kaplan and mauldin - 2008 - auditor rotation and the appearance of independence - evidence from non-professional investors

College of Business, 331 Cornell Hall, University of Missouri, Columbia, MO 65211, USA Abstract We examine the impact of audit firm versus partner rotation on non-professional investors’

Auditor rotation and the appearance

of independence: Evidence from non-professional investors

Steven E Kaplana,*, Elaine G Mauldinb,1

a

W.P Carey School of Business, Arizona State University, Tempe, AZ 85287-3606, USA

b

Robert J Trulaske, Sr College of Business, 331 Cornell Hall, University of Missouri, Columbia, MO 65211, USA

Abstract

We examine the impact of audit firm versus partner rotation on non-professional investors’ inde-pendence-related perceptions, extending prior research on auditor rotation and independence in fact Arguments for mandatory audit firm rotation continue to be made by regulators and investor groups based, in part, on the idea that firm rotation will incrementally strengthen independence in appear-ance relative to audit partner rotation We report the results of two experiments The first examines 5-year audit firm versus partner rotation under relatively weak or strong audit committees We find

no statistically significant difference in beliefs about how much of an income reducing audit differ-ence management will record, or in beliefs about auditor independdiffer-ence, between the two auditor rotation conditions On the other hand, we find that non-professional investors do believe more

of the audit difference will be recorded, and the auditors will be more independent, under a strong audit committee than a relatively weak audit committee The second experiment provides further evi-dence on audit firm versus partner rotation by examining a setting involving a 26-year audit firm– client relationship Again, no statistically significant differences between the two auditor rotation conditions were found These findings suggest that compared to audit partner rotation, audit firm rotation does not strengthen independence in appearance among non-professional investors and that non-professional investors recognize the value of strong audit committees

Published by Elsevier Ltd

Keywords: Auditor rotation; Auditor independence; Audit committee; Investor judgment

0278-4254/$ - see front matter Published by Elsevier Ltd.

doi:10.1016/j.jaccpubpol.2008.01.004

* Corresponding author Tel.: +1 480 965 6498; fax: +1 480 965 8392.

E-mail addresses: steve.kaplan@asu.edu (S.E Kaplan), mauldin@missouri.edu (E.G Mauldin).

1

Tel.: +1 573 884 0933.

Journal of Accounting and Public Policy 27 (2008) 177–192

www.elsevier.com/locate/jaccpubpol

1 Introduction

The importance of auditor independence, in fact and in appearance, is widely accepted

in theory and practice (Nelson, 2006) Many contend that regulations permitting an unlim-ited period of association between audit firms and their clients represent a threat to

Sarbanes Oxley Act of 2002 (SOX) mandate the rotation of the lead audit partner, but not the audit firm itself, every 5 years However, because of concerns about the threat

to auditor independence, SOX also directed the United States General Accounting Office (GAO) to study the expected effects of mandatory audit firm rotation The report issued

by the GAO indicates that audit firm rotation may strengthen the appearance of indepen-dence among financial statement users, but that the eviindepen-dence is limited and indirect and calls for additional research on the topic (GAO, 2003)

We contribute to this policy issue by reporting the results of two experiments that pro-vide epro-vidence on whether audit firm rotation, relative to the currently required audit part-ner rotation, strengthens the appearance of independence as reflected in non-professional investors’ perceptions.2 The first experiment also includes two levels of audit committee strength, strong or relatively weak Auditors work for and with the audit committee and both are part of the corporate governance mosaic (Cohen et al., 2004) By manipulat-ing this variable, we are able to provide evidence on whether a strong audit committee strengthens non-professional investors’ independence-related perceptions as well as whether the ability of audit firm rotation to enhance these perceptions depends, in part,

on the strength of the audit committee In this experiment, the audit firm–client relation-ship is five years, mimicking the currently required partner rotation However, the ability

of audit firm rotation, relative to audit partner rotation, to enhance non-professional investors’ independence-related perceptions may depend on the length of the audit firm– client relationship In the second experiment, we consider a relatively long audit firm–cli-ent relationship where the currfirm–cli-ent year is year 26 of the relationship

