.
Vollmer, H. (2007). How to do more with numbers. Elementary stakes, framing,
keying, and the three-dimensional character of numerical signs. Accounting,
Organizations and Society, 32(6), 577e600.
Vollmer, H. (2013). The sociology of disruption, disaster and social change. Punctuated
cooperation. Cambridge University Press.
Yang, C., & Modell, S. (2015). Shareholder orientation and the framing of manage-
ment control practices: A field study in a Chinese state-owned enterprise. Ac-
counting, Organizations and Society, 45, 1e23.
 Accounting, Organizations and Society 58 (2017) 50e66

Contents lists available at ScienceDirect

Accounting, Organizations and Society

journal homepage: www.elsevier.com/locate/aos

Audit time pressure and earnings quality: An examination of

accelerated filings
Tamara A. Lambert a, *, Keith L. Jones b, Joseph F. Brazel c, D. Scott Showalter c
a
Lehigh University, Rauch Business Center, 621 Taylor Street, Bethlehem, PA 18015, United States
b
University of Kansas, 1654 Naismith Dr., Lawrence, KS 66045, United States
c
North Carolina State University, Nelson Hall, Raleigh, NC 27695, United States

a r t i c l e i n f o a b s t r a c t

Article history: Using publicly available data from annual reports, we find that SEC rule changes (33-8128 and 33-8644)
Received 14 January 2014 that impose time pressure on the audits of registered firms have a negative impact on earnings quality,
Received in revised form which we interpret as evidence of lower audit quality. Consistent with our predictions, we find that the
1 November 2016
10-K accelerations reduced audit quality only when it actually reduced the number of days from year-end
Accepted 20 March 2017
Available online 24 April 2017
to audit report date, and that this effect was more acute for smaller, accelerated filers and during the
initial deadline change (relative to the second). We also provide insights into the quality of these audits
by conducting a survey of thirty-two retired audit partners. Survey results underscore the challenges
JEL classification:
M40
time pressure imposes on receiving and evaluating complex valuations (such as for derivatives, pensions,
M41 and goodwill) and resolving audit adjustments.
M42 © 2017 Elsevier Ltd. All rights reserved.
M48

Keywords:
Accelerated SEC filings
Audit delay
Earnings quality
Time pressure

1. Introduction
Securities and Exchange Commission (SEC) rules 33-8128 and
33-8644 substantially reduced the 10-K filing period for large
accelerated filers and accelerated filers by 15 days, from 90 days
after fiscal year-end to 60 and 75 days, respectively (SEC 2002,
2005).1 For many firms and their auditors, such regulation led to
* Corresponding author.
E-mail addresses: tal413@lehigh.edu (T.A. Lambert), keithjones@ku.edu
(K.L. Jones), joe_brazel@ncsu.edu (J.F. Brazel), scott_showalter@ncsu.edu
(D.S. Showalter).
1
According to rule 33-8128 (SEC 2002), an accelerated filer (AF) is a firm that
meets the following conditions at the end of its fiscal year: 1) Its common equity
public float (the part of equity not held by management or large shareholders) was
$75M or more as of the last business day of its most recently completed second
fiscal quarter; 2) The company has been subject to the reporting requirements of
Section 13(a) or 15(d) of the Exchange Act for a period of at least 12 calendar
months; 3) The company has previously filed at least one annual report pursuant to
Section 13(a) or 15(d) of the Exchange Act; and 4) The company is not eligible to use
Forms 10-KSB and 10-QSB. A large accelerated filer (LAF) is defined as an AF with a
worldwide market value of outstanding voting and non-voting common equity
held by non-affiliates of $700M or more (SEC 2005). A non-accelerated filer (NAF) is
a firm that does not meet the definition of an AF.
exogenously imposed year-end time pressure to meet the new fil-
ing deadlines. This setting provides a natural experiment that we
use to provide archival evidence on the effect of time pressure on
audit/earnings quality. We also provide rich qualitative information
related to the pressure audit firms experienced during the accel-
eration periods, areas in which time pressure resulted in audit
difficulties, the ways in which audit firms attempted to alleviate the
pressure, and the resulting quality of accelerated audits. The com-
bination of our archival and qualitative data allows us to further
explore the impact of regulatory-induced pressure on audit firms
and contribute to an emerging stream of literature that explores the
impact of controversial regulatory changes on the quality of infor-
mation supplied to financial statement users.
Experimental and survey research has shown that increasing
audit time pressure may limit the extent to which auditors employ
a questioning mind and critically evaluate evidence (e.g., McDaniel,
1990; Otley & Pierce, 1996; Willett & Page, 1996; Asare, Trompeter,
& Wright, 2000; Braun, 2000; Coram, Ng, & Woodliff, 2004; Nelson,
2009; PCAOB, 2012). Archival evidence illustrates that fewer audit
hours, in general, are associated with lower quality earnings
(Caramanis & Lennox, 2008). Our setting is one in which we know
the extent and source of time pressure and allows us to contribute

http://dx.doi.org/10.1016/j.aos.2017.03.003
0361-3682/© 2017 Elsevier Ltd. All rights reserved.
 T.A. Lambert et al. / Accounting, Organizations and Society 58 (2017) 50e66
archival results to complement and triangulate existing experi-
mental/survey research exploring the effect of time pressure on
audit quality.
We extend prior studies which make use of this setting (e.g.,
Bryant-Kutcher, Peng, & Weber, 2013; Doyle & Magilke, 2013;
Impink, Lubberink, van Praag, & Veenman, 2011; Krishnan &
Yang, 2009) by examining whether it is the extent of time
pressure placed on the audits that is associated with lower
earnings quality. We also identify conditions under which firms
that were seemingly affected by the acceleration did not expe-
rience a reduction in earnings equality (i.e., firms that needed to
file earlier, but did not need to reduce audit delay to do so). Our
analyses enable us to more definitively describe the effects of the
SEC 10-K accelerations and to identify the conditions under
which such accelerations do not impair the quality of earnings.
This allows us to speak to how future accelerations and/or other
regulatory activity may impact financial reporting quality based
on the extent to which they would place time pressure on
contemporary audits. Thus, this research should assist U.S. and
international regulatory organizations considering future accel-
erations of financial reports.
In our study, we first describe how exogenously-induced in-
creases in time pressure may substantially impact the audit
approach and limit year-end testing of account balances and
transactions. We use audit delay, defined as the length of time from
a company's fiscal year-end to the date of the auditor's report
(Ashton, Willingham, & Elliott, 1987), to develop a continuous
measure of audit time pressure and to group firms based on the
extent to which they were most affected by the deadline change.
We categorize firms as: already filing before the new deadline (i.e.,
unaffected by the accelerations) (Group A); needing to file the 10-K
earlier, but not required to reduce audit delay to do so (Group B); or,
those for which complying with the SEC accelerations of 10-K fil-
ings required a reduction in audit delay (Group C). Time pressure is
determined by (1) the extent to which it was necessary to reduce
the audit delay to meet the new deadline (calculated as the prior
year audit delay minus the current year deadline in days) and (2)
whether a firm is classified as being in Group C.
Greater time pressure at year-end increases the likelihood that
auditors are faced with a choice between having their clients miss
the filing deadline because they are unable to obtain sufficient
appropriate evidence by the 10-K filing deadline or performing a
rushed, lower quality audit. We use working capital accruals to
proxy for audit/earnings quality (Dechow & Dichev, 2002) and find
no evidence that the deadline reductions negatively affected the
earnings quality of Group A and B firms. In short, if 10-K accelera-
tions do not put time pressure on the audit, earnings quality is not
significantly affected. Conversely, we do demonstrate that both the
extent of time pressure faced by auditors to meet the accelerated
10-K deadlines, and the audits of Group C firms in general, are
associated with lower earnings quality. We also find that both of
our time pressure measures are positively associated with the
likelihood that the audit is not complete by the accelerated dead-
line (i.e., the firm files their 10-K with the SEC late). We next explore
whether time pressure differentially affected accelerated filers vs.
large accelerated filers, as well as the earnings quality of firms
during the first (75 day deadline) vs. second deadline change (60
day deadline). Our evidence suggests that audit time pressure has a
more negative effect on the earnings quality of accelerated filers
(relative to large accelerated filers), and for the initial deadline
change (relative to the second), during the 10-K acceleration
periods.
We then conduct a survey of audit partners that asked them to
provide both qualitative and quantitative data related to one spe-
cific 10-K acceleration of a client they served. Providing support for
51
our time pressure proxies, we find participants felt a fair amount of
time pressure on these audits and that the pressure they felt was
significantly associated with the number of days by which the audit
report was accelerated. Consistent with our archival analysis, the
vast majority of partners indicated that the audits of large accel-
erated filers were better equipped to handle the time pressure
caused by the accelerations. Providing insight into why (or, via
what avenue) time pressure negatively affected earnings quality,
we find audit time pressure is positively associated with the level of
difficulty associated with resolving audit adjustments. With respect
to identifying “best practices” for ameliorating the effects of time
pressure, partners indicate that working more hours, performing
more interim testing, and rescheduling the audits of non-public
companies were the most effective strategies. On the other hand,
increasing the use of computer assisted audit techniques on the
audit was not deemed to be an effective strategy. Finally, an
exploratory analysis of our survey responses suggests that, for 10-K
accelerations, interim testing and the percentage of partner time
spent at the client are positively associated with the effectiveness of
audit procedures, reducing the difficulty associated with resolving
year-end audit adjustments, and overall audit and financial
reporting quality.
Overall, the combination of our archival and survey-based evi-
dence should inform deliberations by U.S. and non-U.S. regulatory
bodies considering future filing accelerations.2 Regulators should
be acutely aware of the extent to which such accelerations may
impact the amount of time pressure placed on the financial state-
ment audit. Our aforementioned results related to accelerated filers
suggest that caution should be taken before considering a further
reduction for smaller firms (e.g., from 75 to 60 days) or expanding
accelerations to even smaller, non-accelerated filers (who currently
still face a 90-day filing deadline). If such accelerations are under-
taken in the future, audit firms can strive to increase the extent to
which the best practices we identify can be implemented on a
particular audit. In sum, our study will inform audit teams charged
with handling any future events or regulatory acts that place
greater year-end time pressure on the audit team. For example, the
majority of our survey respondents indicated that the recently
proposed PCAOB standard in relation to expanding the content of
the audit report would induce additional year-end time pressure on
the auditor and potentially impair audit quality.
The remainder of the paper is organized as follows. Section 2
provides background and develops our hypothesis and research
question. Sections 3 and 4 describe our archival and survey ana-
lyses, respectively. Section 5 provides our conclusion.
2. Background and theoretical development
2.1. SEC regulation
Shortly after the passage of the Sarbanes-Oxley Act (SOX), the
2
For example, the European Commission has been contemplating a reduction in
audit delay to improve the timeliness of communications between auditors and
stakeholders (see question number 11 in the European Commission Green Paper,
Audit Policy: Lessons from the Crisis (available at: http://ec.europa.eu/internal_
market/consultations/docs/2010/audit/green_paper_audit_en.pdf). In Canada, the
filing deadline for publicly traded companies was reduced from 140 to 90 days in
2004. Over time, Canada may consider emulating the current 10-K filing deadlines
imposed by the SEC. Also, our results can inform SEC deliberations related to
foreign filers. Foreign issuers recently experienced a reduction in their filing
deadline from 6 months to 4 months. It seems logical that the SEC might consider a
further reduction to this deadline (i.e., to 90 days, as originally proposed), and the
current accelerated deadline applies to all foreign issuers regardless of size. Our
findings suggest that separating foreign issuers into different filing groups might be
appropriate.
52 T.A. Lambert et al. / Accounting, Organizations and Society 58 (2017) 50e66

Fig. 1. Proposed and Enacted Changes to the 10-K Filing Deadline for Accelerated Filers*.
* According to rule 33-8128 (SEC 2002), an accelerated filer (AF) is one that meets the following conditions at the end of its fiscal year: 1) Its common equity public float was $75M
or more as of the last business day of its most recently completed second fiscal quarter; 2) The company has been subject to the reporting requirements of Section 13(a) or 15(d) of
the Exchange Act for a period of at least 12 calendar months; 3) The company has previously filed at least one annual report pursuant to Section 13(a) or 15(d) of the Exchange Act;
and 4) the company is not eligible to use Forms 10-KSB and 10-QSB.
a
Rule 33-8128, Acceleration of Periodic Report Filing Dates and Disclosure Concerning Website Access to Reports (SEC 2002).
b
Rule 33-8507, Temporary Postponement of the Final Phase-In Period for Acceleration of Periodic Filing Dates (SEC 2004).
c
Rule 33-8644, Revisions to Accelerated Deadlines for Filing Periodic Reports (SEC 2005).
d
Rule 33-8644 created a new filer category called the Large Accelerated Filer (LAF), which is an accelerated filer with a worldwide market value of outstanding voting and non-
voting common equity by non-affiliates of $700M or more (SEC 2005).

