Prompting the Benefit of the Doubt: The Joint Effect of Auditor‐Client Social Bonds and Measurement Uncertainty on Audit Adjustments

• Published date: 01 September 2017
• Date: 01 September 2017
• DOI: http://doi.org/10.1111/1475-679X.12171

DOI: 10.1111/1475-679X.12171

Journal of Accounting Research

Vol. 55 No. 4 September 2017

Printed in U.S.A.

Prompting the Benefit of the Doubt:

The Joint Effect of Auditor-Client

Social Bonds and Measurement

Uncertainty on Audit Adjustments

STEVEN J. KACHELMEIER

∗AND BEN W. VAN LANDUYT

†

Received 21 September 2016; accepted 19 February 2017

ABSTRACT

We design an incentivized experiment to test whether measurement uncer-

tainty elevates the risk that social bonds between auditors and reporters

compromise audit adjustments. Results indicate that, when audit evidence

is characterized by some residual uncertainty, the adjustments our auditor-

participants require are sensitive to whether auditors have an opportunity to

∗Department of Accounting, McCombs School of Business, University of Texas at Austin;

†Department of Accounting, Eller College of Management, University of Arizona.

Accepted by Philip Berger. We appreciate helpful suggestions on previous versions from

an anonymous reviewer, participants at the 2016 AAA Auditing Section Midyear Meeting,

the 2016 AAA Annual Meeting, the 2016 University of Illinois Auditing Doctoral Consor-

tium and Symposium, and workshop participants at the University of Bern, University of

California at Irvine, Erasmus University, Iowa State University, Rutgers University (Newark),

the University of Texas at Austin, and the University of Waterloo. We also appreciate indi-

vidual comments from Markus Arnold, Bradley Bennett, Efrim Boritz, Jim Cannon, Paul De-

mere, Jeremy Douthit, Christine Earley,Prasart Jongjaroenkamol, Bill Kinney, Zach Kowaleski,

Stephan Kramer, Christine Nolder, Sue Ravenscroft, Marcel van Rinsum, Aruhn Venkat,

Michael Williamson, and Emre Y¨

ucel. We gratefully acknowledge funding from the Eugene

and Dora Bonham Endowment Fund and from the Department of Accounting, McCombs

School of Business. The first author also gratefully acknowledges support from the Randal B.

McDonald Chair in Accounting. We are grateful to Paula Kachelmeier and Xinyu Zhang for

research assistance.

Data availability: Contact the authors.

963

Copyright C, University of Chicago on behalf of the Accounting Research Center,2017

964 S.J.KACHELMEIER AND B.W.VAN LANDUYT

form a modest but friendly social bond with reporters. In contrast, although

auditors do not adjust fully even when misstatements are known with cer-

tainty, social bonding has no effect in this scenario. Accordingly, our exper-

iment contributes beyond the main effects of social bonding and measure-

ment uncertainty demonstrated in prior research by showing that these forces

interact. A practical implication is that regulators and practitioners should

consider both the technical and the social challenges facing audits of com-

plex estimates.

JEL codes: C92; D81; M41; M42

Keywords: auditor independence; social bonds; measurement uncertainty;

accounting estimates; leniency; social identity; experimental economics

1. Introduction

When audit evidence is inconsistent with reported values, auditors

have discretion over the amount of adjustment to require. Separate

streams of research have found that auditors require smaller adjust-

ments when social ties threaten independence (e.g., Bamber and Iyer

[2007], Bauer [2015]) or when the amount of adjustment is char-

acterized by uncertainty (e.g., Wright and Wright [1997], Libby and

Kinney [2000], Braun [2001]) or imprecision (e.g., Griffin [2014]). We

design an incentivized experiment to link these themes. Our key finding is

that friendly, social interactions between reporters and auditors influence

the adjustments our auditor-participants require only when the amount of

misstatement is estimated, not when it is known with certainty.

