bahan kuliah-a test of the auditor reliability framework using lenders judgments

15
A test of the auditor reliability framework using lenders’ judgments F. Todd DeZoort a,, Travis Holt b , Mark H. Taylor c a The University of Alabama, United States b University of Tennessee at Chattanooga, United States c Case Western Reserve University, United States abstract Taylor et al. (2003) challenged the longstanding notion that independence is the capstone of the audit profession by proposing a conceptual framework that emphasizes reliability, rather than independence, as the professional endgame for auditors. Although the reliabil- ity framework has attracted attention from policymakers, it has not been tested empirically in an audit context to assess its validity from a user’s perspective. The objective of this study is to test the auditor reliability framework and its formative ethical constructs (i.e., integrity, expertise, independence, objectivity, and reliability) with a sample of 168 commercial lenders. We also extend the reliability framework to examine the extent that perceived auditor reliability affects lenders’ judgments of financial reporting reliability and default risk in a hypothetical lending scenario. Finally, we evaluate the extent that lenders’ judgments are affected by auditor provision of nonaudit bookkeeping and payroll services to a prospective borrower in violation of current independence rules. The results provide strong empirical support for the relations predicted in the reliability framework. Structural equation model results indicate that auditor integrity is the foundation of the framework, directly affecting lenders’ assessments of auditor expertise, independence, objectivity, and reliability. Further, although integrity and objectivity directly affect perceived auditor reli- ability, independence and expertise only affects reliability indirectly through its impact on objectivity. Finally, we find that lenders perceive no decrease in auditor objectivity or reli- ability when existing independence rules are violated by combining audit services with nonaudit services for prospective borrowers. Ó 2012 Elsevier Ltd. All rights reserved. Introduction Independence has long been considered the capstone of the audit profession because it has been assumed to be the foundation for public trust (Bazerman & Moore, 2011; Jamal & Sunder, 2011; Jeppesen, 1998; Lee, 1993; Lindberg & Beck, 2004; Mautz & Sharaf, 1961; Montgomery, 1940; Nelson, 2006). However, the accounting literature includes numerous challenges to the role of independence and its relationship to other critical ethical constructs within the profession. For example, DeAngelo (1981) and subsequent studies (e.g., Ganguly, Herbold, & Peecher, 2007) asserted that the two pillars of audit quality are actually integrity and competence rather than independence. Jamal and Sunder (2011) questioned whether mandated indepen- dence is even necessary for audit quality, finding in a field study of baseball card certification in an unregulated market that the demand for attester expertise is prioritized over the demand for independence. Taylor, DeZoort, Thomas, and Munn (2003) introduced an auditor reliability framework that defines and organizes critical ethical constructs in the profession (e.g., integrity, independence, expertise, objectivity) to emphasize profes- sional reliability rather than independence as the relevant ‘‘endgame’’ when providing assurance related to financial reporting. They describe auditor reliability in the frame- work as a condition where stakeholders consistently find 0361-3682/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.aos.2012.08.003 Corresponding author. Address: Culverhouse School of Accountancy, Box 870220, Tuscaloosa, AL 35487, United States. Tel.: +1 205 348 6694; fax: +1 205 348 8453. E-mail address: [email protected] (F.T. DeZoort). Accounting, Organizations and Society 37 (2012) 519–533 Contents lists available at SciVerse ScienceDirect Accounting, Organizations and Society journal homepage: www.elsevier.com/locate/aos

Upload: agung-cyberhunter-longtail

Post on 24-Nov-2015

18 views

Category:

Documents


2 download

TRANSCRIPT

  • ew

    or c

    numerous challenges to the role of independence and itsrelationship to other critical ethical constructs within theprofession. For example, DeAngelo (1981) and subsequent

    Taylor, DeZoort, Thomas, and Munn (2003) introducedan auditor reliability framework that denes and organizescritical ethical constructs in the profession (e.g., integrity,independence, expertise, objectivity) to emphasize profes-sional reliability rather than independence as the relevantendgame when providing assurance related to nancialreporting. They describe auditor reliability in the frame-work as a condition where stakeholders consistently nd

    0361-3682/$ - see front matter 2012 Elsevier Ltd. All rights reserved.

    Corresponding author. Address: Culverhouse School of Accountancy,Box 870220, Tuscaloosa, AL 35487, United States. Tel.: +1 205 348 6694;fax: +1 205 348 8453.

    E-mail address: [email protected] (F.T. DeZoort).

    Accounting, Organizations and Society 37 (2012) 519533

    Contents lists available at SciVerse ScienceDirect

    Accounting, Organiza

    journal homepage: www.ehttp://dx.doi.org/10.1016/j.aos.2012.08.003Introduction

    Independence has long been considered the capstone ofthe audit profession because it has been assumed to be thefoundation for public trust (Bazerman & Moore, 2011;Jamal & Sunder, 2011; Jeppesen, 1998; Lee, 1993; Lindberg& Beck, 2004; Mautz & Sharaf, 1961; Montgomery, 1940;Nelson, 2006). However, the accounting literature includes

    studies (e.g., Ganguly, Herbold, & Peecher, 2007) assertedthat the two pillars of audit quality are actually integrityand competence rather than independence. Jamal andSunder (2011) questioned whether mandated indepen-dence is even necessary for audit quality, nding in a eldstudy of baseball card certication in an unregulatedmarket that the demand for attester expertise is prioritizedover the demand for independence.a b s t r a c t

    Taylor et al. (2003) challenged the longstanding notion that independence is the capstoneof the audit profession by proposing a conceptual framework that emphasizes reliability,rather than independence, as the professional endgame for auditors. Although the reliabil-ity framework has attracted attention from policymakers, it has not been tested empiricallyin an audit context to assess its validity from a users perspective. The objective of thisstudy is to test the auditor reliability framework and its formative ethical constructs(i.e., integrity, expertise, independence, objectivity, and reliability) with a sample of 168commercial lenders. We also extend the reliability framework to examine the extent thatperceived auditor reliability affects lenders judgments of nancial reporting reliability anddefault risk in a hypothetical lending scenario. Finally, we evaluate the extent that lendersjudgments are affected by auditor provision of nonaudit bookkeeping and payroll servicesto a prospective borrower in violation of current independence rules. The results providestrong empirical support for the relations predicted in the reliability framework. Structuralequation model results indicate that auditor integrity is the foundation of the framework,directly affecting lenders assessments of auditor expertise, independence, objectivity, andreliability. Further, although integrity and objectivity directly affect perceived auditor reli-ability, independence and expertise only affects reliability indirectly through its impact onobjectivity. Finally, we nd that lenders perceive no decrease in auditor objectivity or reli-ability when existing independence rules are violated by combining audit services withnonaudit services for prospective borrowers.

    2012 Elsevier Ltd. All rights reserved.A test of the auditor reliability fram

    F. Todd DeZoort a,, Travis Holt b, Mark H. Tayla The University of Alabama, United StatesbUniversity of Tennessee at Chattanooga, United StatescCase Western Reserve University, United Statesork using lenders judgments

    tions and Society

    lsev ier .com/ locate/aos

  • the auditors work credible and dependable, even afteracknowledging the inherent limitations of audit and otherattest functions. In this context, auditor reliability pro-motes nancial reporting reliability that has long beenemphasized as a critical qualitative characteristic of nan-cial information.

    In response to the Taylor et al. (2003) reliability frame-work, the American Institute of Certied Public Accoun-tants (AICPA) formed a Reliability Task Force to evaluatethe frameworks potential to improve professional stan-dards and the quality of professional services by shiftingfrom the traditional independence-based model to a reli-ability-based model (DeZoort, Morgan, Ratcliffe, & Taylor,2008).1 Specically, the Task Force recognized that the tradi-tional independence-based model is a problem for clientswho require more than attest services from their auditorsto achieve reliable nancial reporting.2 The Task Force con-cluded that a refocus on reliability could better serve thepublic interest by allowing auditors to provide attest clientswith certain nonattest services (e.g., internal control ser-vices) to improve the reliability of nancial reporting.3 Fur-ther, the reliability framework was used as a basis forspecic Task Force recommendations to the Accountingand Review Services Committee (ARSC), which in turn pro-posed revisions to longstanding standards to allow auditors

    in both compilation and review settings, even when bothattest and nonattest services were provided in violationof current independence rules. The study results also indi-cate that although integrity, expertise, and objectivity havesignicant direct and indirect effects on lender assess-ments of CPA reliability and nancial reporting reliability,independence only affected reliability indirectly throughits impact on objectivity.