SEC issued rule 33-8128, Acceleration of Periodic Report Filing Dates
and Disclosure Concerning Website Access to Reports (SEC 2002). This
rule was intended to substantially shorten the Form 10-K filing
deadline from 90 to 60 days, over a two-year period, for all firms
with outstanding common equity by non-affiliates of $75M or
more. At that time, all firms subject to the deadline change were
known as “accelerated filers”. The deadline went from 90 days to 75
days on December 15, 2003 with a further reduction to 60 days
scheduled for December 15, 2004 (which was later postponed). The
objective of the deadline reduction was to provide investors with
more timely, relevant information (SEC 2002).
The acceleration of filings has been a controversial and heated
topic of discussion. The SEC received 302 comment letters on the
proposal to accelerate the deadlines; 20 supported the acceleration
and 282 opposed it.3 Based on negative public reaction and con-
cerns expressed by filers and auditors over whether they would be
able to file reports on a timely basis, the SEC adopted rule 33-8507
in November of 2004. This rule postponed the final phase-in date of
the 60-day filing deadline to fiscal year-ends on or after December
15, 2005 (SEC 2004).
The SEC passed rule 33-8644 in December of 2005 (SEC 2005).
This rule created two categories of firms subject to the filing
deadline change. A large accelerated filer (LAF) is a firm with a
worldwide market value of outstanding common equity held by
non-affiliates of $700M or more. LAFs became subject to a reduced
60-day deadline after December 15, 2006. Accelerated filers (AFs),
firms with outstanding common equity held by non-affiliates of
between $75M and $700M, remained subject to a 75-day deadline.
Non-accelerated filers (NAFs), firms with outstanding common
equity by non-affiliates of less than $75M, continued to be subject
to the original 90-day deadline. Fig. 1 illustrates the proposed
changes to the 10-K filing deadline and the changes that were ul-
timately enacted.
2.2. Existing research related to SEC rules 33-8128 and 33-8644
In an early examination of the SEC's decision to accelerate fil-
ings, Krishnan and Yang (2009) find no association between levels
of audit delay and earnings quality around the first deadline
change. However, Krishnan and Yang (2009) do not examine the
time pressure faced by particular audits. Impink et al. (2011)
conclude the accelerated deadlines in 2003 and 2006 are not
associated with an overall increased occurrence of late filings. They
also observe that firms with effective internal controls typically met
3
http://www.sec.gov/rules/extra/33-8089summary.htm#P107_2938.
the filing deadlines, while approximately half of the firms with
internal control weaknesses filed late.
Two more recent studies directly consider the impact of
accelerated filings on financial reporting quality. Bryant-Kutcher
et al. (2013) find evidence of an increase in restatements for
accelerated filers (vs. firms that did not need to file their 10-K
earlier to meet the new deadlines). Doyle and Magilke (2013)
focus on the overall usefulness of 10-K filings by measuring the
absolute value of the three-day market reaction to the 10-K filing.
They also compare the reporting quality of firms that were
required to file earlier to meet the new deadlines to firms who
were already filing before the accelerated deadlines. Doyle and
Magilke (2013) illustrate that AFs and LAFs that were required
to accelerate 10-K filings for the first deadline change experienced
significant decreases in reporting quality. On the other hand, they
find that the reporting quality of LAFs actually increased in the
second deadline period.
Our study extends prior research in relation to SEC 10-K accel-
erations and audit time pressure in several important ways. First,
we discuss and identify a mechanism through which earnings
quality might suffer during mandated accelerations (i.e., via
increased time pressure on the audit). To provide context to this
discussion (described in Section 4), we also incorporate qualitative
data (i.e., quotes) obtained from a survey of audit partners that
experienced 10-K accelerations. Second, we empirically examine
whether mandated filing accelerations that did not require a
reduction in observed audit delay affected earnings quality. As
noted above, prior research has not considered the actual effect on
audit delay/time pressure when identifying firms ‘affected’ by the
mandate. Third, we concentrate on firms that were most likely
affected by the SEC's deadline changes and examine whether the
extent of audit time pressure affects the quality of audited earnings,
as well as the ability to file the 10-K on time. By doing so, we
contribute archival evidence to triangulate experimental and sur-
vey research related to audit time pressure.
Fourth, by isolating the mechanism by which accelerations
affect earnings quality, we examine the implications of further
filing accelerations, as well as reductions in the 10-K filing win-
dow for non-accelerated filers. These results should inform future
deliberations by regulatory bodies considering accelerations of
financial reporting, as well as other future regulatory acts that
may place additional year-end time pressure on audits. Last, we
use our survey data to identify best practices with respect to
strategies that were employed to maintain audit quality during
the years of acceleration. We therefore provide practical guidance
to audit firms concerned with alleviating year-end time pressure
on their audits.
 T.A. Lambert et al. / Accounting, Organizations and Society 58 (2017) 50e66 53

2.3. Audit delay, time pressure, and earnings quality team working even longer hours per day than usual. With greater
time pressure, one might expect some negative effects on
The audit report date (i.e., the audit sign-off date) is the date by engagement team morale and effort levels.4
which the auditors have gathered sufficient and appropriate evi- In sum, the SEC has twice reduced 10-K filing deadlines by 15
dence to conclude fieldwork and issue an audit opinion on a days and prior research suggests that greater time pressure on the
company's financial statements (Arens, Elder, & Beasley, 2010). auditor to meet these accelerated filing deadlines will impair audit
Audit delay (or audit report lag) is defined as the length of time quality and adversely affect the quality of earnings. Formally, we
from a company's fiscal year-end to the date of the auditor's report hypothesize:
(Ashton et al., 1987). At some point after the audit report date, the
H1. Greater audit time pressure to meet accelerated filing dead-
client files the audited 10-K with the SEC; this date is termed the
lines will lead to lower earnings quality.
filing date. It is important to note that the financial statements are
finalized by the audit report date (not the filing date). While pre-
vious research has identified important determinants of audit delay
2.4. Filer type and deadline change
(e.g., Whitworth & Lambert, 2014), we focus on the consequences of
regulatory-induced changes to audit delay.
The SEC's decision to have three distinct groups of filers (LAFs,
Mandated reductions in audit delay put time pressure on audits,
AFs, and NAFs) and three separate reporting deadlines (60 days, 75
particularly at year-end when there are a finite number of days to
days, and 90 days, respectively) was controversial. Some companies
complete the audit. Recent survey research identifies time pressure
(e.g., Comcast Corporation) expressed that all public companies
as an impediment to audit quality. One partner interviewed by
should be required to adhere to the same deadline. Others,
Christensen, Glover, Omer, and Shelley (20, 2015) states “[when I
including the AFL-CIO and KPMG LLP, agreed with the notion of
speak of time pressure as an impediment to audit quality], I'm
excluding smaller companies because they may not have the
talking about the findings or the need for information that comes
necessary resources or infrastructure to meet the accelerated
right at the end … and there's a lot of pressure from management,
deadline (SEC 2002). We shed additional light on this issue by
and with some clients from the audit committee, in my opinion, to
examining if the relation between regulatory-induced time pres-
go ahead and sign the audit opinion without getting that evidence.”
sure and earnings quality differs by filer type (AFs vs. LAFs) and
Experimental research has established that increasing time pres-
deadline change.
sure can have a detrimental effect on audit effectiveness and lead to
The American Bar Association (ABA) argued that “large busi-
dysfunctional auditor behavior (e.g., accepting doubtful evidence,
nesses tend to be more complex, often with international opera-
truncating sample selections) (Coram et al., 2004; DeZoort & Lord,
tions, multiple divisions and subsidiaries and investments from
1997; McDaniel, 1990).
other entities from which they often must await reports.”5 Based on
Negotiations between auditors and clients related to audit ad-
this argument, the quality of LAF audits may experience a stronger
justments occur frequently toward the end of an audit and are a
negative impact from exogenously imposed time pressure. On the
crucial link between the quality of the audit and the quality of the
other hand, Pizzini, Lin, Vargus, and Ziegenfuss (2012) provide
financial statements (Salterio, 2012). Reducing audit delay may
evidence that the audits of LAFs use the companies’ internal audit
curtail the window within which the auditor can effectively
functions to reduce audit delay. In addition, the audits of larger
negotiate the outcome of sensitive or contentious accounting issues
companies probably already employ more extensive interim testing
(Salterio, 2012). For example, Braun (2000) concludes that time
as Bamber, Bamber, and Schoderbek (1993) observe that larger
pressure's earliest effects are manifested through the filtering out of
firms exhibit shorter audit delays. For LAFs that did employ sub-
cues related to the qualitative aspects of misstatements. Also,
stantial year-end testing prior to accelerations, Brazel, Carpenter,
Bennett, Hatfield, and Stefaniak (2014) show that auditors' pre-
and Jenkins (2010) find a positive relation between client size
negotiation judgments are more concessionary in the presence of
and the degree to which auditors change the staffing and timing of
deadline pressure. Specifically, auditors react to end-of-
testing. In other words, because auditors of LAFs tend to be present
engagement deadline pressure by conceding more on their initial
at the client throughout the year, it may be (relatively) easier to
negotiation positions. Because the listing of necessary audit-related
shift testing from year-end to interim to meet filing deadlines and
adjustments to the financial statements is not finalized until year-
still maintain a similar level of audit quality. LAF clients also tend to
end, the audit team has less time to fully evaluate the qualitative
be more prestigious and these audits typically have a greater ability
and quantitative aspects of misstatements and other accounting
to procure higher quantities of competent audit staff (if available).
matters with accelerated 10-K deadlines.
Thus, LAFs may be less likely to suffer a reduction in audit/earnings
Auditors often require considerable time after year-end (and
quality due to accelerations.
substantial post-fiscal year-end evidence) to examine client esti-
AFs are smaller, less prestigious, and likely have a larger portion
mations (e.g., the reserve for obsolete inventory, income tax pro-
of their audit work performed after year-end than LAFs (Bamber
vision, fair value of complex financial instruments). The recent
et al., 1993). As such, the audit approaches for AFs were likely
trend from historical cost to fair value accounting makes the year-
heavily affected by an acceleration. However, AFs were not subject
end analysis of client estimates even more crucial and requires
to the second deadline change from 75 days to 60 days. LAFs, on the
more time to verify the estimates at or after year-end. Likewise,
other hand, were subject to both the first and the second deadline
exercising appropriate levels of professional skepticism often re-
quires additional work/time to obtain sufficient and appropriate
evidence (Nelson, 2009; PCAOB 2012). More time pressure on the 4
Of course, auditors may be able to mitigate the increased time pressure
audit team, induced by the need to meet shorter 10-K filing dead- resulting from accelerations by performing more interim procedures, relying more
lines, may limit the extent to which auditors employ a questioning on the client's internal control systems and internal audit function, and using
mind and critically evaluate evidence. Indeed, PCAOB inspections of advanced audit technology (cf., PCAOB 2007, 2010; Brazel & Agoglia, 2007; Pizzini
et al., 2012). Thus, there are strategies available to the auditor that may reduce the
audits during 10-K accelerations noted a failure to apply an
extent to which post-fiscal year-end audit procedures/evidence are needed to
appropriate level of skepticism when conducting tests and evalu- provide an acceptable level of audit quality. We investigate these and other pos-
ating results (PCAOB 2008). Last, with more audit procedures sibilities in Section 4 of this study.
5
compressed into a shorter window, the resulting effect is the audit http://www.sec.gov/rules/proposed/s70802/skeller1.htm.
54 T.A. Lambert et al. / Accounting, Organizations and Society 58 (2017) 50e66
Fig. 2. Matrix of 10-K Filing Status and Audit Report Delay.
* The focus of this paper is on firms in Group C. These firms had audit delays in the prior year that were greater than next year's deadline. Thus, the auditors of these firms
experienced regulatory-induced time pressure as they were forced to reduce their audit delay in order to file on time in the next year. The audits of firms in Group A were likely
unaffected by the deadline change since their audit delay and file delay in the prior year met the next year's deadline. The audits of firms in Group B may have been affected by the
deadline change depending on how close their prior year audit delay was to the next year's deadline. We address the time pressures, changes in audit delay, and earnings quality
experienced by all three groups in Section 3.2 of this study.
changes. The second acceleration of 15 days was proportionally
larger (i.e., a 20% decrease) than the first acceleration (i.e., a 17%
decrease). More importantly, any slack in the reporting processes
for LAFs would have been absorbed in the first deadline change.
Thus, it is possible that the audit/earnings quality of LAFs was more
impaired during the second acceleration. In sum, the extent to
which pressure is associated with lower earnings quality may differ
by filer type and deadline. We therefore examine the following
research question:
RQ. Does the association between audit time pressure and
earnings quality differ by filer type and deadline change?
3. Archival analysis
3.1. Method
3.1.1. Sample
We derive our sample from the years that the two 10-K filing
reductions took effect. The first deadline change took effect for LAFs
and AFs with fiscal years-ending after December 15, 2003. The
second deadline change took effect for all LAFs with fiscal years-
ending after December 15, 2006. Thus, pressure-induced re-
ductions in audit delay occurred during fiscal year-ends from
December 15, 2003 through December 14, 2004 and from
December 15, 2006 through December 14, 2007. To identify firms
whose audit report lags were required to be reduced by one of the
deadline changes, we organize accelerated filers into three groups
as shown in Fig. 2. Firms in Group A filed their prior year 10-K in
time to meet the subsequent year's deadline. The audits of these
firms are not likely affected by the deadline change. There is no
time pressure on the audit due to accelerations, as the prior year
audit was completed and the 10-K was filed prior to the upcoming
deadline. Firms in Group B had their prior year audit completed
before the subsequent year's accelerated deadline (based on the
prior year's audit report date), but did not file their prior year 10-K
within the time frame specified by the new deadline. The audits of
these firms may be affected by the deadline reduction if they have
to accelerate the audit sign-off in order to meet the following year's
10-K filing deadline. Still, it is likely that many Group B firms did not
experience a mandated reduction in audit delay.
Firms in Group C did not complete their prior year audit in time
to meet the subsequent year's accelerated filing deadline. These
firms experienced time pressure from the deadline reductions
because they had to accelerate the sign-off date of their audit (from
the previous year) in order to meet the new filing deadline. In short,
the audits of firms in Group C were most likely affected by the
regulation.6
Audit Analytics is our source for audit delay data that is used to
measure the amount of regulatory-induced time pressure experi-
enced on the audit. Compustat is our source for financial data.
Combining the samples provides 18,200 firm-year observations.
We then exclude financial institutions and regulated industries
because of their unique nature of accounting for accruals (Frankel,
Johnson, & Nelson, 2002; Tucker & Zarowin, 2006), firms that were
not subject to the first filing deadline change (NAFs), and firms that
were not subject to the second deadline reduction (NAFs and AFs).
We remove observations with missing audit report dates, missing
control variables, and those lacking the data necessary to calculate
accruals (our measure of earnings quality). Finally, we delete
foreign issuers, firms that filed late in the prior year, and firms filing
for the first time (i.e., filing an S-1). As shown in Table 1, Group C
consists of 737 firm-year observations (265 from the first filing
deadline change and 472 from the second deadline change). The
737 firm-year observations represent 698 unique firms.7
3.1.2. Models for archival analysis
We test H1 by examining the amount of time pressure on audit
engagements during the time period of the accelerations. We
measure PRESSURE by taking prior year audit delay minus the
length of time available to file the 10-K with the new deadlines (i.e.,
75 days for the first deadline change and 60 days for the second
deadline change). Audit delay is defined as the audit report date
less the fiscal year-end date (Ashton et al., 1987). For example, if an
auditor, in the year prior to the first deadline change, had an audit
sign-off date 85 days after year-end, then PRESSURE would be equal
to 10. At a minimum, this auditor will need to reduce the audit
delay by 10 days in the current year in order for the 10-K to be filed
within 75 days after fiscal year-end.
To test our hypothesis and research question, we focus on
PRESSURE in the years of the deadline change because this setting
provides a natural experiment to examine pressure-induced re-
ductions in audit delay imposed from an exogenous source. By
examining these reductions during the year of the filing deadline
changes, we do not need to disentangle the reason for the reduction
in audit delay and our analyses speak directly to the criticism of the
6
We acknowledge that, for some Group C firms, reducing slack or inefficiencies
in the audit process may have been methods used to meet the upcoming deadline.
The extent that Group C firms fit this description biases against H1.
7
Thirty-nine LAFs fell into Group C during both filing deadline changes and are
included in our sample twice.
 T.A. Lambert et al. / Accounting, Organizations and Society 58 (2017) 50e66 55