In both experiments, MBA students representing non-professional investors were given

a case (based onLibby and Kinney, 2000) describing the discovery of an income decreas-ing audit difference by the auditor and were asked to indicate the earndecreas-ings the company would report, implicitly judging how much of the audit difference would be corrected in the final audited financial statements The tension in the case is based on an earnings man-agement issue;3if the entire amount of the audit difference is recorded, the company will miss analysts’ forecasts by $.02, providing management incentive to not record the entire amount Theoretically, auditors perceived as more independent should be perceived as more likely to ‘‘stand up” to management when engaging in earnings management, result-ing in beliefs that more of the income decreasresult-ing audit difference will be recorded, ceteris

2 We focus on non-professional investors’ perceptions because these perceptions are considered crucial to capital markets’ efficiency and effectiveness ( SEC, 2001; Vickers and McNamee, 2002 ), especially with recent increases in the number of non-professional investors ( Securities Industry Association, 2002; SEC, 2003 ) and because investor confidence continues to be a pressing concern for management as well as regulators ( PWC,

2005 ).

3

Earnings management, defined as managers altering financial reports to mislead investors about the underlying economic performance of the company ( Healy and Wahlen, 1999 ), is a concern for regulators desiring

to protect the credibility of financial statements ( Levitt, 1998 ).

paribus (DeZoort et al., 2003) Consequently, we examine both perceived auditor indepen-dence and earnings beliefs

In the first experiment, we find that participants’ earnings beliefs, and perceived auditor independence, were not significantly different under audit firm or partner rotation, regard-less of audit committee strength In the second experiment, we continue to find no signif-icant difference in participants’ earnings beliefs or auditor independence perceptions between audit firm and partner rotation These results are consistent with archival research on independence in fact (Myers et al., 2003; Johnson et al., 2002) and suggest that, among non-professional investors, independence in appearance is not affected by audit firm compared to audit partner rotation On the other hand, the results from the first experiment show that participants’ beliefs about reported earnings were lower, and per-ceived auditor independence was higher, in the presence of a strong audit committee, com-pared to a relatively weak audit committee These results indicate that audit committees adopting best practices provide perceptual benefits, enhancing the auditors’ perceived independence among non-professional investors

This study contributes to both research and policy setting For research, the study pro-vides direct evidence on perceptions about the appearance of independence from non-pro-fessional investors The task of forming a judgment about how much of an audit difference will be recorded is especially useful because it provides evidence about investors’ beliefs that are not otherwise available Thus, the study complements archival research exploring the role of corporate governance mechanisms in improving financial reporting credibility For policy setting, the study provides ex ante experimental evidence, not available in archi-val studies, about the impact one might expect from mandating audit firm rotation Such

ex ante evidence should be useful to policymakers as they further consider the need for mandated audit firm rotation

The following section reviews prior literature and develops research questions Then, the experimental design is described and the results are presented The last section dis-cusses the results and provides concluding comments

2 Background

Independent external audits play a central role in the financial reporting process of

(AICPA, 1972) Because independence in fact is unobservable, investors’ perceptions mat-ter and perceived lack of independence can be as damaging as an actual violation of inde-pendence (Olazabal and Almer, 2001) The importance of investors’ perceptions is often recognized in regulatory actions such as SOX, which was passed, in part, to restore inves-tor confidence

2.1 Auditor rotation

Hoyle (1978)traces periodic interest in audit firm rotation from the 1939 Congressional investigation into the McKesson and Robbins fraud case to the 1976 issuance of the Met-calf report and investor activists’ testimony calling for audit firm rotation These calls con-tinue to be made by some researchers (Moore et al., 2006), investor groups (Sinnett, 2004),

as a way to protect against threats to auditor independence Historically, audit firms and organizations, have argued against mandatory audit firm rotation asserting that other pro-tections, such as reputation and legal liability, are sufficient to maintain investor confi-dence (AICPA, 1992)