    Although policymakers have been using the reliabilityframework to guide standard-setting (AICPA, 2009a;DeZoort et al., 2008), the literature lacks empirical evidencethat assesses the validity of the frameworks proposed rela-tions in the frameworks original audit context. The auditfunction is unique given its relatively high levels of assur-ance, user reliance, scrutiny, and consequence comparedto other attest functions. For example, prior studies (e.g.,Bartlett, 1991; Gay, Schelluch, & Baines, 1998; Johnson,Pany, &White, 1983; Yardley, 1989) provide consistent evi-dence that nancial statement users desire and perceivegreater assurance from nancial statement audits than

    520 F.T. DeZoort et al. / Accounting, Organizations and Society 37 (2012) 519533to issue compilation and review reports when independenceis impaired due to the provision of nonattest internal controlservices (AICPA, 2009a).4

    Given the Reliability Task Force and ARSC focus on com-pilation and review engagements, the AICPAs Private Com-pany Practice Section commissioned an empirical study(DeZoort & Taylor, 2009) of the reliability framework tofurther evaluate its validity in those attest contexts. Theresults from commercial lenders provide strong initialempirical support for the frameworks predicted relations

    1 The AICPA Reliability Task Force consisted of practitioners, preparers,academics, and third-party users (e.g., bankers).

    2 The AICPA promulgates technical auditing standards for audits ofnonpublic companies and authoritative ethical rules for all membersengaging in public practice. Although the AICPA Reliability Task Force wasprimarily concerned with improving US professional standards, its ndingsare applicable to international professional standards. Professional codes ofconduct (e.g., AICPA, 2010; IFAC, 2010) do not allow accountants to issueaudit reports or review reports if independence is lacking. Accountants canperform compilations when independence is impaired as long as the lack ofindependence is disclosed in their report.

    3 When considering the framework and current independence rules, theTask Force distinguished among independence impairments caused bynancial interests, family relationships, and CPA performance of nonattestservices. Recommendations focused only on the performance of selectnonattest services. The Reliability Task Force reafrmed the importance ofmaintaining family-based independence (e.g., ensuring that an immedi-ate family member of the CPA is not in a key management position at theclient) and nancial-interest independence (e.g., avoiding direct invest-ments in the client) (DeZoort et al., 2008).

    4 The ARSC exposure draft motivated a large number of comment lettersthat reected both strong support and strong opposition to the proposal.Ultimately, the nal standard (AICPA, 2009b) was released without theprovision because of confusion about the specic scope of internal controlservices targeted in the proposal. The AICPA Professional Ethics ExecutiveCommittee (PEEC) is currently considering revisions to Interpretation 101-3to clarify that accountants can perform certain internal control services forclients and maintain independence as long as management takes respon-sibility for the results of those services (AICPA, 2011).from other attest services (e.g., reviews, compilations). Ina lending context, Reinstein and Leibowitz (2010) suggesta specic link between debt levels and desired assurance le-vel, with lenders requiring nancial statement audits ofborrowers with large outstanding balances. The distinctfeatures of the audit functionmotivated the Reliability TaskForce to highlight the need for further research to evaluatethe framework in an audit context (AICPA, 2008).

    This study provides an initial empirical test of the audi-tor reliability framework proposed in Taylor et al. (2003),revised in DeZoort et al. (2008), and subsequently modiedfor testing in compilation and review contexts (DeZoort &Taylor, 2009). We adopt a users perspective in this studyand evaluate lending ofcers perceptions about auditorintegrity, expertise, independence, objectivity, and reliabil-ity in a hypothetical lending scenario and test the validityof the proposed relations among the frameworks forma-tive constructs. We also evaluate the extent to whichperceived auditor reliability affects lenders judgments ofnancial reporting reliability and default risk. Finally, we

    Fig. 1. Reliability framework (DeZoort et al., 2008).

  • consider whether lenders assessments of the reliabilityframework constructs are affected when auditors violateindependence rules by providing both audit and nonauditbookkeeping and payroll services to the prospective

    F.T. DeZoort et al. / Accounting, Organizations and Society 37 (2012) 519533 521borrower.This study is motivated by the need for empirical evi-

    dence to test the reliability framework in its intended auditcontext given the theoretical relations proposed, and poli-cymakers interest, in implementing changes based on theimplications of the framework. For example, the framework(see Fig. 1) describes integrity, expertise, and independenceas formative ethical constructs underlying the pursuit ofobjectivity, which in turn underlies auditor reliability(DeZoort et al., 2008; Taylor et al., 2003). In this context,although relationship-based independence remains animportant construct, it is not a wholly sufcient antecedentto auditor objectivity and reliability given the importanceof integrity and expertise.5 The framework provides for thepossibility of auditor and nancial reporting reliability in sit-uations where auditors lack independence if they have theintegrity and expertise needed to pursue objectivity. Thisstudy provides initial evidence about the extent that usersagree with the frameworks proposed relations and extendstesting of the framework beyond compilation and review set-tings to consider its validity in the distinct audit context.

    We test the reliability framework using a sample of 168experienced commercial lending ofcers who are membersof the Risk Management Association (RMA). Lenders repre-sent a critical user group who can provide an initial empiri-cal perspective on the relative importance of (and relationsamong) the reliability frameworks constructs and how theframework affects perceived nancial statement reliabilityanddefault risk judgments. The extant accounting literatureprovides numerous studies (e.g., Ashton & Ashton, 1995;Beaulieu, 1994; Danos, Holt, & Imhoff, 1989; Holder-Webb& Sharma, 2010; Minnis, 2011) focused on lenders judg-ments because of their role as crucial users of audited nan-cial statements. For example, Holder-Webb and Sharma(2010) studied how commercial lenders decisions were af-fected by their perceptions of reporting reliability, ndingthat lenders are indeed sensitive to nancial condition andperceived nancial reporting reliability.

    The results provide strong initial support for the auditorreliability framework. Specically, we nd that lendersassessments of auditor expertise, integrity, and indepen-dence directly affect their assessments of auditor objectiv-ity. The results also indicate that lenders assessments ofauditor objectivity positively affect their perceptions ofauditor reliability. Additionally, we nd that perceptionsof auditor reliability are positively associated with assess-ments of nancial reporting reliability, and that nancialreporting reliability, in turn, is negatively related to assess-ments of default risk. These results are robust whenauditors provide both audit services and nonaudit book-keeping and payroll services in violation of current auditorindependence rules.

    5 Taylor et al. (2003) developed the framework to address concerns aboutindependence being the cornerstone of the profession and continuedconfusion about the meaning of independence and other key ethicalconstructs in the profession.The next section develops the studys hypotheses re-lated to the reliability framework and provision of nonau-dit services. The third section describes the studysresearch method. The fourth section provides the results,followed by discussion of the studys limitations andimplications.

    Theory and hypotheses

    The rst set of hypotheses focuses on the reliabilityframeworks formative constructs. Specically, we testthe frameworks predicted relations among auditor integ-rity, expertise, independence, objectivity, and reliability.We also extend the reliability framework analysis to pre-dict that lenders assessments of auditor reliability will di-rectly affect their assessments of borrower nancialstatement reliability and lenders default risk judgments.

    Integrity effects

    Taylor et al. (2003) identied integrity as a key founda-tion component in the pursuit of auditor reliability. Integ-rity in this context is a core virtue in an individualscharacter that reects the ability to be straightforward,honest, and forthright in all professional and businessactivities. After AICPA Reliability Task Force deliberations,DeZoort et al. (2008) revised the initial framework to rec-ognize integrity as the sole ethical foundation. Specically,the framework suggests that auditor integrity directlyunderlies expertise and independence needed to pursueobjectivity. For example, integrity is posited to directlyunderlie auditors good faith efforts to acquire and main-tain the knowledge and experience needed to understandclients, industries, and engagements to ultimately providereliable services. Hardwig (1994) discusses the rationalityof trusting experts and highlights that integrity is a criti-cal antecedent to expertise for users because the underly-ing integrity of experts provides the basis for trust that isneeded to rely on an experts judgments and decisions.Although this trust is sometimes broken, judgments aboutan experts integrity remain part of a users justication forrelying on the expert. Accordingly, we predict that lendersperceptions of auditor integrity will be linked directly totheir assessments of auditor expertise. Stated formally:

    Hypothesis 1a. Lenders assessments of auditor integritywill be positively related to their assessments of auditorexpertise.

    The framework also proposes a direct link betweenintegrity and independence in the pursuit of objectivityand reliability. This link stems from the professional stan-dards (e.g., AU section 220) which require auditors to behonest and forthright when evaluating the nancial andnonnancial relationships with clients that can create con-icts of interest and undermine objectivity and reliability.The extant research literature supports this link with evi-dence that auditors level of moral development affectstheir independence judgments. For example, Sweeneyand Roberts (1997) found that auditors with increasedlevels of moral development were more sensitive to ethical

  • issues and less likely to rely solely on technical standardswhen making independence judgments. From a users per-spective, auditor integrity should be a critical antecedentto efforts to comply with existing independence rules

    sue when providing attest services to clients.6 Taylor et al.