Table 1
Sample selection.

1st Deadline Reductiona 2nd Deadline Reductionb Total

Net Observations after Combining Compustat & Audit Analytics 9,481 8,719 18,200
Financial Institutions (6000-6999) & Regulated Utilities (4900-4999) 3,025 2,862 5,887
NAFs (and AFs with regard to the 2nd Deadline Reduction) 2,920 4,709 7,629
Missing PY or CY Audit Report Date or File Report Date 334 21 355
Missing Control Variables 409 103 512
Missing Information to Calculate Discretionary Accruals 136 59 195
Filed Late in Prior Year, Foreign Issuers and S-1's 520 80 600

2,137 885 3,022

Group A (i.e., PY File Delay < New Filing Deadline) 558 214 772c
Group B (i.e., PY Audit Delay < New Filing Deadline) 1,314 199 1,513c
Group C (i.e., PY Audit Delay > New Filing Deadline) 265 472 737c

2,137 885 3,022

a
First filing deadline reduction from 90 days to 75 days for all accelerated filers (all firms with outstanding common equity by non-affiliates greater than $75M). Deadline
change affected fiscal year-ends from December 15, 2003eDecember 14, 2004.
b
Second filing deadline reduction from 75 days to 60 days for all Large Accelerated Filers (firms with outstanding common equity by non-affiliates greater than $700M). The
deadline change affected fiscal year-ends from December 15, 2006eDecember 14, 2007.
c
Amount represents firm-year observations. Group A had 662 unique firms, Group B had 1406 unique firms, and Group C had 698 unique firms.

filing deadline change creating unwanted (i.e., harmful) time
pressure on the audit. We use a continuous measure across all
groups by allowing some firms (i.e., those in Groups A and B) to
have negative audit time pressure (i.e., they had extra slack because
their year-end audit work was completed earlier). We additionally
focus on firms in each Group (i.e., A, B, and C). We expect firms in
Group C (i.e. the firms with positive PRESSURE) will be the most
adversely affected in relation to earnings quality. We create an in-
dicator variable for each group (i.e., GROUP A, GROUP B, and GROUP
C) set equal to one if the firm is in the corresponding group and zero
otherwise. As our first test of H1, we estimate the following
regression model8:
PRESSURE, GROUP B, and GROUP C. If the coefficient on
ACCEL_FILER is significant, then the dependent variable (LATE_-
FILER or AbsChWCACC) is significantly affected if the firm is an
AF relative to an LAF (i.e., an increase in the intercept for being an
AF). If an interaction is significant (e.g., PRESSURE*ACCEL_FILER
or GROUP C*ACCEL_FILER), then the relation between the
dependent variable and PRESSURE or GROUP C is different for
AFs relative to LAFs. The second indicator (SECOND) is set equal
to one if the firm year observation falls during the second
deadline change and zero if it falls in the first deadline change. As
above, we also interact SECOND with PRESSURE, GROUP B and
GROUP C.

AbsChWCACCt (or LATE_FILERt) ¼ b0 þ b1PRESSUREt (or b1GROUP B þ b2GROUP C) þ b3ChSEASONt þ b4ChLOSSt þ b5ChCFOt þ b6ChB/Mt þ
b7ChLgMVEt þ b8ChLEVERAGEt þ b9ChFINANCEt þ b10ChDISTRESSt b11ChGROWTHt þ b12ChRESTRUCTUREt þ b13ChEXTRAORDINARYt þ
b14ChACQUISTIONt þ b15ChQUALIFIEDt þ b16ChGOING_CONCERNt þ b17ChAUDITORt þ b18ChLEADERt þ b19ChLgAUDIT_FEESt
þ b20Year þ b21-33Industry þ ε (1)

Because our theory suggests the 10-K acceleration will induce
a change in audit delay or time pressure on some audits, we
employ a “changes” model above.9 H1 is supported if the coef-
ficient on PRESSURE and/or GROUP C is positive and significant.
We test our RQ by incorporating two indicator variables into
Model (1). The first is set equal to one if the firm is an AF and zero
for an LAF (ACCEL_FILER). We also interact ACCEL_FILER with
Our primary dependent variable measures the change in the
quality of earnings disseminated to external users during 10-K ac-
celerations. We employ Dechow and Dichev’s (2002) measure of
working capital accruals to measure earnings quality. We estimate
Model (2) cross-sectionally and use the residual (WCACC) from
Model (2) as a measure of accrual quality:
DWCit ¼ b0 þ b1CFOit-1 þ b2CFOit þ b3CFOitþ1 þ εit (2)

8
Consistent with Peterson (2009), we control for standard error bias due to
repeated measures of firm and year by including year dummies in the regression
and performing a cluster regression with a firm identifier as a repeated measure
(SAS procedure proc surveyreg with a gvkey cluster). When the dependent variable
is LATE_FILER, the model is a logistic regression modeling the probability of being a
late filer. LATE_FILER equals 1 if the company filed after the filing deadline and 0 if
the company filed on time.
9
See Appendix A in the Online Appendix for robustness checks employing
alternative model specifications which (a) utilize a “levels” model, (b) consider the
extent to which PRESSURE (versus less audit-related variables examined by prior
research) drives our results, (c) examine an alternative dependent variable as a
measure of earnings quality, and (d) include both PRESSURE and GROUP C in the
same regression.
where DWC is the change in working capital from year t-1 to year
t.10 Our variable of interest is the absolute value of the change
(AbsChWCACC) in the residual (WCACC) from year t-1 to year t
10
DWC is computed as DAccounts Receivable þ DInventory - DAccounts Payable -
DTaxes Payable þ DOther Assets. More specifically, DWC is computed from Com-
pustat items as DWC ¼ - (RECCH þ INVCH þ APALCH þ TXACH þ AOLOCH). All
variables in Model (2) are scaled by average total assets. Dechow and Dichev (2002)
require at least eight years of data to estimate Model (2). We do not have eight
years of data for all of the firms in our sample. Therefore, consistent with Jones,
Krishnan, and Melendrez (2008), we estimate Model (2) cross-sectionally and use
the residual from Model (2) as a measure of accrual quality.
56 T.A. Lambert et al. / Accounting, Organizations and Society 58 (2017) 50e66
under the assumption that earnings that are managed from
otherwise neutral earnings in either direction meet the definition
of poor earnings quality.
Our secondary dependent variable examines the case where
substantial time pressure at year-end may inhibit the ability of the
auditor to obtain sufficient and appropriate evidence to provide an
opinion before the accelerated deadline. When the SEC proposed
accelerating 10-K deadlines, comments on the proposal noted that
audit firms currently did not have enough slack in their audit and
reporting timeline to meet the new deadline change (SEC 2002). As
discussed in our development of H1, substantial time pressure at
year-end may inhibit the ability of the auditor to obtain sufficient
and appropriate evidence to provide an opinion before the accel-
erated deadline. Under these conditions, auditors may be faced
with a choice between having their clients miss the filing deadline
or performing a rushed, lower quality audit. Thus, in conjunction
with studying earnings quality, we consider whether time pressure
affects the likelihood that the firm files its 10-K after the acceler-
ated deadline (LATE_FILER). We expect the greater time pressure
induced by 10-K accelerations will increase the likelihood that the
audit is not complete at the deadline and the firm files its 10-K late
with the SEC (LATE_FILER). LATE_FILER is set equal to one if the firm
filed late and zero otherwise.
As previously noted, PRESSURE measures the extent to which
audit delay, at a minimum, must be reduced/changed to meet the
new filing deadline and we examine changes in earnings quality. As
such, we use change measures for all control variables (with the
exception of our year and industry indicator variables). Consistent
with prior research (Knechel & Payne, 2001; Lo  pez & Peters, 2012),
we include an indicator variable (ChSEASON) to classify changes in
audits of companies from busy season year-ends (i.e., fiscal year-
ends during the months of December and January) to non-busy
season year-ends (or vice versa). We include an indicator variable
as to whether or not the firm switched from a loss (ChLOSS) in the
prior year to a profit (or vice versa) because firms are expected to
manipulate accruals in a systematically different way during loss
 pez & Peters, 2012). Because discre-
years (Frankel et al., 2002; Lo
tionary accrual models do not completely extract nondiscretionary
accruals correlated with firm performance (Frankel et al., 2002), we
include change in cash flow from operations (ChCFO) (e.g., DeFond
& Subramanyam, 1998). We control for growth opportunities using
the change in the ratio of book-to-market value (ChB/M) (e.g.,
Frankel et al., 2002); size using change in the natural log of the
market value of equity (ChLgMVE) (Frankel et al., 2002; Kothari,
Leone, & Wasley, 2005); and change in leverage (ChLEVERAGE)
using the change in the ratio of total liabilities to total assets
(Frankel et al., 2002).
We control for changes in the company's financing (ChFINANCE)
due to concerns over the effect changes in the firm's capital
structure might have on working capital accruals (Ashbaugh,
LaFond, & Mayhew, 2003). ChFINANCE is equal to one if the num-
ber of outstanding shares increased by at least 10% or long-term
debt increased by at least 20% during the current year, but not
the prior year (or vice versa), and zero otherwise. We use
Zmijewski’s (1984) measure of financial distress (a weighted probit
bankruptcy prediction model) to control for financial condition due
to concerns that discretionary accrual models overestimate ac-
cruals for poorly performing companies (Dechow, Sloan, &
Sweeney, 1995; Kothari et al., 2005). ChDISTRESS equals the
change in financial distress from the prior year to the current year.
We control for changes in firm growth (ChGROWTH) by taking the
change in sales growth rate from the prior year to the current year.
Prior research finds that growth is positively associated with
discretionary accruals (Menon & Williams, 2004). Following Carver,
Hollingsworth, and Stanley (2011), we control for ChRESTRUCTURE,
which is coded as a 1 if the firm reported restructuring charges in
either the current or prior year (but not both) and zero otherwise.
We also control for variables that the prior literature has found
to impact audit delay (e.g., Bamber et al., 1993; Schwartz & Soo,
1996). ChEXTRAORDINARY equals one if the company reported an
extraordinary item in the prior year and not the current year (or
vice versa), and zero otherwise. ChACQUISITION equals one if the
company made an acquisition in the prior year and not the current
year (or vice versa), and zero otherwise. ChQUALIFIED equals one if
the company's audit opinion had an explanatory paragraph (other
than going concern) in the prior year and not the current year (or
vice versa), and zero otherwise. ChGOING_CONCERN is coded as
one if the audit report disclosed doubt about the entity's ability to
continue as a going concern in either the current or prior year (but
not both), and zero otherwise. ChAUDITOR is equal to one for any
switch in auditor during the year, and zero otherwise. ChLEADER
equals one if the company was audited by a national industry leader
in the prior year and not the current year (or vice versa), and zero
otherwise.
ChLgAUDIT_FEES is the natural log of the change in audit fees. It
reasons that if auditors are facing increased time pressure, they
would likely increase staff or increase fees to compensate for higher
engagement risk. Change in audit fees controls for increases in
audit firm effort or perceived audit risk. Finally, we include an in-
dicator variable to control for the year, and following Ashbaugh
et al. (2003), we include thirteen indicator variables to control for
industry. 11
3.2. Results
3.2.1. Descriptive statistics
Table 2 provides descriptive statistics for our variables of in-
terest and other, related variables for our sample. Panels A, B, and C
provide descriptive statistics for Groups A, B, and C, respectively
(Fig. 2 defines Groups A, B, and C).12 The absolute value of the
change in working capital accruals (AbsChWCACC) is highest for
both Group B and C firms (approximately 0.05 for both groups).
Accruals are lower (0.04), as expected, for Group A firms (firms
where the prior year 10-K was filed before the accelerated dead-
line). In addition, Group C firms filed late 22% of the time (LATE_-
FILER), Group B firms filed late 16% of the time, and Group A firms
filed late only 5% of the time.
As noted in Fig. 2 and Table 2 Panels A and B, PRESSURE does not
exist for firms in Groups A and B (mean PRESSURE for both groups
is negative, indicating prior year audit delays were below the
accelerated deadlines). Audit delay actually increased for Groups A
and B (i.e., the mean of ChDELAY is 6.4 days for Group A firms and
7.2 days for Group B firms). Consistent with the notion of little time
pressure being associated with the audits of Group A and B firms,
we observe in Panels A and B substantial time periods between
DELAY and FILEDELAY (16 and 21 days for Group A and B firms,
respectively). On the other hand, PRESSURE is present for Group C
firms. Mean PRESSURE for Group C is 9 days (i.e., on average, prior
year audit delay was 9 days in excess of the new deadlines). Group C
firms, on average, experienced a substantial reduction in audit
delay of 10.7 days (and 15.8 days if we exclude firms in Group C that
filed late).
11
Industries were divided into the following groups: SIC 0100e1499, SIC 1500-
1999, SIC 2000-2199, SIC 2200-2399, SIC 2400-2799, SIC 2800-3299, SIC 3300-
3499, SIC 3500-3999, SIC 4000-4899, SIC 4900, SIC 5000-5299, SIC 5300-5999, and
SIC 7000-7999. For parsimony, year and industry indicator variables are not
tabulated.
12
See Appendix B in the Online Appendix for descriptive statistics for our control
variables by Group.
 T.A. Lambert et al. / Accounting, Organizations and Society 58 (2017) 50e66 57