TheGAO’s report (2003)concluded that the currently required audit partner rotation enhances auditor independence in fact as well as audit firm rotation,4but indicated that mandatory audit firm rotation may yet be necessary to strengthen auditor independence

in appearance (GAO, 2003) This conjecture, was based, in part, on the results from a sur-vey of key stakeholders, where 65% (38%) of the public company (public accounting firm) members responded that mandatory audit firm rotation would strengthen perceptions of auditor independence (GAO, 2003) However, this survey evidence does not directly rep-resent the beliefs of investors In an experimental study, Jennings et al (2006)find that judges perceive higher auditor independence when the firm was rotated than when the partner was rotated, but again this does not directly represent the beliefs of investors.5 Recently,Raiborn et al (2006, 39)contends that, relative to audit partner rotation, man-datory audit firm rotation ‘‘would increase the public perception of auditor independence” (italics from original)

Threats to auditor independence may theoretically stem from sociological or economic pressures Under economic theory, audit firms have an economic incentive to keep an ongoing relationship with an audit client in order to obtain profits from future audits Regarding independence in fact, when the audit firm anticipates an ongoing future rela-tionship the audit firm may audit less vigilantly and be more predisposed to go along with aggressive and/or questionable accounting methods to protect their stream of quasi-rents (Watts and Zimmerman, 1986) Mandating audit firm rotation limits the stream of profits from future audits, lessening the economic incentives for the audit firm to act in a less than independent fashion By lessening these economic incentives, mandating a limited time horizon is expected to strengthen perceptions of auditor independence The impact of less-ening economic incentives must occur at a firm level to be effective since all partners (cur-rent and rotating) are likely to be impacted by the firm’s economic incentives

Under a sociological perspective,Moore et al (2006)introduce the term ‘‘moral seduc-tion” to describe how, over time, clients exert a ‘‘gradual accumulation of pressures” to

‘‘encourage complacency among practitioners” such that auditors will be more likely to

‘‘slant their conclusions” (Moore et al., 2006, 11) Bamber and Iyer (2007) provide evi-dence consistent with this concern on an individual auditor basis Their study shows that the length of auditors’ tenure with a client is associated with the extent to which auditors identify with a client, which in turn, is associated with greater acceptance of client pre-ferred outcomes

To examine the contention that audit firm rotation strengthens the appearance of inde-pendence, we use a situation in which management has incentives to manage earnings to

4 Arel et al (2006) and Hatfield et al (2006) provide some evidence on auditor rotation and independence in fact Arel et al (2006) find that experienced auditors believed their audit firm would be more likely to issue a modified audit report under firm rotation than no rotation Hatfield et al (2006) also find positive results comparing firm to no rotation, but find that audit partner versus audit firm rotation results in similar judgments.

5

It is also worth noting that the case included information that non-audit services were more than three times the audit fee It is possible that judges’ perceptions of auditor independence may interact with the level of non-audit fees and that their findings may not generalize to a setting with lower non-non-audit service fees.

meet or beat the analysts’ consensus forecast and the auditors have identified an unre-corded accounting difference If audit firm rotation strengthens the appearance of indepen-dence by reducing perceived economic and/or sociological pressures compared to audit partner rotation, then non-professional investors will perceive higher auditor indepen-dence and will expect reported earnings to be lower, stated as follows:

Research question 1a: Will non-professional investors expect reported earnings to be lower when the audit firm is rotated than when the audit partner is rotated?