    2003, p. 262). Ultimately, expertise with underlying integ-rity should help auditors pursue objectivity needed tomanage the variety of technical, client, and industry issuesthat arise when providing client services.7 From a users

    objectivity.

    522 F.T. DeZoort et al. / Accounting, Organizations and Society 37 (2012) 519533(2003) linked auditor integrity directly to objectivity by sug-gesting that integrity should help professionals maintainclear focus on truthful reporting and be sensitive to dis-honest manipulated reporting behavior (p. 262). Accord-ingly, we predict that lenders assessments of auditorintegrity will be related directly to their assessments ofauditor objectivity. Stated formally:

    Hypothesis 1c. Lenders assessments of auditor integritywill be positively related to their assessments of auditorobjectivity.

    Expertise effects

    The reliability framework also describes auditor exper-tise as a critical foundation element and antecedent toobjectivity. The accounting literature (e.g., Bonner & Lewis,1990; Libby & Tan, 1994) denes expertise as the knowl-edge, ability, and experience needed to achieve task-specic superior performance. The research literature alsoprovides evidence of a link between professional expertiseand objectivity. For example, the performance review liter-ature in accounting (e.g., Tan & Jamal, 2001) providesevidence that reviewer expertise is directly related to theability to control biases and objectively evaluate the workof subordinates.

    The reliability framework links expertise directly toobjectivity because technical knowledge, ability, andexperience help auditors providing attest services toexercise due diligence and properly ll their responsibili-ties in a competent and procient manner (Taylor et al.,

    6 The reliability framework recognizes that complete objectivity isimpossible given the subjectivity inherent in judgment-based domains(Taylor et al., 2003). Accordingly, objectivity in this context represents theability to manage subjectivity.and avoid proscribed conicts of interest when providingaudit services. Specically, in the context of this study,lenders assessments of auditor integrity should be directlyrelated to their condence that auditors will work to com-ply with existing independence rules to protect the publicinterest. Stated formally:

    Hypothesis 1b. Lenders assessments of auditor integritywill be positively related to their assessments of auditorindependence.

    Given the importance of integrity in the framework andprofession, we also test the extent that auditor integrity isdirectly related to perceived auditor objectivity after con-trolling for integrity effects on perceived expertise andindependence. Objectivity refers to the professionals abil-ity to control the biases, conicts, and inappropriate inu-ences that can undermine professional judgment andreliability. As such, objectivity represents the pursuit of abias-free state of mind that all professionals should pur-Objectivity effects on auditor reliability

    The reliability framework proposes that auditor reliabil-ity ows directly from the pursuit of objectivity reectingunderlying integrity, expertise, and independence whenproviding professional services to clients (DeZoort et al.,2008; Taylor et al., 2003). In this study, we test theframeworks reliability in appearance dimension byevaluating the extent that lenders assessments of auditorobjectivity directly affect their overall assessment of auditor

    7 Expertise without integrity represents an enormous risk for stakehold-ers. DeZoort et al. (2008) highlight this risk as a motivating factor in theirrevision of the Taylor et al. (2003) reliability framework to proposeintegrity as the frameworks sole foundation.Independence effects

    Taylor et al. (2003) describe persisting confusion be-tween independence and objectivity in both the profes-sional literature and in practice. The reliability frameworkdenes independence as it exists within professional stan-dards: as a construct focused on the relationships (e.g.,nancial, family) that professionals have with their clients.Alternatively, objectivity refers to the auditors state ofmind and ability tomanage the cognitive biases and subjec-tivity that inherently affect professional judgments anddecisions. Accordingly, current independence rules focuson proscribing nancial and non-nancial relationshipswith the client that have the potential to create conictsof interest that can impair auditors pursuit of objectivity.

    Although relationship-based independence and objec-tivity are clearly related, the framework highlights theimportance of distinguishing between the constructs andrecognizing that independence is one of several antecedentsto objectivity. We test this proposed relation using lendersassessments of auditor independence and objectivity. Assuggested in the reliability framework, lenders assessmentsof auditor independence shouldbepositively related to theirassessments of auditor objectivity. Stated formally:

    Hypothesis 1e. Lenders assessments of auditor indepen-dence will be positively related to their assessments ofauditor objectivity.perspective, the perceived expertise of auditors should di-rectly affect judgments about auditors ability to understandand evaluate complex issues and manage the judgment anddecision biases that undermine objectivity. Accordingly, wepredict that lenders assessments of auditor expertise shouldbe directly related to their assessments of auditor objectivitywhen evaluating a prospective borrowers credit application.Stated formally:

    Hypothesis 1d. Lenders assessments of auditor expertisewill be positively related to their assessments of auditor

  • sider how auditor reliability and its formative constructs

    to impair auditor independence and undermine nancialreporting quality (e.g., Bazerman & Moore, 2011; Frankel,

    F.T. DeZoort et al. / Accounting, Organizations and Society 37 (2012) 519533 523affect lenders judgments of nancial statement reliability.The accounting literature provides a number of studieslinking the quality of accountants work (e.g., audit quality)to nancial statement reliability (e.g., Fischbacher &Stefani, 2007; Johnson, Kurana, & Reynolds, 2002; Stanley& DeZoort, 2007). For example, Fischbacher and Stefani(2007) found that managers were less likely to manipulatenancial statements in the presence of increased auditoreffort resulting in improvements in audited nancial infor-mation. From a lenders perspective, nancial statementsprovide information that is used to assess a companys cur-rent nancial condition and future cash ows. However,information asymmetry between nancial statement pre-parers and users increases uncertainty for lenders andthe cost of lending (Watts & Zimmerman, 1986). Audit ser-vices should reduce lender uncertainty about nancialinformation provided by the prospective borrower.Accordingly, to the extent that audits are designed to in-crease the credibility of nancial statements, we expectperceived auditor reliability to directly affect lendersassessments of nancial reporting reliability. In addition,lenders assessments of nancial reporting reliabilityshould affect their assessment of a loan applicants creditworthiness. Specically, we expect an inverse relation be-tween perceived nancial reporting reliability and the loanapplicants risk of default on the loan. Stated formally:

    Hypothesis 2. Lenders assessments of auditor reliabilitywill be positively related to their assessments of aprospective borrowers nancial statement reliability.

    Hypothesis 3. Lenders assessments of a prospective bor-rowers nancial statement reliability will be negativelyrelated to their assessments of the borrowers default risk.

    In summary, we test the reliability frameworks pro-posed relations among auditor integrity, expertise, inde-pendence, objectivity, and reliability, including nancialstatement reliability. The conceptual framework in Fig. 2summarizes the studys conceptual model and hypotheses.

    Nonaudit service effects

    Regulators and others have contended that the provi-sion of nonaudit services to audit clients has the potentialreliability. Specically, lenders reliability judgments shouldreect the amount of condence they have in auditors abil-ity to objectively plan and perform the engagement. Accord-ingly, we predict that lender assessments of auditorobjectivity in attest engagements will be directly relatedto their assessments of auditor reliability. Stated formally:

    Hypothesis 1f. Lenders assessments of auditor objectivitywill be positively related to their assessments of auditorreliability.

    Auditor reliability effects on nancial statement reliability

    Beyond testing the reliability framework, we also con-vices is potentially useful during the audit. For example,Joe and Vandervelde (2007) found that auditors gainedknowledge in performing nonaudit tasks. Their resultssuggested that this knowledge could be transferred toaudit tasks (i.e., risk assessment) that may lead to im-proved nancial reporting quality. Additionally, DeZoortand Taylor (2009) found that lender perceptions of CPAreliability and nancial reporting reliability in compila-tion and review engagements did not decrease whenCPAs provided both attest and nonattest bookkeepingand payroll services to a client in violation of indepen-dence rules. Lenders perceived signicantly higher reli-ability in a compilation setting where nonattestservices were provided despite reporting lower perceivedindependence.

    We question whether such ndings will extend to anaudit context given the high level of assurance and the in-creased focus on the importance of independence. The reli-ability framework suggests that perceived auditor integrityand expertise have the potential to offset user concernabout compromised independence resulting from combin-ing audit and nonaudit services. Accordingly, we questionthe extent that the frameworks relations will be affectedwhen auditors provide both audit and nonaudit servicesto clients. Stated formally:

    Research Question. To what extent does the provisionof nonaudit services to audit clients affect lendersframework component assessments and default riskjudgments?