Table 2 PRESSURE and being in Group C led to a higher likelihood of filing

Descriptive statistics. late. As predicted, earnings quality decreased and late filing
Variables N Mean Median Standard Lower Upper increased among the most vulnerable of firms (i.e., firms with
Deviation Quartile Quartile
greater PRESSURE and firms in Group C).
The results in Table 3 related to Group B firms refine our un-
Panel A: Filer Group A
derstanding of the effects of 10-K accelerations. Prior research
AbsChWCACC 772 0.041 0.027 0.048 0.012 0.051 suggests firms “affected” by the accelerations (defined as requiring
LATE_FILER 772 0.049 0.000 0.216 0.000 0.000
a reduction in filing delay as in the case of both Groups B and C)
PRESSURE 772 30.699 32.000 17.805 46.000 14.000
ACCEL_FILER 772 0.315 0.000 0.465 0.000 1.000 typically experienced reductions in financial reporting quality
SECOND 772 0.277 0.000 0.448 0.000 1.000 (Bryant-Kutcher et al., 2013; Doyle & Magilke, 2013). The non-
DELAY 772 46.999 48.000 20.581 34.000 57.000 significant coefficient on GROUP B, when the dependent variable
ChDELAY 772 6.385 1.000 16.056 1.000 10.000 is AbsChWCACC, suggests that earnings quality did not suffer for
FILEDELAY 772 62.948 61.000 34.377 54.000 71.000
ChFILEDELAY 772 2.212 0.000 33.431 3.000 4.000
firms in that group. However, the positive and significant coeffi-
cient on GROUP B and GROUP C for LATE_FILER does illustrate
Panel B: Filer Group B
firms in both groups do have a higher probability of filing late,
AbsChWCACC 1513 0.050 0.031 0.061 0.013 0.066 relative to firms in Group A (i.e., firms whose prior year filing
LATE_FILER 1513 0.155 0.000 0.362 0.000 0.000 occurred before the new deadline). As noted in Table 2, firms in
PRESSURE 1513 28.369 30.000 17.626 44.000 12.000
ACCEL_FILER 1513 0.513 1.000 0.500 0.000 1.000
Group B and Group C represent a substantial percentage of AFs
SECOND 1513 0.132 0.000 0.338 0.000 0.000 and LAFs that were subject to the SEC's acceleration of 10-K filings
DELAY 1513 52.451 54.000 24.079 37.000 65.000 (approximately 75% of our sample). This finding should inform any
ChDELAY 1513 7.206 2.000 18.289 2.000 14.000 future deliberations by regulators on the topic of accelerating 10-K
FILEDELAY 1513 73.888 74.000 26.493 69.000 75.000
filing deadlines. To the extent firms have audit delays that are
ChFILEDELAY 1513 8.448 11.000 26.142 15.000 5.000
greater (less) than any new proposed deadline, the acceleration is
Panel C: Filer Group C likely (unlikely) to impair earnings quality. However, to the extent
AbsChWCACC 737 0.051 0.030 0.073 0.013 0.061 firms have file delays that are greater (less) than any new proposed
LATE_FILER 737 0.218 0.000 0.413 0.000 0.000 deadline (i.e., firms in Group B and Group C), the acceleration is
PRESSURE 737 8.647 9.000 4.450 5.000 12.000
likely (unlikely) to increase the probability of firms filing late. We
ACCEL_FILER 737 0.256 0.000 0.437 0.000 1.000
SECOND 737 0.640 1.000 0.480 0.000 1.000 further investigate the potential effects of future SEC accelerations
DELAY 737 64.697 59.000 28.499 57.000 69.000 in Section 3.2.4.
ChDELAY 737 10.749 12.000 20.113 15.000 5.000
FILEDELAY 737 70.620 60.000 30.562 60.000 75.000
ChFILEDELAY 737 6.758 13.000 30.206 15.000 6.000
3.2.3. Filer type and deadline change
AbsChWCACC is calculated as the absolute value of the change in WCACC. WCACC is Table 4 Panels A and B provides the results for the tests of
calculated as the change in working capital on past, present, and future operating
our research question: Does the association between audit time
cash flows. Change in working capital from year t-1 to year t is computed as: DAc-
counts Receivable þ DInventory e DAccounts Payable e DTaxes Payable þ DOther pressure and earnings quality differ by filer type and deadline
Assets. PRESSURE is calculated as the prior year audit delay minus the current year change? In Panel A, we include in Model (1) an indicator vari-
deadline in days. LATE_FILER is equal to one if the firm filed after the filing deadline able (ACCEL_FILER) set equal to one if the firm is an AF and zero
and zero if the firm filed on time. ACCEL_FILER is an indicator variable equal to one if if the firm is an LAF.13 We also incorporate into Model (1) an
the firm is an Accelerated Filer (AF) and zero if the firm is a Large Accelerated Filer
interaction between ACCEL_FILER and our variables related to
(LAF). SECOND is coded 1 if the firm year observation was during the second
deadline change (i.e., Dec. 2006eNov. 2007) and zero if the firm year observation time pressure (PRESSURE*ACCEL_FILER in the first regression
was during the first deadline change (i.e., Dec. 2003eNov. 2004). DELAY is calculated and GROUP B*ACCEL_FILER and GROUP C*ACCEL_FILER in the
as the number of days from fiscal year-end to the audit report date. ChDELAY is second regression). Note the coefficients on ACCEL_FILER and
calculated as the change in DELAY. FILEDELAY is calculated as the number of days
PRESSURE*ACCEL_FILER are positive and significant in the
from fiscal year-end to the filing of the annual report. ChFILEDELAY is calculated as
the change in FILEDELAY. regression where the dependent variable is AbsChWCACC (p-
value < 0.001 and p-value ¼ 0.039, respectively). This finding
suggests that PRESSURE has a more negative effect on earnings
With respect to changes in the period of time between fiscal quality for AFs (i.e., when ACCEL_FILER ¼ 1), relative to LAFs. In
year-end and the filing of the 10-K with the SEC, file delay increased addition, the coefficients on ACCEL_FILER and GROUP
slightly for Group A firms (the mean of ChFILEDELAY in Panel A is C*ACCEL_FILER are also positive and at least marginally signifi-
2.2). File delay for Group B firms was reduced by 8.4 days, which is cant (p-value < 0.001, and p-value ¼ 0.090 respectively), which
expected, as these firms had to reduce their file delay, but not suggests the negative effect on earnings quality for being in
necessarily audit delay, to meet the new deadlines. Thus, Group B Group C, relative to Group A, is greater when the firm is also an
firms primarily reduced the slack between audit delay and file AF, relative to an LAF.
delay in order to meet the new deadlines rather than reducing both Table 4, Panel A also provides results related to the probability of
audit and file delay. Group C firms reduced both audit delay (10.7 filing late. When the dependent variable is LATE_FILER, the coeffi-
days) and file delay (6.8 days) to meet the new deadlines. cient on ACCEL_FILER is positive and significant (p-value < 0.001);
however, the coefficient on PRESSURE*ACCEL_FILER is not signifi-
3.2.2. Hypothesis 1 testing cant (p-value ¼ 0.106). This illustrates that AFs (vs. LAFs) are not
Table 3 provides results of H1 testing. When the absolute value
of the change in working capital accruals (AbsChWCACC) is the
13
dependent variable, the coefficients on PRESSURE and GROUP C are We classify each firm's filer status based on its classification in Audit Analytics.
positive and significant (p's < 0.001). Thus, greater PRESSURE and Per our discussion with personnel at Audit Analytics, firms self-report their filer
status classification. When a firm does not self-report, Audit Analytics classifies the
being in GROUP C led to lower earnings quality. When LATE_FILER firm's status as unknown. In those cases, we classify an AF as a firm with a market
is the dependent variable, the coefficients on PRESSURE and Group value of equity (MVE) between $75M and $700M and a LAF as a firm with MVE
C are also positive and significant (p's < 0.001). Thus, higher greater than $700M (see footnote 1).
58 T.A. Lambert et al. / Accounting, Organizations and Society 58 (2017) 50e66

Table 3
Pressure and groups on working capital accruals and the probability of being a late filer.

Variables Predicted DV ¼ AbsChWCACC DV ¼ LATE_FILER (0 or 1)

Sign
Parameter p-value Parameter p-value Parameter p-value Parameter p-value
Estimate Estimate Estimate Estimate

INTERCEPT ? 0.0358 <0.0001 0.0290 <0.0001 2.2263 <0.0001 3.5501 <0.0001

PRESSURE + 0.0003 <0.001 0.0282 <0.0001
GROUP B ? 0.0027 0.252 1.1571 <0.0001
GROUP C + 0.0135 <0.0001 1.8678 <0.0001
ChSEASON ? 0.0138 0.473 0.0149 0.436 0.6634 0.297 0.8599 0.189
ChLOSS ? 0.0071 0.058 0.0073 0.051 0.3074 0.034 0.2987 0.040
ChCFO e 0.0258 0.415 0.0249 0.431 0.5978 0.201 0.7041 0.128
ChB/M e 0.0089 0.003 0.0088 0.004 0.1366 0.078 0.1443 0.069
ChLgMVE e 0.0202 <0.0001 0.0205 <0.0001 0.3440 0.009 0.3560 0.006
ChLEVERAGE + 0.0239 0.262 0.0249 0.242 1.0869 0.019 1.2655 0.007
ChFINANCE ? 0.0036 0.117 0.0037 0.109 0.3721 0.002 0.3563 0.003
ChDISTRESS e 0.0040 0.223 0.0038 0.239 0.0414 0.498 0.0577 0.351
ChGROWTH + 0.0037 0.225 0.0037 0.228 0.1579 0.074 0.1473 0.076
ChRESTRUCTURE ? 0.0016 0.507 0.0019 0.430 0.0235 0.868 0.0270 0.849
ChEXTRAORDINARY ? 0.0031 0.510 0.0025 0.591 0.3300 0.159 0.4290 0.079
ChACQUISITION ? 0.0057 0.017 0.0058 0.014 0.0463 0.744 0.0395 0.778
ChQUALIFIED ? 0.0023 0.266 0.0025 0.231 0.1974 0.103 0.1380 0.259
ChGOING_CONCERN ? 0.0629 0.003 0.0643 0.002 0.1747 0.685 0.3609 0.408
ChAUDITOR ? 0.0025 0.646 0.0028 0.615 1.6292 <0.0001 1.6451 <0.0001
ChLEADER ? 0.0008 0.920 0.0006 0.937 0.1533 0.838 0.0115 0.989
ChLgAUDIT_FEES ? 0.0048 0.123 0.0049 0.115 1.0093 <0.0001 0.9863 <0.0001
Year and Industry dummies Not Reported Not Reported