Research question 1b: Will non-professional investors perceive the auditor more inde-pendent when the audit firm is rotated than when the audit partner is rotated? 2.2 Audit committee strength

Prior research suggests that the role of the audit firm is complex because the firm inter-acts with both the audit committee and management (Cohen et al., 2004) The audit com-mittee appoints the audit firm and oversees the financial reporting process, including resolving any differences between management and the audit firm Prior research generally supports the idea that stronger audit committees are associated with positive financial reporting outcomes (Abbott et al., 2004; DeZoort and Salterio, 2001; Xie et al., 2003) and also finds that stronger audit committees are positively associated with greater sup-port for the auditor’s position on accounting issues (DeZoort and Salterio, 2001; DeZoort

et al., 2003) If a strong audit committee strengthens the appearance of independence, then non-professional investors will perceive higher auditor independence and will expect reported earnings to be lower, stated as follows:

Research question 2a: Will non-professional investors expect reported earnings to be lower when the audit committee is strong than when it is relatively weak?

Research question 2b: Will non-professional investors perceive the auditor more inde-pendent when the audit committee is strong than when it is relatively weak?

The extent to which audit firm rotation enhances independence-related perceptions might interact with the strength of the audit committee because both the audit firm and

et al., 2004) It might be the case that any perceptual benefits might diminish when the audit committee is strong The audit committee would presumably be expected to closely monitor the work of the audit firm, such that the form of audit rotation would no longer

be consequential For example, in a case involving the discovery of fraud triggering sub-stantial decline in the stock price,Jennings et al (2006)find that judges’ auditor liability perceptions were less when the firm was rotated compared to partner rotation, but only under conditions of relatively weak corporate governance On the other hand, for inves-tors a strong audit committee may interact with audit firm rotation to provide even higher perceptions of independence This discussion leads to our final research question, stated as follows:

Research question 3a: Will the effect of auditor rotation on non-professional investors’ reported earnings beliefs be different when the audit committee is strong than relatively weak?

Research question 3b: Will the effect of auditor rotation on non-professional investors’ perceptions of auditor independence be different when the audit committee is strong than relatively weak?

3 Methodology and results for Experiment 1

3.1 Procedures and task

rotation and audit committee strength Experimental materials included a cover sheet, case materials, and additional questions The cover sheet requested participation and provided general instructions including a guarantee of anonymity and a definition of an audit difference

medium-sized publicly traded auto parts manufacturer, Capital Auto Parts (CAP) Information about the company, pre-audit financial statement balances, earnings per share (EPS), and analysts’ consensus forecast was constant across all treatments The case also described the discovery of an audit difference related to management’s estimate of the inventory obsolescence allowance, stating: ‘‘the auditor believes that the recorded allow-ance is outside a reasonable range by an amount that overstates current earnings per share

by $.03.” The case was designed such that under traditional benchmarks the audit differ-ence would be considered quantitatively immaterial (Brody et al., 2003) However, fully adjusting earnings for the audit difference would result in earnings $.02 below the analysts’ forecast, providing a context where management has incentives to not record the audit difference

After reading the case, each participant indicated ‘‘the most likely EPS amount CAP would finally report in the audited financial statements for the year” by circling one of four amounts ranging from $1.07 (all the audit difference corrected) to $1.10 (none of the audit difference corrected) We use responses to this question, the earnings belief, as the first dependent measure For the second dependent measure, perceived auditor independence,

we use responses from one question in the additional questions where participants were asked to assess the independence of CAP’s auditor on a seven point response scale anchored by ‘‘low independence” (coded 1) and ‘‘high independence” (coded 7) Other additional questions included questions about the manipulated variables, and the partic-ipant’s background and attitudes

3.2 Treatment variables

For the auditor rotation treatment, participants were randomly assigned to one of two conditions, partner or firm rotation Reflecting current regulation, the partner rotation condition indicated the external audit firm rotates the audit partner in charge of the engagement every 5 years For the firm rotation condition, the case indicated the audit committee has a policy of rotating the external audit firm every 5 years This time frame was chosen to be consistent with the partner rotation condition and with various investor and legislative proposals recommending firm rotation every 5–7 years (Moore et al., 2006; Zeff, 2003; Marshall, 2002; Hoyle, 1978) This time frame is also within the range of man-datory audit firm rotation existing outside the United States (Zeff, 2003) Regarding the

year in the rotation cycle, the case indicated that since this was the audit partner’s (external audit firm’s) fifth year with the client, a new audit partner (external audit firm) would take over the engagement next year Setting the case in the last year prior to rotation is consis-tent withArel et al (2006) In the year prior to rotation the audit partner and firm has the least to lose in terms of future quasi rents (DeAngelo, 1981)