    Method

    Participants

    The studys participants were commercial lender mem-bers of the Risk Management Association (RMA). RMA of-cials reviewed the research materials and e-mailed lendermembers to endorse the study and request participation.The request for participation provided a link to the websitecontaining the research instrument. After 2 weeks, a secondrequest was sent to the RMAmembers asking them for par-ticipation if they had not already completed the study.8

    8 No signicant differences were found among responses from partici-pants who responded after the rst and second requests.Johnson, & Nelson, 2002; Levitt, 2000; Simunic, 1984).For example, Simunic (1984) suggests that the provisionof nonaudit services by auditors leads to economic bondingbetween the auditor and client. Because of this bonding,auditors may be less willing to criticize the work providedfrom these services. Therefore, the resulting objectivityimpairment can have detrimental effects on overall nan-cial reporting quality.

    However, the literature also includes arguments thatthe provision of select nonaudit services to audit clientscan improve the quality of nancial reporting (AICPA,2009a; DeZoort et al., 2008; Joe & Vandervelde, 2007).Information obtained from the provision of nonaudit ser-

  • 524 F.T. DeZoort et al. / Accounting, Organizations and Society 37 (2012) 519533One hundred eighty-seven commercial lenders pro-vided complete responses in the study.9 We cannot deter-mine the specic number of requests sent or the responserate for the study because data collection was facilitatedby RMA contacts. As indicated in Table 1, the participants

    Fig. 2. Conceptual framework. Constructs and hypotheses below the dashed liand revised by DeZoort et al. (2008).

    9 We omitted 18 participants who provided incomplete responses.However, our results are qualitatively similar if these participants are leftin the analysis.

    Table 1Demographics (n = 168).

    Number of cities represented 99Professional designationNo 74%Yes 26%

    GenderFemale 28%Male 72%

    Size of nancial institution (assets)Less than $1 billion 25%$1 billion - $5 billion 26%Greater than $5 billion 49%

    Lending experience (years) 13.99EducationBachelors 67%Masters 33%average 14 years of lending experience and represent 99 cit-ies in the United States and Canada. Almost half (49%) workin nancial institutions with more than $5 billion in assets. Amajority of the participants were male (72%), lacking a pro-fessional license (74%), and possessing a bachelors degree(or equivalent) as their highest level of education (67%).We found no signicant demographic effects.

    Research instrument

    The online research instrument (see Appendix A) wasadapted from DeZoort and Taylor (2009) and administeredusing PsychData.10 After completing an online consent form,participants received general instructions and were toldthey were evaluating an application from a hypotheticalprivately-held tool manufacturer (hereafter Tools) torenance a $1.5 million line of credit. Company backgroundinformation (see Appendix A) described Tools as an average

    ne represent the Reliability Framework introduced by Taylor et al. (2003)

    10 The research instrument was developed in consultation with groups oflenders, researchers, and representatives from the AICPA and the RMA toensure case realism, understandability, and content validity for theframeworks measurement items. We then pretested the instrument with53 MBA students. The development and pretest procedures led to minormodications to the instrument. The study and all research materials wereapproved by the Institutional Review Boards at the researchersuniversities.

  • (medium) risk company with effective internal controls,competent management and directors, and stable nancialhealth and growth in recent years. Participants were toldthat Tools was a new prospective customer for them. Sum-mary nancial information provided unaudited account bal-ances and performance results. The background information

    accounting rm description, participants were told the

    where 1 = Very unrealistic and 9 = Very realistic) and

    four items. The expertise items focused on various dimen-sions of expertise, including experience, knowledge, andability. Similarly, the independence items were written tomeasure perceived independence related to both feescharged and the scope of services provided. The reliabilitystatements were split with two items focused on auditorreliability and two items focused on nancial statementreliability. Response items were randomized and measuredusing a ve-point Likert scale anchored Strongly Dis-agree and Strongly Agree. Each scale contained nega-

    with audit, compilation, or review services for the past

    F.T. DeZoort et al. / Accounting, Organizations and Society 37 (2012) 519533 525understandable (overall mean = 6.61 on a nine-point scalewhere 1 = Very difcult to understand and 9 = Very easyto understand). The results did not differ signicantly be-tween treatment groups.

    Dependent variables

    We developed a series of 20 items to measure the reli-ability framework constructs (see Appendix A). Speci-cally, integrity (items 14), expertise (items 58),independence (items 912), objectivity (items 1316),and reliability (items 1720) were each measured with

    11 Given our focus on auditor characteristics in this study, we developed aborrower and lending scenario that, aside from audit-based manipulation,was as normal as possible. For example, our work with lenders duringinstrument development indicated that the vast majority of commerciallines of credit are secured. However, the benets provided by a secured lineof credit (e.g., the existence and valuation of collateral assets) become lesscertain in the presence of compromised auditor independence.accounting rm has provided audit services for Tools forthe past 3 years. The instrument provided the most recentaudit report providing an unqualied opinion on the nan-cial statements.

    Independent variables

    We manipulated the provision of nonaudit service lev-els at two levels randomly between subjects (see Appen-dix, Panel B). Participants assigned to the audit onlygroup were told that the public accounting rm only pro-vided audit services, while participants in the audit andnonaudit services group were told that their accountingrm provided audit and nonaudit bookkeeping and payrollservices. We developed a service-based independenceproblem in this study rather than a nancial interest orfamily conict independence problem because of recentpolicymaker interest in the area (DeZoort et al., 2008).The focus on auditor provision of bookkeeping and payrollservices as our specic independence problem came afterconsulting with a number of lenders and regulators duringstudy development about types of nonaudit services thatwould create a transparent auditor independence problem.

    Overall, the participants found the research scenario tobe both realistic (overall mean = 6.63 on a nine-point scalealso indicated that the renanced line of credit would be se-cured by Tools accounts receivables and inventory.11

    The instrument then described Tools public accountingrm as a regional public accounting rm that had been inbusiness for 20 years and provided a full menu of servicesto their clients. In addition, the instrument stated that theaccounting rm had approximately 80 professionals work-ing in four ofces in the United States. After reading the3 years. Second, we asked participants to indicate whetherthe accounting rm provided audit services only to Toolsor if it provided both audit and nonaudit bookkeepingand payroll services. The manipulation check questionshad a 92% pass rate, leaving 168 participants for theremainder of the analysis after excluding individuals whomissed at least one manipulation check question.12

    After the manipulation checks, we asked participants toassess the default risk associated with approving Tool Com-panys application to renance its line of credit using a nine-point scale anchored 1 = No Risk and 9 = A Great Deal ofRisk.13 The nding of an overall loan applicant default riskmean of 5.06 indicates that the participants assessed defaultrisk for Tools at a moderate level as intended in the case. Wealsoasked for assessments of theamountof assuranceprovidedby audits, reviews, and compilations and found signicant ex-pected differences among all three engagement types (auditmean = 7.58, review mean = 5.08, compilation mean = 3.47 ona scale anchored 1 = no assurance and 9 = total assurance)at the 0.01 level. The instrument concluded with a series ofdemographics questions used to evaluate the sample and as-sess potential differences due to individual characteristics.

    Results

    Descriptive results and measure validation

    Table 2 provides descriptive statistics for the measure-ment items used in the study. We used a multitrait matrix(Campbell & Fiske, 1959) to evaluate the construct validityof the reliability framework measures used in the study.14

    12 The studys results are qualitatively similar if all of the participants areincluded in the analysis.13 The literature (e.g., Booth, 1992; Minnis, 2011) provides evidence thatlender reliance on audited nancial statements is particularly high whenevaluating risk and setting interest rates.14 Construct validity is the extent that a scale measures the theoreticalconstruct it purports to measure. Campbell and Fiske (1959) describeconstruct validity in terms of convergent validity (i.e., the degree thatmeasures of theoretically related construct are related in fact) anddiscriminant validity (i.e., the degree that measures of theoretically distinctconstructs are not related in fact).tively worded items to control for response bias.

    Manipulation checks and other questions

    We used two multiple-choice questions to assess par-ticipants understanding of the studys experimental mate-rials. First, we asked participants to indicate whether theaccounting rm in the scenario provided Tools Company

  • eresta

    526 F.T. DeZoort et al. / Accounting, Organizations and Society 37 (2012) 519533Table 2Descriptive statistics for framework measures.