R2 0.146 0.146
No. of obs 3022 3022 3022 3022

This table provides the results of estimating Model (1) on the sample of all Accelerated Filers (AFs) and Large Accelerated Filers (LAFs) in the years of the filing deadline changes
(see Table 1 sample selection). Model (1) is a multivariate regression when the dependent variable is AbsChWCACC and a logistic regression when the dependent variable is
LATE_FILER. AbsChWCACC is calculated as the absolute value of the change in WCACC. WCACC is calculated as the change in working capital on past, present, and future
operating cash flows. Change in working capital from year t-1 to year t is computed as: DAccounts Receivable þ DInventory e DAccounts Payable e DTaxes Payable þ DOther
Assets. LATE_FILER is equal to one if the firm filed after the filing deadline and zero if the firm filed on time. PRESSURE is calculated as the prior year audit delay minus the
current year deadline in days. GROUP A is an indicator variable equal to one if the firm is in Group A and zero otherwise. GROUP B is an indicator variable equal to one if the firm
is in Group B and zero otherwise. GROUP C is an indicator variable equal to one if the firm is in Group C and zero otherwise. ChSEASON is coded 1 if the firm changed from
having a busy season year-end to a non-busy season year-end (or vice versa) and 0 otherwise. ChLOSS is coded 1 if the firm switched from operating at a loss in the prior year to
having net income in the current year (or vice versa) and 0 otherwise. ChCFO is calculated as the change in cash flow from operations. ChB/M is calculated as the change in book
to market ratio. ChLgMVE is the change in the natural log of market value of equity. ChLEVERAGE is calculated as the change in LEVERAGE, which is equal to total liabilities
divided by total assets. ChFINANCE is coded 1 if the number of outstanding shares increased by at least 10% or long-term debt increased by at least 20% during the current year,
but not in the prior year (or vice versa) and 0 otherwise. ChDISTRESS is calculated as the change in DISTRESS, which is calculated based on Zmijewski (1984). ChGROWTH is
calculated as the change in sales growth rate. ChRESTRUCTURE is coded 1 if the firm reported restructuring charges in either the current or prior year (but not both) and
0 otherwise. ChEXTRAORDINARY is coded 1 if the firm reported extraordinary item(s) in either the current or prior year (but not both) and 0 otherwise. ChACQUISITION is
coded 1 if the firm experienced an acquisition in either the current or prior year (but not both) and 0 otherwise. ChQUALIFIED is equal to 1 if the company's audit opinion had
an explanatory paragraph (other than going concern) in either the prior or current year (but not both) and 0 otherwise. ChGOING_CONCERN is coded 1 if the audit report
disclosed doubt about the entity's ability to continue as a going concern in either the current or prior year (but not both) and 0 otherwise. ChAUDITOR is coded 1 if an auditor
change occurred during the current year and 0 otherwise. ChLEADER is coded as 1 if the company was audited by a national industry leader in the prior year or current year
(but not both) and 0 otherwise. ChLgAUDIT_FEES is calculated as the change in the natural log of audit fees. All variables are winsorized at the 1st and 99th percentiles and p-
values are one-tailed when there are hypothesized directional expectations. All other p-values are two-tailed. Variables of interest are in bold.

differentially affected by PRESSURE in terms of whether or not they
file late. The coefficients on ACCEL_FILER in the last regression is
positive and significant (p-value ¼ 0.053) while the coefficient on
GROUP C*ACCEL_FILER is not significant (p-value ¼ 0.209). We
observe that AFs (vs. LAFs) are not differentially affected by being in
Group C relative to Group A in terms of whether or not they file late.
Panel B of Table 4 provides results of tests for differences be-
tween the first deadline change (affecting AFs and LAFs) and the
second deadline change (only affecting LAFs). In Panel B, we include
in Model (1) an indicator variable (SECOND) set equal to one if the
firm observation is part of the second deadline change and zero if
the observation is a part of the first deadline change. We also
incorporate into Model (1) an interaction between SECOND and our
variables related to time pressure (PRESSURE*SECOND in the first
regression and GROUP B*SECOND and GROUP C*SECOND in the
second regression). When the dependent variable is AbsChWCACC,
consistent with Table 3 the coefficient on PRESSURE in the first
regression is positive and significant (p-value ¼ 0.001). However,
the coefficient on SECOND is negative and significant (p-
value < 0.001), which suggests firms in the first deadline change
had lower earnings quality relative to firms in the second deadline
change.14 The coefficient on PRESSURE*SECOND is not significant
(p-value ¼ 0.747), which indicates the negative effect that PRES-
SURE has on earnings quality is not different between the first and
second deadline. Alternatively, the coefficient on GROUP C*SECOND
is negative and significant (p-value ¼ 0.006), which illustrates the
negative effect on earnings quality for being in Group C is smaller
during the second deadline change, relative to the first deadline
change. This suggests that audit firms may have learned from the
first deadline change to identify best practices to employ during the
second deadline (see section 4.2.3 for an analysis of best practices
and lessons learned).
14
Untabulated results of regressions including SECOND without the interaction
terms also reveal a negative, significant coefficient on SECOND for all models pre-
sented in Panel B of Table 4 (all p-values  0.070). This also implies that firms
exhibited lower earnings quality during the first accelerated filing time period than
they did during the second time period. Results for our independent variables in
these regressions (i.e., PRESSURE, GROUP B, and GROUP C) remain substantially
unchanged from those reported in Table 3.
Table 4
Pressure and groups on working capital accruals and the probability of being a late filer.

Variables Predicted DV ¼ AbsChWCACC DV ¼ LATE_FILER (0 or 1)

Sign
Parameter p-value Parameter p-value Parameter p-value Parameter p-value
Estimate Estimate Estimate Estimate

Panel A: Filer Type

INTERCEPT ? 0.0342 <0.0001 0.0316 <0.0001 2.3627 <0.0001 3.2634 <0.0001

PRESSURE + 0.0001 0.222 0.0358 <0.0001
GROUP B ? 0.0016 0.460 0.4467 0.113
GROUP C + 0.0063 0.012 1.4519 <0.0001
ACCEL_FILER ? 0.0267 <0.0001 0.0170 <0.0001 1.2566 <0.0001 0.6829 0.053
PRESSURE*ACCEL_FILER ? 0.0003 0.039 ¡0.0121 0.106
GROUP B*ACCEL_FILER ? 0.0000 0.993 1.0438 0.008
GROUP C*ACCEL_FILER ? 0.0152 0.090 0.5260 0.209
ChSEASON ? 0.0135 0.463 0.0151 0.402 0.7072 0.334 0.9667 0.195
ChLOSS ? 0.0063 0.086 0.0063 0.085 0.2637 0.073 0.2788 0.062
ChCFO e 0.0256 0.410 0.0248 0.423 0.6151 0.148 0.7022 0.094

T.A. Lambert et al. / Accounting, Organizations and Society 58 (2017) 50e66

ChB/M e 0.0093 0.002 0.0092 0.002 0.1301 0.074 0.1307 0.089
ChLgMVE e 0.0187 <0.0001 0.0190 <0.0001 0.2526 0.031 0.2902 0.013
ChLEVERAGE + 0.0227 0.281 0.0232 0.267 1.0940 0.009 1.2203 0.004
ChFINANCE ? 0.0043 0.055 0.0043 0.059 0.3329 0.007 0.3219 0.009
ChDISTRESS e 0.0040 0.211 0.0039 0.225 0.0390 0.468 0.0523 0.337
ChGROWTH + 0.0033 0.277 0.0033 0.284 0.1284 0.107 0.1243 0.114
ChRESTRUCTURE ? 0.0012 0.604 0.0013 0.575 0.0360 0.804 0.0326 0.823
ChEXTRAORDINARY ? 0.0019 0.684 0.0016 0.725 0.4261 0.083 0.4641 0.065
ChACQUISITION ? 0.0054 0.021 0.0055 0.018 0.0523 0.718 0.0485 0.737
ChQUALIFIED ? 0.0025 0.226 0.0028 0.183 0.2031 0.100 0.1625 0.195
ChGOING_CONCERN ? 0.0582 0.005 0.0581 0.005 0.0049 0.990 0.1428 0.723
ChAUDITOR ? 0.0016 0.770 0.0015 0.783 1.6121 <0.0001 1.6860 <0.0001
ChLEADER ? 0.0018 0.811 0.0025 0.749 0.1922 0.779 0.0531 0.944
ChLgAUDIT_FEES ? 0.0055 0.077 0.0058 0.063 1.0539 <0.0001 1.0258 <0.0001
Year and Industry dummies Not Reported Not Reported
R2 0.158 0.159
No. of obs 3022 3022 3022 3022

Panel B: Deadline Change

INTERCEPT ? 0.0532 <0.0001 0.0401 <0.0001 1.6040 <0.0001 3.4722 <0.0001

PRESSURE + 0.0003 0.001 0.0251 <0.0001
GROUP B ? 0.0049 0.113 1.4121 <0.0001
GROUP C + 0.0221 <0.0001 1.9391 <0.0001
SECOND ? ¡0.0173 <0.0001 ¡0.0045 0.053 ¡0.8557 <0.0001 0.5650 0.136
PRESSURE*SECOND ? 0.0000 0.747 0.0493 0.001
GROUP B*SECOND ? ¡0.0075 0.067 ¡1.8325 <0.001
GROUP C*SECOND ? ¡0.0181 0.006 ¡0.6394 0.128
ChSEASON ? 0.0138 0.473 0.0155 0.419 0.6464 0.302 0.8623 0.199
ChLOSS ? 0.0071 0.057 0.0074 0.048 0.2946 0.073 0.3117 0.034
ChCFO e 0.0258 0.415 0.0245 0.437 0.5970 0.148 0.7248 0.119
ChB/M e 0.0090 0.003 0.0089 0.004 0.1395 0.074 0.1411 0.077
ChLgMVE e 0.0202 <0.0001 0.0204 <0.0001 0.3437 0.031 0.3582 0.006
ChLEVERAGE + 0.0239 0.262 0.0250 0.236 1.0693 0.009 1.2645 0.006
ChFINANCE ? 0.0036 0.118 0.0036 0.113 0.3695 0.007 0.3553 0.004
ChDISTRESS e 0.0040 0.223 0.0038 0.244 0.0414 0.468 0.0599 0.336
ChGROWTH + 0.0037 0.225 0.0036 0.234 0.1551 0.107 0.1461 0.078
ChRESTRUCTURE ? 0.0016 0.508 0.0018 0.463 0.0305 0.804 0.0311 0.828
ChEXTRAORDINARY ? 0.0031 0.508 0.0029 0.529 0.3399 0.083 0.4056 0.093
ChACQUISITION ? 0.0056 0.017 0.0057 0.016 0.0230 0.718 0.0350 0.803
ChQUALIFIED ? 0.0023 0.266 0.0027 0.198 0.1994 0.100 0.1320 0.283
ChGOING_CONCERN ? 0.0628 0.003 0.0621 0.003 0.2268 0.990 0.4095 0.347

59
(continued on next page)
 Table 4 (continued )

Variables Predicted DV ¼ AbsChWCACC DV ¼ LATE_FILER (0 or 1) 60

Sign
Parameter p-value Parameter p-value Parameter p-value Parameter p-value
Estimate Estimate Estimate Estimate

ChAUDITOR ? 0.0025 0.644 0.0029 0.602 1.6124 <0.0001 1.6688 <0.0001

ChLEADER ? 0.0008 0.916 0.0014 0.865 0.1498 0.779 0.0165 0.984
ChLgAUDIT_FEES ? 0.0048 0.122 0.0053 0.087 0.9862 <0.0001 0.9818 <0.0001
Year and Industry dummies Not Reported Not Reported
R2 0.142 0.144
No. of obs 3022 3022 3022 3022

This table provides the results of estimating Model (1) on the sample of all Accelerated Filers (AFs) and Large Accelerated Filers (LAFs) in the years of the filing deadline changes (see Table 1 sample selection). Model (1) is a
multivariate regression when the dependent variable is AbsChWCACC and a logistic regression when the dependent variable is LATE_FILER. AbsChWCACC is calculated as the absolute value of the change in WCACC. WCACC is
calculated as the change in working capital on past, present, and future operating cash flows. Change in working capital from year t-1 to year t is computed as: DAccounts Receivable þ DInventory e DAccounts Payable e DTaxes
Payable þ DOther Assets. LATE_FILER is equal to one if the firm filed after the filing deadline and zero if the firm filed on time. PRESSURE is calculated as the prior year audit delay minus the current year deadline in days. GROUP A
is an indicator variable equal to one if the firm is in Group A and zero otherwise. GROUP B is an indicator variable equal to one if the firm is in Group B and zero otherwise. GROUP C is an indicator variable equal to one if the firm is
in Group C and zero otherwise. ACCEL_FILER is an indicator variable equal to one if the firm is an Accelerated Filer (AF) and zero if the firm is a Large Accelerated Filer (LAF). SECOND is coded 1 if the firm year observation was
during the second deadline change (i.e., Dec. 2006eNov. 2007) and zero if the firm year observation was during the first deadline change (i.e., Dec. 2003eNov. 2004). ChSEASON is coded 1 if the firm changed from having a busy
season year-end to a non-busy season year-end (or vice versa) and 0 otherwise. ChLOSS is coded 1 if the firm switched from operating at a loss in the prior year to having net income in the current year (or vice versa) and
0 otherwise. ChCFO is calculated as the change in cash flow from operations. ChB/M is calculated as the change in book to market ratio. ChLgMVE is the change in the natural log of market value of equity. ChLEVERAGE is
calculated as the change in LEVERAGE, which is equal to total liabilities divided by total assets. ChFINANCE is coded 1 if the number of outstanding shares increased by at least 10% or long-term debt increased by at least 20%
during the current year, but not in the prior year (or vice versa) and 0 otherwise. ChDISTRESS is calculated as the change in DISTRESS, which is calculated based on Zmijewski (1984). ChGROWTH is calculated as the change in
sales growth rate. ChRESTRUCTURE is coded 1 if the firm reported restructuring charges in either the current or prior year (but not both) and 0 otherwise. ChEXTRAORDINARY is coded 1 if the firm reported extraordinary item(s)
in either the current or prior year (but not both) and 0 otherwise. ChACQUISITION is coded 1 if the firm experienced an acquisition in either the current or prior year (but not both) and 0 otherwise. ChQUALIFIED is equal to 1 if the
company's audit opinion had an explanatory paragraph (other than going concern) in either the prior or current year (but not both) and 0 otherwise. ChGOING_CONCERN is coded 1 if the audit report disclosed doubt about the
entity's ability to continue as a going concern in either the current or prior year (but not both) and 0 otherwise. ChAUDITOR is coded 1 if an auditor change occurred during the current year and 0 otherwise. ChLEADER is coded as
1 if the company was audited by a national industry leader in the prior year or current year (but not both) and 0 otherwise. ChLgAUDIT_FEES is calculated as the change in the natural log of audit fees. All variables are winsorized
at the 1st and 99th percentiles and p-values are one-tailed when there are hypothesized directional expectations. All other p-values are two-tailed. Variables of interest are in bold.