The link between auditor-client social bonds and measurement uncer-

tainty is important because the effect of each construct is likely to depend

on the other. In particular, our study suggests that social bonds alone are in-

sufficient to impair auditor independence. That is, although stronger social

bonds between participants in the auditor and reporter roles decrease the

extent of misreporting in our experiment, they do not influence audit ad-

justments when the amount of misstatement is known with certainty. When

the amount of misstatement is estimated, however, the same social bonds re-

sult in smaller audit adjustments, even after controlling for lower levels of

misreporting.

From the perspective of measurement uncertainty, our findings add

a social dimension to the technical challenges often associated with au-

diting complex estimates (e.g., Bratten et al. [2013], PCAOB [2014],

Griffith, Hammersley, and Kadous [2015], Griffith et al. [2015], Glover,

Taylor, and Wu [2017]). Specifically, our study suggests that even seem-

ingly innocuous social interactions between auditors and clients can lead

auditors to interpret uncertainty in the client’s favor. While prior studies

show that measurement uncertainty or subjectivity can lead auditors to be

PROMPTING THE BENEFIT OF THE DOUBT 965

more lenient (e.g., Wright and Wright [1997], Libby and Kinney [2000],

Braun [2001]), our results suggest that uncertainty alone does not impair

auditor judgments. Rather, we observe a significant downward shift in

auditor-imposed adjustments only when social bonds and measurement un-

certainty are both present.

For theoretical insight, we draw on work from social psychology that

links uncertainty to leniency. Specifically, Ganzach and Krantz [1991] de-

velop psychological arguments supporting their finding that uncertainty

leads to more lenient social judgments, as is well-captured by the famil-

iar adage of giving someone the “benefit of the doubt.” Ganzach and

Krantz [1991] do not, however, examine the potential for social bonds to

strengthen the association between uncertainty and leniency. Accordingly, we

also draw on different literatures involving social identity theory (e.g., Tajfel

[1978], Tajfel and Turner [1979], Tajfel [1981], Ashforth, Harrison, and

Corley [2008]), social comparison theory (e.g., Festinger [1954],

Haunschild [1994], Podolny [1994]), and betrayal aversion (e.g., Bohnet

and Zeckhauser [2004], Bohnet et al. [2008]) to develop our interactive

prediction that stronger social bonds magnify the tendency for uncer-

tainty to promote leniency. In an audit context, this reasoning implies that

auditor-client social bonds will influence audit adjustments primarily when

misstatements are estimated, not when misstatements are known with cer-

tainty.

To capture these notions in a ceteris paribus manner, we design an incen-

tivized experiment in the traditions of experimental economics, in which

we assign student participants to roles analogous to reporters and auditors.

Before participants learn about the reporter-auditor game, they complete

an unrelated exercise involving trivia questions. Our first manipulation is

whether participants answer the trivia questions individually or in pairs,

capturing the presence or absence of an opportunity to form a modest

but friendly social bond. After completing the trivia contest, participants

convene at individual computer stations for a separate, interactive task in

which reporter-participants (hereafter, “reporters”) are compensated for

aggressive reporting and auditor-participants (hereafter, “auditors”) decide

how much of the report to adjust, given a signal that is analogous to au-

dit evidence. Reporter-auditor pairs are either the same pairs that com-

pleted the trivia contest together or newly formed pairs in the condition in

which participants completed the trivia contest individually. Incentives for

the reporter-auditor game reflect the usual tensions in such environments:

reporters profit from unadjusted aggressive reporting, but face a penalty

if adjusted by the auditor. In turn, auditors incur costs to adjust aggressive

reporting, but also face the possibility of a monetary penalty, analogous to

a legal settlement or fine, that increases in probability as the amount of

unadjusted aggressive reporting increases.

Our experiment also manipulates measurement uncertainty. Specifically,

the auditor receives evidence that either reveals the extent of the reporter’s