    Framework component

    IntegrityAPC makes truthful claimsI do not believe what professionals at APC tell mea

    I trust the professionals at APCThe professionals at APC are honest

    ExpertiseAPC has a great deal of experienceAPC lacks the expertise needed in the Tools engagementa

    APC has the knowledge needed to complete the Tools engagementAPC has the ability to do the Tools engagement

    IndependenceAPC is independent in its relationship with ToolsThe scope of services provided by APC for Tools creates a conict of intThe fees paid by Tools to APC create a conict of interesta

    APC has no conict of interest in the Tools engagement

    ObjectivityAPC professionals are biased when providing services to ToolsaFor example, we examined correlations among all of themeasurement items to assess their convergent and discrim-inant validity related to the frameworks formative con-structs. The results provide evidence of validity, withcoefcients consistently higher among items designed tomeasure the same construct than among items measuringdifferent constructs. We also nd evidence of internal con-sistency (reliability) for the measures with Cronbachs alphaexceeding 0.800 for all of the framework constructs.

    Table 3 presents pairwise correlations among theframework constructs and default risk judgments. The re-sults show signicant correlations among the frameworkconstructs and nancial statement reliability (p < 0.01).Pairwise correlations among the framework variablesrange from 0.349 (IndependenceExpertise) to 0.704 (Audi-tor ReliabilityF/S Reliability). As expected, the results alsoshow consistent negative correlations between the frame-works constructs and default risk.

    APC professionals provide unbiased services to ToolsAPCs professionals lack objectivity in their work related to Toolsa

    APCs professionals are objective when providing services to ToolsAuditor ReliabilityI can rely on APCs work when evaluating Tools credit applicationAPCs work with Tools Company should not be relied upona

    F/S ReliabilityTools nancial statements can be relied upon when making a credit decisionI cannot rely on Tools nancial statementsa

    All responses for the reliability framework (i.e., integrity, expertise, independence1 = Strongly Disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, and 5 = Stron

    a Indicates the negatively-worded item mean score was reverse-coded for con

    Table 3Correlation matrix.

    Integrity Expertise Independence

    IntegrityExpertise 0.462Independence 0.452 0.349Objectivity 0.548 0.499 0.670Auditor reliability 0.525 0.477 0.414F/S Reliability 0.497 0.406 0.369Default risk 0.339 0.294 0.293

    Notes: Bold (Pearson) correlations indicate signicance at 0.01 level.Mean SD Cronbachs a

    0.8573.68 0.6153.99 0.7023.65 0.6753.47 0.650

    0.8253.57 0.6643.79 0.6593.75 0.6443.80 0.623

    0.8713.66 0.9783.59 0.9313.65 0.8503.44 0.849

    0.8133.56 0.894SEM results

    We used AMOS 18 to conduct structural equation mod-eling (SEM) with maximum likelihood estimation to testthe hypothesized relations summarized in Fig. 2. Table 4provides the measurement model results. We identiedeach latent variable by setting the loading of one measure-ment item for each framework component to 1 because thelatent variables are not measured directly. The conrma-tory factor analysis provides evidence of convergent valid-ity (Hair, Anderson, Tatham, & Black, 1995) by revealingsignicant standardized loadings that are consistentlygreater than 0.700 (p < 0.001 for all items).

    Fig. 3 provides the t statistics and standardized param-eter estimates for the structural model. Overall, the resultsindicate that the model ts the data well (v2(df) =14.694(10), p = 0.144, v2/df = 1.469, GFI = 0.976, AGFI =0.933, CFI = 0.993, RMSEA = 0.051 (90% CI = 00.102),

    3.59 0.8143.76 0.7753.61 0.712

    0.8163.84 0.7523.82 0.766

    0.8893.98 0.8523.97 0.823

    , objectivity, reliability) items are measured using a ve-point scale wheregly Agree.sistency.

    Objectivity Auditor reliability F/S Reliability

    0.5900.562 0.704

    0.337 0.329 0.345

  • Table 4Measurement model.

    Framework component

    IntegrityAPC makes truthful claimsI do not believe what professionals at APC tell mea

    I trust the professionals at APCThe professionals at APC are honest

    ExpertiseAPC has a great deal of experienceAPC lacks the expertise needed in the Tools engagementa

    APC has the knowledge needed to complete the Tools engagementAPC has the ability to do the Tools engagementIndependence

    eresta

    cision

    F.T. DeZoort et al. / Accounting, Organizations and Society 37 (2012) 519533 527SRMR = 0.051).15 In addition, the sample-to-parameter ratio

    APC is independent in its relationship with ToolsThe scope of services provided by APC for Tools creates a conict of intThe fees paid by Tools to APC create a conict of interesta

    APC has no conict of interest in the Tools engagement

    ObjectivityAPC professionals are biased when providing services to Toolsa

    APC professionals provide unbiased services to ToolsAPCs professionals lack objectivity in their work related to Toolsa

    APCs professionals are objective when providing services to Tools

    Auditor reliabilityI can rely on APCs work when evaluating Tools credit applicationAPCs work with Tools Company should not be relied upona

    F/S statement reliabilityTools nancial statements can be relied upon when making a credit deI cannot rely on Tools nancial statementsa

    All t-values are signicant at p < 0.001.a Indicates that a parameter is xed at 1.0 in the original solution.of 6.72 (168/25) suggests the sample size is reasonable forthe specied model.16

    The rst three hypotheses (H1aH1c) predict that lend-ers assessment of auditor integrity will affect their assess-ments of auditor expertise, independence, and objectivity.As predicted in H1a, the models path coefcients in Fig. 3indicate that Integrity is positively associated with Expertise(path coefcient = 0.562, p < 0.001). The results also indi-cate a signicant positive link between Integrity and Inde-pendence (path coefcient = 0.552, p < 0.001), providingsupport for H1b. Finally, the signicant positive link be-tween Integrity and Objectivity (path coefcient = 0.286,p < 0.001) provides support for H1c. These results providestrong evidence that integrity is the foundation of the reli-ability framework.

    The results in Fig. 3 also show the effects of lendersassessments of auditor expertise and independence on per-

    15 The literature (e.g., Hooper, Coughlan, & Mullen, 2008; Iacobucci, 2010;Kline, 2011) describes a wide variety of model t indices and somedisparity in opinion about thresholds for good t. However, commonrules of thumb for acceptable model t are that the p-value for the chi-square test statistic should be greater than 0.05 (Iacobucci, 2010); GFI,AGFI, and CFI should be greater than 0.90 (Bentler & Bonnet, 1980; Hu &Bentler, 1998); RMSEA should be less than 0.06 (Hooper et al., 2008; Kline,2011); v2/df should be less than 5 (Arbuckle & Wothke, 1999); and SRMRshould be less than 0.08 (Hooper et al., 2008; Hu & Bentler, 1998; Iacobucci,2010; Kenny, 2011).16 The literature (e.g., Grafton, Lillis, & Widener, 2010; Hair et al., 1995)suggests that a subject-to-parameter ratio of 5 or more is adequate forstructural equation modeling.ceptions of auditor objectivity. As predicted in H1d, we

    Standardized loading t-Value

    0.707 a

    0.767 7.5330.779 7.6000.753 7.462

    0.706 a

    0.812 8.8160.862 9.1850.827 8.938

    0.850 a

    0.872 14.2410.758 11.4570.825 13.053

    0.712 a

    0.789 8.7930.781 8.7210.778 8.695

    0.811 a

    0.836 13.000

    0.893 a

    0.897 16.308nd a signicant link between Expertise and Objectivity(path coefcient = 0.156, p = 0.04). The signicant positiverelation between Independence and Objectivity (path coef-cient = 0.580, p < 0.001) provide support for H1e. Finally,the ndings support H1f by showing that perceived Objec-tivity directly affects perceived Auditor Reliability (pathcoefcient = 0.358, p < 0.001). These ndings providestrong support for the relations posited in the auditor reli-ability framework.

    The structural model results in Fig. 3 also highlight sev-eral links among the framework constructs that extend thehypothesis tests and provide additional insight into howthe reliability framework works. For example, althoughthe reliability framework predicts an indirect relationshipbetween Integrity and Auditor Reliability through Objectiv-ity, the results clearly show that a direct relationship alsoexists. Specically, the signicant direct link betweenIntegrity and Auditor Reliability (path coefcient = 0.253,p < 0.01) provides further evidence of the inuence ofintegrity in the framework. Alternatively, we nd insignif-icant direct links with Auditor Reliability for both Expertiseand Independence at the 0.05 level (path coef-cients = 0.101 and 0.081, respectively).

    In addition to providing evidence in support of the audi-tor reliability framework, the results also provide evidencethat auditor reliability strongly inuences lenders assess-ments of nancial statement reliability and, in turn,borrower default risk. Specically, Fig. 3 results show thatH2 is supported by the signicant positive linkbetween Auditor Reliability and F/S Reliability (path

  • 528 F.T. DeZoort et al. / Accounting, Organizations and Society 37 (2012) 519533coefcient = 0.802, p < 0.001). In addition, we nd theexpected H3 negative relation between F/S Reliability andDefault Risk (path coefcient = 0.361, p < 0.001). Theseadditional ndings highlight the frameworks potential togo beyond understanding users perceptions of auditorreliability and its formative constructs to also understand-ing the extent that auditor reliability affects nancial state-ment reliability and important user judgments.