15
filing the 10-K late.

filing deadlines by 15 days.

T.A. Lambert et al. / Accounting, Organizations and Society 58 (2017) 50e66

3.2.4. Implications for the future

our evidence suggests that earnings quality will suffer.

quality earnings being supplied to market participants.
technology advances and users demand access to information on a
The Wall Street Reform and Consumer Protection Act (Dodd-
of filing late. When the dependent variable is LATE_FILER, the
Table 4, Panel B also provides results related to the probability

As noted previously and described in Fig. 1, the SEC has twice accelerated 10-K
of current slack for NAFs (mean ¼ 0.7 days) and AFs (mean ¼ 0.2
category of filer.15 As we observe above, any future accelerations
To assess the potential impact of future reductions in 10-K filing

higher than our current sample of Group C firms (note mean

of slack (days between audit report date and filing date) for NAFs
audit report date and potential new filing deadline) and the amount
described in Fig. 2 and in our analyses). Table 5 also presents
deadlines, we examine the potential PRESSURE that AFs and NAFs
more likely to have lower earnings quality). Thus, future accelera-
are likely more troublesome during accelerations (e.g., they are
larly in the cases of AFs and NAFs. Our results in Section 3.2.3
regulators) will propose future 10-K filing accelerations, particu-
that the SEC desires for these firms to face a shorter deadline. As
relation between audit time pressure and earnings quality (tested

nally slated by the SEC to have a 60-day deadline, which indicates

(NAFs) from Section 404(b) of the Sarbanes-Oxley Act, which clears

(e.g., a reduction from 90 to 75 days). In addition, AFs were origi-

the way for the SEC to impose filing deadline reductions on NAFs
Frank) created a permanent exemption for non-accelerated filers
predominately to our measure of accruals and, to a lesser extent,
change (which affected both AFs and LAFs). Our conclusion relates
did not affect the probability of filing late, relative to firms in
line. Finally, the coefficients on SECOND and SECOND*GROUP C are
the first deadline change. However, rising PRESSURE minimizes

days) suggests audit delay is the barrier to filing earlier and that
deadline. Mean pressure in days is 8.6 and 9.5 for NAFs and AFs,
descriptive statistics for the amount of pressure (days between
audit delays greater than 60 days. Thus, the audits of a majority of
Based on 2014 filing data, in Table 5 we illustrate that 56% of
celerations induce time PRESSURE on the audits of AFs and NAFs,
earnings quality. On the other hand, to the extent that future ac-
that do not induce PRESSURE on the auditor are not likely to impact
might experience, assuming a future 15-day reduction for each
tions of 10-K deadlines for smaller NAFs or AFs could lead to lower
indicate that, in general, the audits of smaller (vs. larger) firms

and AFs whose audit delay is greater than the new potential filing
deadline reduction (they would be considered Group C firms as
NAFs had audit delays greater than 75 days and 78% of AFs had
by H1) was more acute for AFs and during the first deadline
which suggests that being in GROUP C during the second deadline
the reduction in the probability of filing late in the second dead-
value < 0.001). These findings imply that being in the second
although the coefficient on SECOND is negative and significant (p-
coefficients on PRESSURE and PRESSURE*SECOND are positive and

Group A. Overall, our results support the notion that the negative
significant (p-value < 0.001 and p-value ¼ 0.001 respectively)

PRESSURE for Group C firms in Table 3, Panel C ¼ 8.6 days). The lack
respectively. Thus, mean pressure for these firms would be slightly
NAFs and AFs would be substantially affected by another 15-day
timelier basis, it is reasonable to assume the SEC (and other foreign
not significant (p-value ¼ 0.136 and p-value of 0.128 respectively),
deadline change decreases the probability of filing late relative to
 T.A. Lambert et al. / Accounting, Organizations and Society 58 (2017) 50e66 61

Table 5
Analysis of pressure and slack for NAFs and AFs in fiscal year 2014.

Non Accelerated Filers (NAFs)

NAFs filing between 75 and 90 days (N ¼ 921 or 56% of all NAFs that filed on time) Mean Median Std Dev

Pressure
Days between Audit Report Date and Potential New Deadline (i.e., 75 days) 8.6 10 4.84
Slack
Days between Audit Report Date and Filing Date 0.7 0.0 1.73

Accelerated Filers (AFs)

AFs filing between 60 and 75 days (N ¼ 1003 or 78% of all AFs that filed on time) Mean Median Std Dev

Pressure
Days between Audit Report Date and Potential New Deadline (i.e., 60 days) 9.5 11 4.73
Slack
Days between Audit Report Date and Filing Date 0.2 0.0 0.74

Data includes all NAFs and AFs in Audit Analytics that filed a 10-K with a 2014 fiscal year-end. Firms self-report their filer status. We classify a NAF as a firm with a market value
of equity (MVE) of less than $75M and an AF as a firm with a market value of equity between $75M and $700M.

these audits are currently being performed as quickly as possible
(to limit filing delays). In short, these data suggest that a 15-day
acceleration of 10-K filing deadlines for NAFs and AFs would
require large, pressure-induced reductions in audit delay. Our study
finds that such reductions lead to reductions in the quality of
earnings provided to financial statement users.16
Select ONE fiscal year-end audit for which you served as an audit
partner and that was required to accelerate its 10-K filings be-
tween the years of 2003e2004 or 2006-2007. You will be asked
Table 6
Survey descriptive statistics.

4. Survey analysis Participant Variables N Mean Standard Deviation

SERVED 32 100.00 N/A

4.1. Method ClientEXPERIENCE 31 2.90 1.38
IndEXPERIENCE 32 19.13 8.29
PercTIME 32 42.91 23.33
In order to obtain a richer understanding of the effects of
PercFAC 32 73.36 24.03
accelerating 10-K deadlines on audit/earnings quality, we con- PercBIGFOUR 32 100.00 N/A
ducted a survey that was completed by thirty-two retired Big Four PercLARGEOFFICE 32 53.13 N/A
audit partners who served as a partner on either an AF or LAF OtherACCELERATORS 30 3.17 1.60
during the years of acceleration (see Appendix C of the Online Audit Variables N Mean Standard Deviation
Appendix for the full survey).17 As recommended by Bloomfield,
PercLAF 32 71.88 N/A
Nelson, and Soltes (2016), we use our survey to provide context PercFIRST 32 59.37 N/A
for our archival results and to suggest directions for the develop- Perc10-KEARLY 32 93.75 N/A
ment of new theory. We use retired audit partners as our partici- PercAUDITEARLY 32 71.88 N/A
pants for two reasons. First, all of them were senior partners most DaysPRESSURE 31 8.53 7.56
ScalePRESSURE 32 7.34 2.01
responsible for, and knowledgeable about, audit quality during the
PercREV>$10B 32 37.49 N/A
10-K acceleration time period (between 2003 and 2007), which PercSEASON 32 87.50 N/A
allows our participants to provide informed responses on matters
SERVED is the percent of participants that indicated that they served as a partner on
that are hard to observe using other methods. Second, we expect an accelerated or large accelerated filer during one of the acceleration periods.
retired partners to be much more likely to provide candid re- ClientEXPERIENCE is the participant's response to: At the time of the 10-K filing
sponses (vs. active partners presently employed by the firms). deadline change, for how many years had you served on the audit engagement that
Survey participants were given the following instructions: you chose”? IndEXPERIENCE is the response to the following question: At the time
of the acceleration, for how many years had you served companies in the client's
industry? PercTIME is the response to: Approximately what percentage of your time
(measured on an annual basis) did you work on this client? PercFAC is the response
16
Data from our survey of audit partners (described in Section 4) is consistent to: Approximately what percentage of your hours charged to this client did you
with the notion that future accelerations for NAFs and AFs could impair audit spend at the client's facilities? PercBIGFOUR is the percent of respondents that
quality. We asked respondents to rate the potential impact on audit quality worked for a Big Four firm. PercLARGEOFFICE is the percent of respondents that
(1 ¼ Very Negative Impact, 5 ¼ No Impact, and 10 ¼ Very Positive Impact) if the 10- indicated that they worked for a large (>500 Professionals) office. OtherACCELER-
K filing deadlines for NAFs and AFs were reduced by 15 days (as described above). ATORS is the response to: How many different clients did you have that were
For both NAFs and AFs, seventy-seven percent of respondents provided a response required to accelerate the filing of their 10-K? PercLAF is the percent of respondents
below five (on the “Very Negative Impact” side of the scale). Mean responses for that selected the audit of a Large Accelerated Filer (vs. Accelerated Filer) to provide
NAFs and AFs (3.25 and 3.81, respectively) were also significantly lower than the data. PercFIRST is the percent of respondents that selected an audit affected by the
mid-point of five (all p-values  0.010). first acceleration (vs. the second acceleration). Perc10-KEARLY is the percent of
17
Obtaining our sample began with contacting current and retired national firm audits selected where the 10-K filing date had to be accelerated to meet the new
professionals at all of the Big Four accounting firms. We asked these professionals deadline. PercAUDITEARLY is the percent of audits selected where the audit report
to provide our online (Qualtrics) survey link directly to ten to twenty retired date had to be accelerated to meet the new deadline. DaysPRESSURE is the response
partners that served on an AF or LAF. The link was distributed to the retired to the following: Please recall the approximate number of days that the audit report
partners by the national firm professionals via an e-mail message provided by the date was accelerated. ScalePRESSURE is the response to: In the year of acceleration,
authors. Forty-nine partners started the survey. Four participants did not serve on the level of time pressure on the audit team was (measured on a scale where
an AF or LAF during the acceleration time period and were exited from the survey. 1 ¼ Very Low and 10 ¼ Very High). PercREV>$10B is the percent of audits selected
Thirteen participants did not complete the survey. Our final sample size of thirty- where the client's revenues exceeded $10 Billion. PercSEASON is the percent of
two is comparable to previous studies that have attempted to obtain qualitative audits selected that had a fiscal year-end either in December or January. When
data from audit professionals (e.g., Hirst & Koonce, 1996; Trompeter & Wright, presenting a standard deviation does not provide any value (e.g., SERVED, Perc-
2010). BIGFOUR) we denote this with “N/A”.
62 T.A. Lambert et al. / Accounting, Organizations and Society 58 (2017) 50e66
to recall aspects of that audit during the year of acceleration.
Therefore, if you served more than one client that accelerated its
10-K filing in response to SEC rules 33-8128 and 33-8644, please
(if possible) select the audit that:
 You can best recall.
 Required the 10-K to be filed earlier than in the previous year, in
order to meet the new accelerated 10-K filing
 Required that audit report be dated earlier than in previous year,
in order for the client to meet the new accelerated 10-K filing
deadline.18
The survey asked participants a series of open-ended questions
concerning the topics that may or may have not been an issue
during their selected audit (e.g., engagement team morale,
resolving audit adjustments).19 In addition to obtaining the afore-
mentioned qualitative data, the survey elicited quantitative data
related to time pressure on the audit, measures of audit/earnings
quality, strategies employed to maintain audit quality, de-
mographic data corresponding to the participant and the audit
selected, and other miscellaneous questions. We use these data to
identify the strategies used to maintain an acceptable level of audit
quality during accelerations (i.e., best practices and lessons
learned) and to provide additional insight into our archival results.
Participants took, on average, 42.07 minutes to complete the survey
instrument.20
Descriptive statistics for the participants and the audits they
selected are presented in Table 6. Of particular note is that partic-
ipants devoted a high percentage of their time to the audits they
selected (PercTIME and PercFAC) and they had several other clients
that were required to accelerate their 10-K filing as well (Other-
ACCELERATORS). Thus, these participants were likely aware of the
specific audit issues, strategies, and outcomes associated with their
selected audit and the general dynamics of auditing under accel-
erated conditions. The majority of clients selected by our re-
spondents were LAFs and were audits where both the 10-K (Perc10-
KEARLY) and the audit report date (PercAUDITEARLY) had to be
accelerated to meet the new deadline. Therefore, as one might
expect, participants reported a fair amount of time pressure on the
audits they selected (DaysPRESSURE and ScalePRESSURE). While
not tabulated, participants selected audit clients from a variety of
industries, with the most common being manufacturing and in-
dustrial products (n ¼ 12).
4.2. Results
4.2.1. Perceptions related to time pressure and audit quality
Our survey analysis provides quantitative and qualitative evi-
dence about audit partners’ experiences with the acceleration of
10-K filing deadlines. When we asked our survey respondents if
they experienced any additional time pressure during the audit
they selected, one respondent noted the following:
Shortening the amount of time to complete the audit by 20 days
added extreme pressure. Not only because the end date was 20
days quicker but we were expected to be substantially complete
18
Because our analyses focus on firms that accelerated both the 10-K and audit
report (Group C firms in Fig. 2), we asked participants to recall an audit that would
be classified as a Group C audit.
19
We include the exact questions in Appendix D of our Online Appendix, as well
as additional, substantive responses not included in this study's text.
20
When calculating this average, we exclude three participants who took over
2 hours to complete the instrument and most likely did not complete the survey at
one sitting.
well in advance of the filing date in order to meet with the audit
committee on audit issues and findings. Because the audit
committee needed our views in advance of the audit committee
meeting, that shortened the time even more for us to do our
work.
Our survey data confirms that PRESSURE used in our archival
analysis represents an adequate proxy for the time pressure
perceived by the audit team. A non-tabulated analysis reveals a
significant, positive correlation (p-value ¼ 0.023) between Day-
sPRESSURE and ScalePRESSURE (see variables in Table 6).
In Section 2.3 we describe how time pressure induced by 10-K
accelerations could affect the following factors: the audit adjust-
ment process, testing accounts at year-end, professional skepti-
cism, and morale/effort. We asked respondents to comment on
these factors and obtained the following:
It was difficult to align client personnel schedules to make
certain the audit firm personnel understood the client's ac-
counting for complex transactions and confer about potential
audit adjustments.
The entire audit support system was built on the basis of a 90
day year-end close. So attorneys, actuaries, valuation specialists,
audit committee members all had to adjust their schedules
forward and, frankly, many of them did not which caused the
last two weeks of the audit to be very difficult.
I was very concerned about the risk that long hours might
adversely impact the degree of professional skepticism main-
tained by the staff. Our auditors were very busy and they
recognized that pushing the client for more answers in areas
being audited today would only delay the client's delivery of
schedules needed for audit areas scheduled to be started
tomorrow.
It caused significant morale issues due to increased time pres-
sures on multiple client audit closings. It got to a point where
our firm executives were working/spending significant amounts
of time to persuade Generation X & Y professionals to “hang in
there” and not quit.
Survey responses also underscored how time pressure imposes
challenges on receiving and evaluating complex valuations (e.g.,
derivatives, pensions, and goodwill) as follows:
The audit required involvement by specialists in the areas of
pensions and derivatives as well as an SEC reviewing partner.
Those specialists were exceptionally busy when I needed them,
given the accelerated filing timeframes impacting so many cli-
ents of the firm.
It was difficult for the audit team to get comfortable with the
valuation of derivatives, due to the (then) size of our firm's
available pool of experts and tools for valuation of swaps, collars,
etc.
Valuation work put significant stress on the client's internal
valuation team to gather the required valuation support. This
led to erosion in relationships with some of the client's valua-
tion people.
Receipt of third party valuation, actuarial, and goodwill
impairment assessments were completed closer to reporting
deadlines.
Valuations at year-end were critical and the time required for
validation of them put us in a squeeze to meet earlier signoff.
 T.A. Lambert et al. / Accounting, Organizations and Society 58 (2017) 50e66 63