For the audit committee strength treatment, participants were randomly assigned to either a strong or relatively weak audit committee condition Though SOX included sev-eral provisions meant to strengthen the effectiveness of audit committees,6they fell short of the Blue Ribbon Committee (BRC) on Improving the Effectiveness of Corporate Audit

To maintain realism, however, both treatments were designed to be within current regulations

Based on previous research (DeZoort et al., 2002; Cohen et al., 2004), the manipulation

of audit committee strength incorporates multiple measures.DeZoort et al (2002)contend that overall audit committee effectiveness is multi-dimensional, including composition (financial expertise and independence), diligence (number of meetings7), resources (num-ber of mem(num-bers and the committee’s access to management, external auditors, and internal auditors), and authority (committee responsibilities) Under the strong audit committee treatment, all five members of the audit committee were independent outside directors who also were financial experts as defined by the SEC (and two of the members were CPAs) Further, the audit committee met with senior members of management, the vice president of internal audit, and the external auditors in private, separate sessions six times during the year In contrast, under the relatively weak audit committee treatment, all three audit committee members were independent outside directors, but only one was consid-ered a financial expert as defined by the SEC and none had a background in either accounting or finance Further, the audit committee met only once a year, together with senior members of management as well as the vice president of internal audit and the

3.3 Manipulation checks and demographics

Participants were 163 MBA students who were randomly assigned to treatment

the additional questions were used to test participants’ understanding of the treatment

6 Section 301 requires all committee members to be independent of management and that the audit committee is directly responsible for the appointment, compensation, and oversight of the external auditor Section 407 requires companies to disclose the audit committee member(s) that is a financial expert or indicate the absence of

a financial expert on the audit committee.

7 More frequent meetings are associated with higher effort ( DeZoort et al., 2002 ).

8 Neither SOX nor Nasdaq listing requirements specify either a required frequency of meetings or who should

be included in the meetings The NYSE listing requirements also do not specify meeting frequency but, since November 2004, do specify that the audit committee should meet separately with management, the internal auditors, and the external auditors The experimental instrument specifies that the audit committee is ‘‘compliant with current regulations” implicitly placing the company in a Nasdaq setting.

9

Four students did not complete the case questionnaire and were dropped from the study.

variables, auditor rotation and audit committee strength For the analyses shown, partic-ipants incorrectly answering either of the manipulation check questions (38) were dropped because of a lack of ‘‘inclusion importance” (Yates, 1990, p 376) AsTan and Yates (1995,

p 315) contend: ‘‘if a decision maker never acknowledges the existence of a particular dimension, then the decision maker cannot possibly respond to that dimension” There

is no significant difference in the number of participants dropped between conditions When all participants are included in the analyses (not shown), the significance, or lack thereof, of each variable remains the same

Table 1, Panel A, presents a demographic profile of the 125 participants included in the results Participants’ mean age was about 30 and their mean work experience was about 7 years The majority of participants was male and owns or has owned stock Participants’ mean rating of their own ability to understand financial reporting was above the mid-point This demographic profile suggests that the participants are capable of completing the task and are a reasonable proxy for non-professional investors There is no significant difference across treatment condition for any of the demographic variables (p > 10) sug-gesting that randomization of participants to condition was effective In sensitivity analy-ses, not shown, we include each of the background variables as a control variable in the models reported in the results The only variable that is significant is age and the reported results are unaffected so we do not report these covariates

Table 1, Panel B, presents a profile of participants’ attitudes relevant to materiality of the audit difference, task realism and importance As shown, participants’ mean assess-ment of materiality was 5.31, on the highly material end of the scale suggesting that par-ticipants understood the impact of the audit difference on the company’s ability to meet the forecast Participants’ mean rating of task realism was 5.00, above the mid-point of the seven-point scale Finally, participants indicated that their assessments of financial statement reliability influenced their investment decisions (mean 5.84), suggesting their beliefs about financial reporting outcomes are important There is no significant difference across treatment condition for these attitude variables (p > 10) further suggesting that randomization of participants to conditions was effective Again, sensitivity analyses

Table 1

Experiment 1 – Profile of 125 participants’ background and attitudes

Mean Std.

dev.