    Nonaudit service effects

    We evaluated nonaudit service effects on each of theframework constructs in the structural model. Table 5, Pa-nel A, provides the means and standard deviations for theframework components across NONAUDIT condition. Uni-variate comparisons indicate that perceived CPA indepen-dence, objectivity, and auditor reliability are signicantlylower (p < 0.01 in all cases) when auditors provide nonauditbookkeeping and payroll services in addition to nancialstatement audit services. The NONAUDIT manipulationdid not signicantly affect perceived auditor integrity,expertise, nancial statement reliability, or default risk.

    The ANCOVA results in Table 5, Panel B, indicate that theNONAUDIT treatment did not signicantly affect lendersassessments of auditor integrity or expertise. However, asexpected, the lenders did report signicantly higher inde-pendence for auditors who provide only audit services to

    Fig. 3. Structural model results. Model t: v2(df) = 14.694(10), p = 0.144), v2/df =0.102), SRMR = 0.051. , , Signicant at the 0.001, 0.01, and 0.05 levels, respthe loan applicant than for auditors who provide bothaudit and nonaudit services (audit mean = 16.07 vs. audit/nonaudit mean = 13.04, p < 0.001). Interestingly, despitethe decrease in perceived independence when auditors alsoprovide nonaudit services, we nd no correspondingdecrease in lenders assessments of auditor objectivity,auditor reliability, or nancial statement reliability aftercontrolling for underlying framework variables.

    We also conducted a multigroup moderation test inAMOS that compares structural model performance forthe two nonaudit services groups. Specically, we used achi-square difference test to compare an unconstrainedmodel (where structural paths were allowed to differ forthe two nonaudit service groups) to a constrained model(where structural paths were constrained to be the samefor the two nonaudit service groups). Table 5, Panel C, re-ports the path comparison results for the two nonauditservices groups. The chi-square difference test reveals asignicant difference in model performance for the twogroups (p < 0.05). Follow-up path comparisons reveal thatNONAUDIT signicantly moderates the IntegrityIndepen-dence path (z = 3.642, p < 0.001) and the ObjectivityAuditorReliability path (z = 3.164, p < 0.01). However, the paths aresignicant (p < 0.01) for both groups and simply strongerfor the nonaudit services group than for the no nonauditservices group. No other signicant between-group pathdifferences were found.

    1.469, GFI = 0.976, AGFI = 0.933, CFI = 0.993, RMSEA (90% CI) = 0.051 (0 ectively. Dashed lines represent insignicant links.

  • ided

    dence

    **

    F.T. DeZoort et al. / Accounting, Organizations and Society 37 (2012) 519533 529Table 5Nonaudit service effects.

    Panel A: Means (standard deviations)Nonaudit services prov

    No

    Integrity 15.05 (1.91)Expertise 14.95 (2.16)Independence 15.74 (1.87) ***

    Objectivity 15.12 (1.97) ***

    Auditor reliability 7.96 (1.06) **

    F/S Reliability 8.15 (1.25)Default risk 4.88 (1.31)

    Panel B: ANCOVA resultsDependent variable

    Integrity Expertise Indepen

    Test variableNonaudit 1.642 0.516 43.361*Supplemental analysis

    In addition to extending the DeZoort et al. (2008) reli-ability framework in Fig. 1 to test both direct and indirecteffects of integrity, expertise, and independence on reli-ability, we also evaluated a number of alternative struc-tural models to assess their relative performance. Forexample, we tested the structural model without nancialstatement reliability (H2) and default risk (H3) to evaluateFig. 1 reliability framework initially proposed in DeZoortet al. (2008). The reduced model had similar t (e.g.,v2(df) = 10.363(5), p = 0.066, v2/df = 2.073, GFI = 0.980,AGFI = 0.916, CFI = 0.992, RMSEA = 0.060 (90% CI = 0.0000.012), SRMR = 0.0407) and path signicance, suggestingthat the base reliability proposed in the literature standsreasonably on its own and that nancial statement reliabil-ity and default risk are not driving model t.

    CovariatesIntegrity 59.997* 72.898***

    ExpertiseIndependenceObjectivityAuditor reliability

    Observations 168 168 168F 1.642 30.001*** 64.518***

    Adj. R2 .014 .278 .453

    Panel C: Structural model path comparisonsPath coefcients

    No nonaudit services provi

    LinkIntegrity? Expertise .443***

    Integrity? Independence .460***

    Integrity? Objectivity .327***

    Integrity? Auditor reliability .262*

    Expertise? Objectivity .176*

    Independence? Objectivity .398***

    Expertise? Auditor reliability .158Independence? Auditor reliability .142Objectivity? Auditor reliability .227**

    Auditor reliability? F/S reliability .735***

    Auditor reliability? Default risk .352***

    Variables denitions: Nonaudit = provision of nonaudit services, with 1 = only au* Signicant at the 0.05 levels.** Signicant at the 0.01 levels.*** Signicant at the 0.001 levels.Yes Overall

    14.63 (2.16) 14.84 (2.04)14.84 (2.17) 14.90 (2.16)12.92 (3.54) 14.35 (3.15)13.92 (2.95) 14.52 (2.57)7.48 (1.49) 7.73 (1.31)7.86 (1.82) 8.01 (1.56)5.17 (1.41) 5.02 (1.36)

    Objectivity Auditor reliability F/S Reliability

    1.903 0.043 0.683We also tested a variety of alternative models that mod-ify the framework relations posited in Taylor et al. (2003)and DeZoort et al. (2008) to assess their relative perfor-mance. First, we compared our main model to an alterna-tive model with independence, integrity, and expertisespecied as the three foundational constructs. This alterna-tive model is inferior to the main model (AIC = 189.585 vs.50.694 for the main model) and poorly t (e.g., v2(df) =139.282(12), p = 0.000, v2/df = 11.607, GFI = 0.769,AGFI = 0.568, RMSEA = 0.244 (90% CI = 0.2110.278),SRMR = 0.268).17 Second, we tested an alternative modelwith independence (rather than objectivity) as the primary

    15.253*** 9.802** 2.5361.390 2.341 0.13789.334*** 1.979 0.187

    25.654*** 4.214*

    97.699***

    168 168 16869.912*** 42.730*** 58.847***

    .645 .583 .675

    ded Nonaudit services provided z-Value

    .673*** 1.269

    .648*** 3.642***

    .263** .157

    .283** .570

    .138 .799

    .630*** 1.010.005 1.578.009 1.131.661*** 3.164**

    .881*** 1.673.350*** .755

    dit services provided and 2 = audit and nonaudit services provided.

    17 The AIC (Akaike Information Criterion) is a commonly-used predictivet index for comparing alternative non-hierarchical models estimated withthe same data (Kline, 2011). The model with the smallest AIC value hasbetter t and the greatest chance to replicate.

  • mediator. The ndings indicate that this alternative modelalso is inferior to the main model (AIC = 116.041) and poorlyt (e.g., v2(df) = 86.041 (10), p = 0.000, v2/df = 8.604,GFI = 0.891, AGFI = 0.765, RMSEA = 0.183 (90% CI = 0.1480.221), SRMR = 0.147). Finally, we evaluated alternative

    Specically, although the provision of nonaudit services

    530 F.T. DeZoort et al. / Accounting, Organizations and Society 37 (2012) 519533to an audit client decreases lenders assessments of auditorindependence, it does not also decrease assessed auditorobjectivity or reliability.

    Collectively, this study has a number of implications forresearch, policy, and practice. From a research perspective,the results extend the reliability literature in accounting byproviding initial direct empirical evidence about the valid-ity of the reliability framework in an audit context. DeZoort

    18 We tested alternative models with independence and objectivity itemscombined into one factor because the two constructs are closely related inthe literature (Taylor et al., 2003) and because exploratory factor analysisprovided some evidence that the items for the two constructs loaded onone factor.models with expertise and integrity as the foundation con-structs (DeAngelo, 1981; Ganguly et al., 2007), with onlyindependence or expertise specied as the foundation con-struct, and with independence and objectivity combined asone factor.18 Similar to the original Taylor et al. (2003) mod-el, these other alternative models also had higher AIC scoresthan the main model and considerably weaker goodness oft results.