Table 7
Best practices.

Strategies Not Attempted Not At All (1,2,3,4) Middle (5 and 6) To A Great Extent (7,8,9,10) Response Mean H0: Avg Rating ¼ 5.74

1. MoreHOURS 3 6 2 21 7.24 ***

2. InterimTESTING 2 4 5 21 6.93 ***
3. RescheduleAUDITS 3 6 7 16 6.48 *
4. InternalCONTROLS 4 5 11 12 6.21
5. MoreSENIORITY 8 5 8 11 6.17
6. ClientCLOSE 4 7 7 14 6.11
7. ConsultNATIONAL 4 5 13 10 6.00
8. OmitlowvaluePROCEDURES 4 8 10 10 5.96
9. ItAUDIT 8 7 9 8 5.79
10. MoreSTAFF 9 6 9 8 5.65
11. IA 4 7 13 8 5.46
12. IncreasingFEE 6 10 9 7 5.42
13. OUTSOURCE 16 4 9 3 5.38
14. IndEXPERIENCE 9 7 10 6 5.17
15. OtherSPECIALISTS 10 8 10 4 5.09
16. CAATS 5 10 12 5 4.96 *

Thirty-two survey participants were asked to indicate, for the audit they selected, how much each of the above strategies enhanced their audit team's ability to maintain
an acceptable level of audit quality in the year of the acceleration. Participants responded on a scale where 1 ¼ Not At All and 10 ¼ To A Great Extent. If the strategy was
not attempted, participants could indicate that the strategy was not attempted. Columns 2-5 present the number of participants responding not attempted, on the “Not
At All” side of the scale (1-4), in the middle of the scale (5 and 6), and on the “To A Great Extent” side of the scale (7-10). The mean response for the sample across all
participants and strategies (excluding observations where a strategy was not attempted) is 5.74. The final column reports the results of a t-test of the null hypothesis
that each strategy's average response is equal to the mean response for the entire sample (5.74). *** and * denote rejection at the 1% and 10% levels, respectively. The
strategies were measured with the following prompts: MoreHOURS ¼ Expanding the number of hours worked per day/week. InterimTESTING ¼ Performing additional
audit procedures before fiscal year-end (interim testing). RescheduleAUDITS ¼ Rescheduling the audits of non-public clients in your office to allocate human resources
to accelerated and large accelerated filer audits. InternalCONTROLS ¼ Additional testing and reliance on your client's internal controls over financial reporting.
MoreSENIORITY ¼ Increasing the seniority of the professionals assigned to the engagement. ClientCLOSE ¼ Requesting the client to modify their year-end closing
procedures to provide for additional audit time. ConsultNATIONAL ¼ Consulting with the national office on accounting and auditing issues.
OmitlowvaluePROCEDURES ¼ Omitting “low value” audit procedures that were performed in the prior year. ItAUDIT ¼ Increasing the number of IT audit specialists on
the engagement team. MoreSTAFF ¼ Increasing the number of audit professionals on the engagement team, regardless of industry expertise. IA ¼ Increasing reliance on
your client's internal audit function. IncreasingFEE ¼ Increasing the audit fee to deal with the SEC mandated acceleration. OUTSOURCE ¼ Outsourcing audit work to a
foreign country (i.e., offshoring). IndEXPERIENCE ¼ Increasing the number of audit professionals on the engagement team with expertise in the industry of your client.
OtherSPECIALISTS ¼ Increasing the number of other specialists (other than IT) on the engagement team. CAATS ¼ Increasing the use of computer assisted audit
techniques (e.g., IDEA) on the audit.