Minimum Maximum

Panel A: Profile of participants’ background

Ability to understand financial reporting (1, very low to 7, very

high)

Panel B: Profile of participants’ task attitudes

Materiality of audit difference (1, not material at all to 7, highly

material)

Task realism (1, very unrealistic to 7, very realistic) 5.09 1.17 1 7

My assessment of the trustworthiness or reliability of the annual

audited financial statements strongly influences my investment

decision (1, strongly disagree to 7, strongly agree)

(not shown) reveal no effects on the reported results when the variables are each included

as controls in the reported models

3.4 Analyses of research questions

General linear model analysis of variance (GLM) is used to present evidence relative to the research questions because it is appropriate for testing unbalanced designs (Milliken and Johnson, 1992) In the first analysis, the dependent variable is the participant’s earn-ings belief.10The independent variables are auditor rotation (partner or firm), audit com-mittee strength (strong or relatively weak), and the interaction term Table 2, Panel A presents descriptive statistics by condition for the dependent variable and Panel B presents multivariate results

Table 2

Experiment 1 – Analysis of auditor rotation and audit committee strength on most likely reported (audited) EPS a

Panel A: Descriptive statistics

Audit committee strength Auditor rotation

Panel B: GLM ANOVA – Most likely reported EPS

a Most likely reported (audited) EPS is either $1.07, $1.08, $1.09, or $1.10.

10 Two other analyses were used to test the sensitivity of the results to alternative specifications (not shown) First, we transformed the dependent variable to ranks and re-performed the GLM analysis reported Second, using logistic regression, the dependent variable was calculated as a dichotomous variable taking the value of one for reported EPS of $1.09 or $1.10 (meet or beat analysts’ forecast) and zero for reported EPS of $1.07 or $1.08 (miss analysts’ forecast) The statistical results of both sensitivity analyses are substantially equivalent to those reported.

Research question 1a asks whether earnings beliefs will be lower when the audit firm is rotated compared to when the audit partner is rotated As shown inTable 2Panel A, mean (median) estimated EPS is 1.083 (1.08) and 1.085 (1.09) under the audit partner and audit firm rotation, respectively Neither the means nor the medians are significantly different (p > 10), indicating audit firm rotation is not perceived to constrain earnings management

to a greater extent than audit partner rotation Consistent with these univariate results, the multivariate analysis in Panel B finds the main effect for auditor rotation insignificant (p = 24)

Research question 2a asks whether earnings beliefs will be lower when the audit

mean (median) estimated EPS is 1.081 (1.07) and 1.086 (1.09) under the strong and rela-tively weak audit committee, respecrela-tively As shown in Panel B, the main effect for audit committee strength is significant (p = 02), indicating that a strong audit committee is per-ceived to constrain earnings management to a greater extent than a relatively weak audit committee Research question 3a asks whether the effect of auditor rotation on earning beliefs depends on audit committee strength Panel B shows that the interaction between auditor rotation and audit committee strength is not significant (p = 83), indicating that the effect of auditor rotation was not significantly different by level of audit committee strength

In the second analysis related to research questions 1b to 3b, the dependent variable is the participant’s assessment of the auditor’s perceived independence Descriptive statistics

Table 3

Experiment 1 – Analysis of auditor rotation and audit committee strength on perceived independencea Panel A: Descriptive statistics

Audit committee strength Auditor rotation

Panel B: GLM ANOVA – Perceived external auditor independence

a

Participant beliefs about the independence of CAP’s external auditor, answered on a seven-point scale anchored on 1 – low independence and 7 – high independence.