    Discussion and conclusion

    This study extends the reliability literature in account-ing (e.g., DeZoort et al., 2008; Taylor et al., 2003) by provid-ing an initial empirical test of the reliability framework inan audit context. Overall, the results provide support forthe frameworks posited relations. For example, the pri-mary and alternative model results provide strong evi-dence that integrity is the foundation of the framework,with lenders assessments of auditor integrity directlyaffecting their assessments of auditor expertise, indepen-dence, objectivity, and reliability. The ndings also indicatethat lenders assessments of auditor expertise and inde-pendence directly affect their assessments of auditorobjectivity, which in turn directly affects assessments ofauditor reliability.

    In addition to testing the frameworks primary rela-tions, we also consider the effects of auditor reliability onlenders assessments of nancial reporting reliability andcorresponding default risk. As predicted, the results reveala signicant positive relation between lenders assess-ments of auditor reliability and their assessments of over-all nancial reporting reliability for the credit applicant inthe case. Further, we nd a signicant inverse relation be-tween lenders assessments of nancial reporting reliabil-ity and borrower default risk, providing initial evidencelinking the reliability framework to critical business judg-ments by users of nancial statements and auditorservices.

    Finally, the studys results provide evidence that thereliability framework is robust when auditors provide bothaudit and nonaudit services to the prospective borrower.and Taylor (2009) provided initial support for the frame-work in compilation and review engagements, but the lit-erature presents serious questions about the frameworksvalidity in an audit context (AICPA, 2008). Our results re-lated to the frameworks formative constructs help synthe-size and extend the extant literature on the effects ofintegrity on expertise (e.g., Hardwig, 1994), integrity onindependence (e.g., Sweeney & Roberts, 1997), expertiseon objectivity (e.g., Tan & Jamal, 2001), and auditor reli-ability on nancial reporting reliability (e.g., Fischbacher& Stefani, 2007; Johnson et al., 2002; Stanley & DeZoort,2007). The studys ndings also contribute to the indepen-dence literature in accounting (see Anandarajan, Kleinman,and Palmon (2010) for a recent review) by distinguishingempirically between independence and objectivity andproviding evidence that users do not perceive indepen-dence to be the most important ethical construct for audi-tors in pursuit of reliability. Further, our nding that theprovision of nonaudit services to audit clients did not neg-atively affect lender perceptions of auditor objectivity andreliability are consistent with DeZoort and Taylors (2009)ndings in compilation and review engagements and com-plement prior research suggesting functional effects ofcombining audit and nonaudit services for clients (e.g.,Ashbaugh, LaFond, & Mayhew, 2003; Davis & Hollie,2008; Dopuch, King, & Schwartz, 2003; Joe & Vandervelde,2007; Kinney, Palmrose, & Scholz, 2004).

    From policy and practice perspectives, our results relatedirectly to the ongoing debate (e.g., Bazerman & Moore,2011; Gul, Tsui, & Dhaliwal, 2006; Kohlbeck, Looknanan-Brown, & Trainor, 2010; Nelson, 2006) about widespreadbanning of nonaudit services for audit clients followingthe Sarbanes-Oxley Act of 2002 push to strengthen auditorindependence. In this study, although the provision of non-audit services signicantly affected lenders assessments ofauditor independence, it did not signicantly affect theirassessments of auditor objectivity and reliability.

    Finally, the results reinforce the importance of the rolethat auditor and nancial reporting reliability play in len-der judgment and decision-making. Specically, thestudys ndings highlight the effects of perceived auditorintegrity, expertise, and independence on lenders defaultrisk assessments through their effects on (auditor andnancial reporting) reliability. The results also highlightthat lenders nd auditors credible even when auditor inde-pendence is impaired by providing nonaudit internal con-trol services. Further, the results suggest that theinformation contained in these reports has the potentialto reduce the costs of borrowing.

    We highlight several limitations to consider whenevaluating the studys implications. For example, althoughTaylor et al. (2003) discuss the importance of reliability infact and in appearance, this initial study focuses only onusers perceptions of auditor reliability (i.e., reliability inappearance). Future research is needed to develop mea-sures and test the extent to which the frameworks positedrelations hold when the formative constructs (e.g., integ-rity, expertise, independence) are measured directly usingauditors. In addition, future research also should extendframework assessment across user groups (e.g., investors,analysts) and reliance domains (e.g., regulatory, country)

  • it would be se sAc ventory. Tools rde

    F.T. DeZoort et al. / Accounting, Organizations and Society 37 (2012) 519533 531given fundamental differences in objectives, relationships,and stakes. Although lenders represent a highly relevantaudit user group (Reinstein & Leibowitz, 2010), they pos-sess a different power relationship than absentee owners/investors to the extent that they have the ability to requestand receive more information from borrowers. Alterna-tively, although investors can always seek other invest-ment opportunities, they are relatively limited in theirability to request additional information frommanagement.

    Future research also should evaluate whether theframework results hold across different types of auditorindependence violations. This studys focus on service-based independence involving bookkeeping and payrollservices was motivated by specic policymaker interest.However, questions remain about how users would re-spond to framework measures in settings involving inde-pendence issues related to the provision of alternativenonaudit services, nancial interests, and family conicts.

    Further, although our experiment was framed in termsof a nonpublic company audit to enhance comparabilitywith other studies of the framework in other attest settings(e.g., DeZoort & Taylor, 2009), the possibility exists thatuser perceptions would be systematically different in a set-ting involving a public company to the extent that expo-sure and risk are different. Our framework results in thisstudy are similar the DeZoort and Taylor (2009) ndingsin compilation and review settings, providing support fortheoretical suggestion that the framework should performsimilarly across user groups and domains. However, addi-tional research is needed to further assess these issues.

    Finally, despite our testing of alternative models, thisstudy does not establish that the reliability framework pro-posed in the literature (e.g., DeZoort et al., 2008; Tayloret al., 2003) is the optimal model for organizing key ethicalconstructs in the profession. The theoretical links proposedin the framework are intended to be robust, but future re-search involving triangulated methods should continue tochallenge the conceptual relations proposed in the reliabil-ity framework and continue pursuit of a comprehensiveethical framework that organizes key constructs andguides future research, policy, and practice.

    Acknowledgements

    We gratefully acknowledge the generous support re-ceived from the AICPAs Private Companies Practice Sec-tion and the Risk Management Association. We thankseminar and workshop participants at Case Western Re-serve University, Kennesaw State University, The Univer-sity of Alabama, the 2010 National Auditing Conference,and the 2010 AAA Annual Meeting. We also thank MarkPeecher (editor), Mark Koziel, Chuck Landes, Tom Ratcliffe,and Mark Zmiewski for their generous support throughoutthe entire research process and Doug Boyle, Chris Burant,Tim Fogarty, George Franke, Dana Hermanson, Tim Louw-ers, Doug Prawitt, Jonathan Stanley, Anne Wilkins, andYi-Jing Wu for their helpful comments and suggestions. Fi-nally, we thank participating Risk Management Associationmembers for their willingness to share their valuable timeand insights.Panel. B. Experimental treatments

    APC services providedTools accounting rm is Andrews, Parker, and Carlson, LLP(APC). APC is a regional public accounting rm established20 years ago. The rm provides a full menu of accounting,assurance, and tax services and currently employs approx-imately 80 professionals working in four ofces in the Uni-ted States. APC has provided audit services to Tools for thepast 3 years. In addition, APC also provides nonattest book-keeping and payroll services (or provides no other attestor nonattest services) to Tools. The Tools engagement isadequately staffed and supervised in accordance with rmand AICPA quality control standards. Prior year engage-ments have produced no signicant disagreements. Belowis a copy of the most recent audit report issued by APC:

    AUDIT REPORTTo the Board of DirectorsMidwest Tools & Supply CompanyWe have audited the accompanying balance sheet of

    Midwest Tools & Supply Company as of December 31,counts Receivable and Inbt.has no othe

    The renanced line of cred cured by ToolTotal equity $2.2 million

    Total liabilities $3.0 million

    Current liabilities $1.0 million

    Total assets $5.2 million

    PP&E (net) $2.0 million

    Current assets $2.4 million

    Inventory $1.4 millionNet income $0.5 millionAccounts receivable (net) $1.0 millionAppendix A. Experimental materials

    Panel. A. Background information

    Midwest Tools & Supply CompanyPlease assume that you are evaluating an application fromMidwest Tools & Supply (Tools) Company to renance a$1.5 million line of credit. Tools is a new prospective cus-tomer for your bank and you have not worked with thecompany in any other setting.

    Tools Company BackgroundTools is a privately-held tool manufacturer that sells to dis-tributors and select retailers. Tools has been in business for10 years and employs 60 people. The company has had sta-ble nancial health and growth in recent years. Back-ground research indicates that Tools is an average(medium) risk company with effective internal controlsover nancial reporting and competent management anddirectors.