Consistent with our archival analysis that implies the audits of
LAFs were less affected by accelerations, when we asked our survey
respondents if they felt that the audits of LAFs or AFs were better
able to handle any time pressures caused by accelerations, eighty-
seven percent of our respondents indicated the audits of LAFs were
better equipped. When we asked why this was the case, we ob-
tained the following:
Typically, more established companies, longer history of
reporting, identical IT platforms, more experienced and more
accounting staffs, stronger internal controls. Also, more estab-
lished boards and experienced audit committees.
4.2.2. Further investigation of the time pressure and earnings
quality relation
In an attempt to pinpoint why we observe the time pressure and
earnings quality relations as described in Table 3, we analyze data
obtained from our survey participants. We obtained three mea-
sures of time pressure for the audits selected by our participants,
DaysPRESSURE (similar to PRESSURE), PercAUDITEARLY (similar to
GROUP C) and ScalePRESSURE (a more subjective assessment of
time pressure; see Table 6 for variable definitions). We also
measured several process and outcome variables related to audit/
earnings quality (e.g., the effectiveness of year-end audit proced-
ures, overall level of audit quality). Survey respondents were asked
to rate these process and outcome variables for the year of accel-
eration (when compared to their average client that year) on a scale
ranging from 1 ¼ Lower to 10 ¼ Higher. In three non-tabulated
ordinal regressions, controlling for the variables listed in Table 6,
we observe that DaysPRESSURE, PercAUDITEARLY, and Scale-
PRESSURE are all positively associated with the level of difficulty
associated with resolving audit adjustments (all p-
values < 0.019).21 In short, our survey data delves deeper into the
H1 relation and provides evidence that greater time pressure on the
auditor significantly increases the complexity of resolving financial
statement adjustments identified during the audit.
4.2.3. Best practices and lessons learned
Given that future accelerations will likely induce substantial
time pressure on audits, we use our survey data to identify stra-
tegies that enhanced the ability of audit teams to maintain an
acceptable level of audit quality during 10-K accelerations. As such,
we attempt to identify some best practices, and also some strate-
gies that may be less effective. Table 7 lists, in order of mean re-
sponses, the sixteen potential strategies we examined (e.g.,
expanding the number of hours worked per day/week, Mor-
eHOURS). Participants were asked how much each of the strategies
enhanced their audit team's ability to maintain an acceptable level
of audit quality in the year of the acceleration and responded on a
scale where 1 ¼ Not At All and 10 ¼ To A Great Extent. If the strategy
was not attempted, participants could denote this fact instead of
providing a response to the scale. For each strategy, Columns 2-5 of
Table 7 present the number of participants responding: not
attempted, on the “not at all” side of the scale (1-4), in the middle of
the scale (5 and 6), and on the “to a great extent” side of the scale
(7-10), respectively. The mean response for the sample across all
participants and strategies (excluding observations where a strat-
egy was not attempted) is 5.74. The final column reports the results
of a t-test of the null hypothesis where each strategy's average
21
No other process or outcome variables were significantly associated with all
three of our survey's measures of time pressure.
64 T.A. Lambert et al. / Accounting, Organizations and Society 58 (2017) 50e66
response is equal to the mean response for the entire sample across
all participants and strategies (5.74). Note in the final column that
the means for MoreHOURS, InterimTESTING (performing addi-
tional audit procedures before fiscal year-end), and Reschedu-
leAUDITS (rescheduling the audits of non-public clients) are
significantly greater than 5.74 and the mean for CAATS is signifi-
cantly lower than 5.74.
Some items of particular importance are illustrated in Table 7.
First, working more hours, performing more interim testing, and
rescheduling the audits of non-public companies to better allocate
human resources to AFs and LAFs appear to be perceived as the
most effective strategies (best practices). We also note the vast
majority of audits in our sample employed these techniques (note
the low frequencies in the “Not Attempted” column for these
strategies). Second, consistent with the notion that simply adding
additional human resources would be an ineffective means of
dealing with 10-K accelerations, we observe that ItAUDIT
(increasing the number of IT audit specialists), MoreSTAFF
(increasing the number of audit professionals on the engagement
team, regardless of industry expertise), IndEXPERIENCE (increasing
the number of audit professionals on the engagement team with
expertise in the industry of the client), and OtherSPECIALISTS
(increasing the number of other non-IT specialists on the engage-
ment team) are all ranked in the bottom half of strategies. Also, all
four of these strategies were not attempted by a large number
(between 8 and 10) of our survey respondents.
Third, while increasing the audit fee to deal with the 10-K ac-
celeration was attempted by most participants, more participants
viewed the strategy on the “Not At All” side (10 participants) than
the “To A Great Extent” side of the scale (7 participants). The pre-
viously mentioned human resource constraints may have made
additional billing ineffective because few additional human re-
sources could be acquired with the additional fees. Fourth,
outsourcing or off-shoring of selected audit tasks was not used on
sixteen engagements in our sample. This may be due to many
participants viewing the strategy as less effective (note the rank of
the strategy is 13th) or that, at the time of the 10-K accelerations,
outsourcing was a less predominant practice (though it may be a
means of relieving time pressure in the future). The low ranking
could also indicate that year-end procedures are not the type of
procedures that can be outsourced. Fifth, the use of computer
assisted audit techniques (CAATS) was used by the vast majority of
our respondents, but was deemed the least effective strategy.
CAATS was the only strategy with a mean response that was
significantly below the overall sample mean of 5.74 and ten par-
ticipants responded on the “Not At All” side of the scale (vs. five
were on the other side of the scale). Clearly, given the role IT sys-
tems (e.g., SAP, Oracle) play in the production of financial state-
ments (e.g., Brazel & Dang, 2009), developing more effective CAATS
and data analytics to alleviate time pressure represents a sub-
stantial opportunity for audit firms. Last, we received the following
response when we asked participants if they employed any addi-
tional strategies to effectively deal with the time pressure caused
by accelerations:
I don't know that there were ways to effectively deal with the
time pressure other than to try to increase the “esprit de corps”
through catering lunches and dinners, having more team
meetings, making sure as the partner that you personally
thanked the team members regularly and openly and trying to
get as many out of the client's office as possible for some per-
sonal time whenever possible.
In sum, these quantitative and qualitative data from our survey
analysis should inform audit teams charged with handling future
events or regulatory actions that place greater year-end time
pressure on the audit team.22
4.2.4. Exploratory analysis of survey data
Given our unique dataset of survey responses under PRESSURE,
we perform an analysis examining bivariate correlations among our
quantitative survey data (described in Tables 6 and 7). Given the
exploratory nature of this analysis, future research can study the
relations discussed herein using more focused and rigorous
methods. Upon review of the correlations (not tabulated), two
interesting patterns emerge from the survey data. In our setting,
the benefit derived from shifting audit testing from year-end to
interim (best practice InterimTESTING in Table 7) and the per-
centage of the partner's (respondent's) time working on the client
(PercTIME in Table 6) appear to be crucial factors. For the engage-
ments recalled by respondents, we observe that the value obtained
from InterimTESTING to be positively associated with the effec-
tiveness of interim and year-end procedures, as well as the overall
audit and financial reporting quality perceived by the participant.
InterimTESTING also appears to reduce the difficulty associated
with resolving audit adjustments at year-end (two-tailed p-
values < 0.10).
PercTIME is positively associated with client size (as one would
expect) and appears to have a positive impact on all of our process
and outcome variables related to audit/earnings quality (e.g.,
increasing the effectiveness of year-end audit procedures,
decreasing difficulty associated with audit adjustments) (two-
tailed p-values < 0.10). This supplements our archival findings
related to LAFs vs. AFs and suggests that a primary way that audit
firms were able to minimize the effect of time pressure on LAFs
during deadline changes was to devote more partner hours to these
larger engagements. This strategy could have had a negative impact
on smaller, AF engagements as they may have suffered from a
shortage of partner hours. In addition, it is interesting to note that
PercTIME is negatively associated with the responding partner's
level of industry expertise (IndEXPERIENCE in Table 6) and posi-
tively associated with the time they spent at the client's facilities
(PercFAC in Table 6). Thus, spending more time on a given client/
being present in the field appear to be effective ways of compen-
sating for a lack of industry expertise, as well as a means of
maintaining an acceptable level of audit quality in settings where
PRESSURE exists. This finding complements other studies that
highlight the importance of face-to-face interactions amongst
members of the engagement team (e.g., Agoglia, Brazel, Hatfield, &
Jackson, 2010; Dennis & Johnstone, 2016).
5. Conclusion
SEC rules 33-8128 and 33-8644 substantially reduced the 10-K
filing period for large accelerated (LAFs) and accelerated filers
(AFs) from 90 to 60 and 75 days, respectively (SEC 2002, 2005). We
investigate the effects of this regulation by examining under what
contexts 10-K filing accelerations have been associated with lower
earnings quality. Our study's empirical archival evidence tri-
angulates prior experimental and survey research examining the
detrimental effects of audit time pressure. Also, qualitative data
obtained from our survey of audit partners adds a rich context to
our discussion of time pressure/audit quality and provides insights
regarding best practices when post year-end audit time is reduced.
Overall, our findings provide support for claims by auditors and
preparers that accelerations of 10-K filings have the capacity to
22
See Appendix E of the Online Appendix for analyses of strategies listed in
Table 7 using available archival measures.
 T.A. Lambert et al. / Accounting, Organizations and Society 58 (2017) 50e66
reduce the quality of financial information supplied to external
users. However, these adverse effects appear to depend on the
context of the acceleration. In particular, accelerations may be
problematic if they mandate substantial reductions in audit delay
and, in turn, induce time pressure on the audit engagement team at
year-end. Conversely, if a10-K acceleration does not impact a firm's
audit delay, the effect on earnings quality is not significant. We also
explore whether time pressure differentially affected AFs vs. LAFs,
as well as the earnings quality of firms during the first (75 day
deadline) vs. the second deadline change (60 day deadline). We
observe a more negative effect of pressure on the earnings quality
of AFs (relative to LAFs) and for the initial deadline change (relative
to the second).
We use survey data to support our archival proxies and results.
These data also illustrate that audit time pressure during acceler-
ations increased the level of difficulty associated with resolving
year-end financial statement adjustments. We also provide some
“best practices” for ameliorating the effects of time pressure (e.g.,
increasing interim testing and rescheduling the audits of non-
public companies), as well as some less effective methods (e.g.,
use of computer assisted audit techniques).
Given our findings and the current audit report dates of NAFs
and AFs, regulators may want to exercise caution before acceler-
ating 10-K filings for AFs and NAFs in the future. Overall, our
archival and survey evidence should also be of considerable use to
the SEC and foreign regulators if they consider further accelerations
of financial statement reporting, as well as other future acts that
may place additional year-end time pressure on audits.
Data availability
The archival data used in this study are publicly available from
the sources indicated in the text. Contact the authors for access to
the survey data used in this study.
Acknowledgements
We would like to thank Robert Bloomfield (editor), two anon-
ymous referees, Rashad Abdel-Khalik, Philip Berger, Scott Bronson,
Frank Buckless, Randy Elder, Pieter Elgers, Barbara Grein, Michelle
Hanlon, Linda Hughen, Brad Lail, Elisa Lee, Kathleen Linn, Kevin
Melendrez, Roger Meuwissen, Don Pagach, Mark Peecher, Gary
Peters, Ray Pfeiffer, Marycobb Randall, Scott Richardson, Steve
Salterio, Roger Silvers, Justin Vaughan, and workshop participants
at the University of Illinois, University of Massachusetts, Virginia
Tech University, North Carolina State University, Northern Illinois
University, University of Wisconsin-Milwaukee, Texas Tech Uni-
versity, and Drexel University for their helpful comments. We also
thank the four audit firms and the partners that completed our
survey.
Appendix A. Online Appendix
Online Appendix related to this article can be found at http://dx.
doi.org/10.1016/j.aos.2017.03.003.
References
Agoglia, C. P., Brazel, J. F., Hatfield, R. C., & Jackson, S. B. (2010). How do audit
workpaper reviewers cope with the conflicting pressures of detecting mis-
statements and balancing client workloads? Auditing: A Journal of Practice &
Theory, 9(3), 75e91.
Arens, A. A., Elder, R. J., & Beasley, M. S. (2010). Auditing and assurance services, an
integrated approach. Upper Saddle River, NJ: Prentice Hall.
Asare, S. K., Trompeter, G. M., & Wright, A. M. (2000). The effect of accountability
and time budgets on auditors' testing strategies. Contemporary Accounting
Research, 17(4), 539e560.
65
Ashbaugh, H., LaFond, R., & Mayhew, B. W. (2003). Do nonaudit services compro-
mise auditor Independence? Further evidence. The Accounting Review, 78(3),
611e639.
Ashton, R. H., Willingham, J. J., & Elliott, R. K. (1987). An empirical analysis of audit
delay. Journal of Accounting Research, 25(2), 275e292.
Bamber, E. M., Bamber, L. S., & Schoderbek, M. P. (1993). Audit structure and other
determinants of audit report Lag: An empirical analysis. Auditing: A Journal of
Practice & Theory, 12(1), 1e23.
Bennett, G. B., Hatfield, R. C., & Stefaniak, C. M. (2014). The effect of deadline
pressure on pre-negotiation positions: A comparison of auditors and client
management. Contemporary Accounting Research, 32(4), 1507e1528.
Bloomfield, R., Nelson, M. W., & Soltes, E. (2016). Gathering data for financial
reporting research. Journal of Accounting Research, 54(2), 341e395.
Braun, R. L. (2000). The effect of time pressure on auditor attention to qualitative
aspects of misstatements indicative of potential fraudulent financial reporting.
Accounting, Organizations and Society, 20, 243e259.
Brazel, J. F., & Agoglia, C. P. (2007). An examination of auditor planning judgments in
a complex accounting information system environment. Contemporary Ac-
counting Research, 24(4), 1059e1083.
Brazel, J. F., Carpenter, T. D., & Jenkins, J. G. (2010). Auditors' use of brainstorming in
the consideration of Fraud: Reports from the field. The Accounting Review, 85(4),
1273e1301.
Brazel, J. F., & Dang, L. (2009). The effect of ERP system implementations on the
management of earnings and earnings release dates. Journal of Information
Systems, 22(2), 1e21.
Bryant-Kutcher, L., Peng, E. Y., & Weber, D. (2013). Regulating the timing of
Disclosure: Insights from the acceleration of 10-K filing deadlines. Journal of
Accounting and Public Policy, 32(6), 475e494.
Caramanis, C., & Lennox, C. (2008). Audit effort and earnings management. Journal
of Accounting and Economics, 45(1), 116e138.
Carver, B. T., Hollingsworth, C. W., & Stanley, J. D. (2011). Recent auditor downgrade
activity and changes in clients' discretionary accruals. Auditing: A Journal of
Practice & Theory, 30(3), 33e58.
Christensen, B. E., Glover, S. M., Omer, T. C., & Shelley, M. K. (2015). Understanding
audit Quality: Insights from audit professionals and investors. Contemporary
Accounting Research (forthcoming). Available via http://papers.ssrn.com/sol3/
papers.cfm?abstract_id¼2358163.
Coram, P., Ng, J., & Woodliff, D. R. (2004). The effect of misstatement on the pro-
pensity to commit reduced audit quality acts under time budget pressure.
Auditing: A Journal of Practice & Theory, 23(2), 159e167.
Dechow, P. M., & Dichev, I. D. (2002). The quality of accruals and earnings: The role
of accrual estimation errors. The Accounting Review, 77(4), 35e59.
Dechow, P. M., Sloan, R. G., & Sweeney, A. P. (1995). Detecting earnings manage-
ment. The Accounting Review, 70(2), 193e225.
DeFond, M. L., & Subramanyam, K. R. (1998). Auditor changes and discretionary
accruals. Journal of Accounting and Economics, 25(1), 35e67.
Dennis, S. A., & Johnstone, K. M. (2016). Contemporary insights from a fraud brain-
storming field survey. Working Paper. University of Kentucky. Available via
http://papers.ssrn.com/sol3/papers.cfm?abstract_id¼2746546.
DeZoort, F. T., & Lord, A. T. (1997). A review and synthesis of pressure effects
research in accounting. Journal of Accounting Literature, 16, 28e85.
Doyle, J. T., & Magilke, M. J. (2013). Decision usefulness and accelerated filing
deadlines. Journal of Accounting Research, 51(3), 549e581.
Frankel, R. M., Johnson, M. F., & Nelson, K. K. (2002). The relationship between
auditors' fees for nonaudit services and earnings management. The Accounting
Review, 77(Supplement), 71e105.
Hirst, E., & Koonce, L. (1996). Audit analytical procedures: A field investigation.
Contemporary Accounting Research, 13(2), 457e486.
Impink, J., Lubberink, M., van Praag, B., & Veenman, D. (2011). Did accelerated filing
requirements and SOX section 404 affect the timeliness of 10-K filings? Review
of Accounting Studies, 17, 227e253.
Jones, K., Krishnan, G., & Melendrez, K. (2008). Do models of discretionary accruals
detect actual cases of fraudulent and restated earnings? Contemporary Ac-
counting Research, 25(2), 499e531.
Knechel, W. R., & Payne, J. L. (2001). Additional evidence on audit report lag.
Auditing: A Journal of Practice & Theory, 20(1), 137e146.
Kothari, S. P., Leone, A., & Wasley, C. (2005). Performance matched discretionary
accrual measures. Journal of Accounting and Economics, 39(1), 163e167.
Krishnan, J., & Yang, J. S. (2009). Recent trends in audit report and earnings
announcement lags. Accounting Horizons, 23(3), 265e288.
pez, D. M., & Peters, G. F. (2012). The effect of workload compression on audit
Lo
quality. Auditing: A Journal of Practice & Theory, 31(4), 139e166.
McDaniel, L. S. (1990). The effects of time pressure and audit program structure on
audit performance. Journal of Accounting Research, 28(2), 267e285.
Menon, K., & Williams, D. D. (2004). Former audit partners and abnormal accruals.
The Accounting Review, 79(October), 1095e1118.
Nelson, M. (2009). A model and literature review of professional skepticism in
auditing. Auditing: A Journal of Practice & Theory, 28(2), 1e34.
Otley, D. T., & Pierce, B. J. (1996). The operation of control systems in large audit
firms. Auditing: A Journal of Practice & Theory, 15(2), 65e84.
Peterson, M. (2009). Estimating standard errors in finance Panel data Sets:
Comparing approaches. The Review of Financial Studies, 22(1), 435e480.
Pizzini, M., Lin, S., Vargus, M., & Ziegenfuss, D. (2012). The impact of internal audit
function quality and contribution on audit delays. Working Paper. San Marcos:
Texas State University. Available via http://papers.ssrn.com/sol3/papers.cfm?
66 T.A. Lambert et al. / Accounting, Organizations and Society 58 (2017) 50e66
abstract_id¼1673490.
Public Company Accounting Oversight Board (PCAOB). (2007). Auditing standard No.
5, an audit of internal control over financial reporting that is integrated with an
audit of financial statements. Washington, DC: AICPA.
Public Company Accounting Oversight Board (PCAOB). (2008). Report on the PCAOB's
2004, 2005, 2006, and 2007 inspections of domestic annually inspected firms.
Washington, D.C: PCAOB.
Public Company Accounting Oversight Board (PCAOB). (2010). Auditing standard No.
13, the Auditor's responses to the risks of material misstatement. Washington, DC:
AICPA.
Public Company Accounting Oversight Board (PCAOB). (2012). Maintaining and
applying professional skepticism in audits. Staff audit practice alert No. 10 (SAPA
10). Washington, D.C: PCAOB.
Salterio, S. E. (2012). Fifteen years in the trenches: Auditor-client negotiations
exposed and explored. Accounting & Finance, 52(S1), 233e286.
Schwartz, K. B., & Soo, B. S. (1996). The association between auditor changes and
reporting lags. Contemporary Accounting Research, 13(1), 353e370.
Securities and Exchange Commission (SEC). (2002). Release No. 33-8128, Accelera-
tion of periodic report filing dates and disclosure concerning website access to
reports. Washington, DC: SEC.
Securities and Exchange Commission (SEC). (2004). Release No. 33-8507, Temporary
postponement of the final phase-in period for acceleration of periodic report filing
dates. Washington, DC: SEC.
Securities and Exchange Commission (SEC). (2005). Release No. 33-8644, Revisions to
accelerated deadlines for filing periodic reports. Washington, DC: SEC.
Trompeter, G., & Wright, A. (2010). The world has changed-have analytical pro-
cedure practices? Contemporary Accounting Research, 27(2), 669e700.
Tucker, J. W., & Zarowin, P. A. (2006). Does income smoothing improve earnings
informativeness? The Accounting Review, 81(1), 251e270.
Whitworth, J. D., & Lambert, T. A. (2014). Office-level characteristics of the Big 4 and
audit report timeliness. Auditing: A Journal of Practice & Theory, 33(3), 129e152.
Willett, C., & Page, M. (1996). A survey of time budget pressure and irregular
auditing practices among newly qualified UK chartered accountants. British
Accounting Review, 28, 101e120.
Zmijewski, M. E. (1984). Methodological issues related to the estimation of financial
distress prediction models. Journal of Accounting Research, 22(Supplement),
59e82.