    Summary (unaudited) Annual Financial Information

    Revenues $6.0 millionPretax income $0.7 million

  • 20 -er enancia -pa m nopinion .

    c gstandard -ic s eaudit to en t.An audi rn -dures th tfo p -ne lreportin nau l esu i lstateme dsi a sev i .W ie rou n

    (15) APCs professionals lack objectivity in their

    Bentler, P., & Bonnet, D. (1980). Signicance tests and goodness-of-t in8,

    6Bo ., f

    nBooth, J. ( g

    thCa , t

    valida ali

    Da , nn y,

    14(3),

    532 F.T. DeZoort et al. / Accounting, Organizations and Society 37 (2012) 519533In our opinion, the nancial statements referred toabove present fairly, in all material respects, the nancialposition of the Company as of December 31, 2007, andthe results of its operations and its cash ows for the yearthen ended in conformity with US generally acceptedaccounting principles.

    Andrews, Parker, and Carlson, LLPFebruary 10, 2008

    Panel. C. Reliability framework questions

    Please respond to the following items related to APC,LLPs professional services for Midwest Tools & SupplyCompany (Tools).

    (1) I trust the professionals at APC.(2) APC makes truthful claims.(3) The professionals at APC are honest.(4) I do not believe what professionals at APC tell

    me.(5) APC has a great deal of experience.(6) APC has the knowledge needed to complete the

    Tools engagement.(7) APC has the ability to do the Tools engagement.(8) APC lacks the expertise needed in the Tools

    engagement.(9) APC is independent in its relationship with

    Tools.(10) APC has no conict of interest in the Tools

    engagement.(11) The scope of services provided by APC for Tools

    creates a conict of interest.(12) The fees paid by Tools to APC create a conict of

    interest.(13) APCs professionals are objective when

    providing services to Tools.(14) APC provides unbiased services to Tools.r opi ion.

    e bel ve that our audit provides a reasonable basis fo

    aluat ng the overall nancial statement presentation

    gnicnts, assessing the accounting principles used annt estimates made by management, as well apport ng the amounts and disclosures in the nancia

    dit ag. Accordingly, we express no such opinion. Aso includes examining, on a test basis, evidencss of the Companys internal control over nancia

    r theat are appropriate in the circumstances, but nourpose of expressing an opinion on the effectiveancia

    t includes consideration of internal control ovel reporting as a basis for designing audit proceancia

    obtain reasonable assurance about whether th

    l statements are free of material misstatemena. Tho

    s generally accepted in the United States of Amere standards require that we plan and perform thWe

    on these nancial statements based on our auditonducted our audit in accordance with auditinnys

    l statements are the responsibility of the Comanagement. Our responsibility is to express as equity, and cash ows for the year then ended. Thes

    07, and the related statement of operations, stockholdk lending decisions. Accounting, Organizations, and Societ235249.nos, P.in baHolt, D., & Imhoff, E. (1989). The use of accounting informatio

    BulletD. T., & Fiske, D. W. (1959). Convergent and discriminantion by the multitrait-multimethod matrix. Psychologicn, 56, 81105.hypompbell1992). Contract costs, bank loans, and the cross-monitorinesis. Journal of Financial Economics, 31, 2541.nner, SAccou00.& Lewis, B. (1990). Determinants of auditor expertise. Journal oting Research, 28(Suppl.), 128.the a588nalysis of covariance structures. Psychological Bulletin, 8work related to Tools.(16) APC is biased when providing services to Tools.(17) I can rely on APCs work when evaluating Tools

    credit application.(18) APCs work with Tools should not be relied

    upon.(19) Tools nancial statements can be relied upon

    when making a credit decision.(20) I cannot rely on Tools nancial statements.

    Negatively-worded item that required reverse coding.All responses measured on a ve-point scale with

    the following labels: Strongly Disagree, Disagree,Neutral, Agree, and Strongly Agree

    Appendix B. Supplementary material

    Supplementary data associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.aos.2012.08.003.

    References

    American Institute of Certied Public Accountants (AICPA) (2011). AICPAProfessional Ethics Executive Committee exposure draft, OmnibusProposal. February 28, 2011. New York, NY: AICPA.

    AICPA (2010). Code of professional conduct. New York, NY: AICPA.AICPA (2009a). Exposure draft: Proposed statements on standards for

    accounting and review services. April 28, 2009. AICPA Accounting andReview Services Committee. New York, NY: AICPA.

    AICPA (2009b). Statement on standards for accounting and review servicesno. 19. Compilation and review engagements. AICPA Accounting andReview Services Committee. New York, NY: AICPA.

    AICPA (2008). Recommendations of the reliability task force: Considerationof an alternative framework for compilation and review engagements.New York, NY: AICPA.

    Anandarajan, A., Kleinman, G., & Palmon, D. (2010). Auditor independenceresearch: Where do we stand? Journal of Accounting, Ethics, and PublicPolicy, 11(1), 131.

    Arbuckle, J., & Wothke, W. (1999). Amos 4.0 users guide. Chicago, IL:SmallWaters Corporation.

    Ashbaugh, H., LaFond, R., & Mayhew, B. (2003). Do nonaudit servicescompromise auditor independence? Further evidence. The AccountingReview, 78(3), 611639.

    Ashton, R. H., & Ashton, A. H. (1995). Judgment and decision-makingresearch in accounting and auditing. New York, NY: CambridgeUniversity Press.

    Bartlett, R. (1991). Perceived levels of assurance of bankers and CPAs: Acomparison. Advances in Accounting, 9, 205225.

    Bazerman, M., & Moore, D. (2011). Is it time for auditor independenceyet? Accounting, Organizations and Society, 36(45), 310312.

    Beaulieu (1994). Commercial lenders use of accounting information ininteraction with source credibility. Contemporary Accounting Research,10(2), 557586.

  • Davis, S., & Hollie, D. (2008). The impact of nonaudit service fee levels oninvestors perception of auditor independence. Behavioral Research inAccounting, 20(1), 3144.

    DeAngelo, L. E. (1981). Auditor size and audit quality. The Journal ofAccounting and Economics, 3, 180199.

    DeZoort, T., Morgan, D., Ratcliffe, T., & Taylor, M. (2008). Refocusing onreliability. The Journal of Accountancy (October), 7478.

    DeZoort, T., & Taylor, M. (2009). An empirical test of the reliabilityframework: Evidence from banking professionals. Prepared by DeZoort,T. and Taylor, M. for the American Institute of Certied PublicAccountants Private Company Practice Section. New York, NY: AICPA.

    Dopuch, N., King, R., & Schwartz, R. (2003). Independence in appearanceand in fact: An experimental investigation. Contemporary AccountingResearch, 20(1), 79114.

    Fischbacher, U., & Stefani, U. (2007). Strategic errors and audit quality: Anexperimental investigation. The Accounting Review (May), 679704.

    Frankel, R., Johnson, M., & Nelson, K. (2002). The relation betweenauditors fees and nonaudit services and earnings management. TheAccounting Review, 77, 71105.

    Ganguly, A., Herbold, J., & Peecher, M. (2007). Assurer reputation forcompetence in a multiservice context. Contemporary AccountingResearch (Spring), 133177.

    Gay, G., Schelluch, P., & Baines, S. (1998). Perceptions of messages

    Johnson, D., Pany, K., & White, R. (1983). Reports and the loan decision:Actions and perceptions. Auditing: A Journal of Practice and Theory, 2,3851.

    Johnson, V., Kurana, I., & Reynolds, K. (2002). Audit-rm tenure and thequality of nancial reports. Contemporary Accounting Research, 19(4),637660.

    Kenny, D. (2011). Measuring model t. .

    Kinney, W. R., Palmrose, Z., & Scholz, S. (2004). Auditor independence,non-audit services and restatements: Was the US government right.Journal of Accounting Research, 42(3), 561588.

    Kline, R. (2011). Principles and practice of structural equation modeling (3rded.). New York, NY: The Guilford Press.

    Kohlbeck, M., Looknanan-Brown, V., & Trainor, J. (2010). Did the SOX banon non-audit services go far enough or too far? Evidence from investorsperceptions of auditor independence. Working paper. Florida AtlanticUniversity.

    Lee, T. A. (1993). Corporate audit theory. London, England: Chapman &Hall.

    Levitt, A. (2000). Renewing the covenant with investors. Remarks deliveredat NYU Center for Law and Business, New York, NY, May 10.

    Libby, R., & Tan, H. T. (1994). Modeling the determinants of audit

    F.T. DeZoort et al. / Accounting, Organizations and Society 37 (2012) 519533 533conveyed by review and audit reports. Accounting, Auditing, andAccountability Journal, 11(4), 472494.

    Grafton, J., Lillis, A., & Widener, S. (2010). The role of performancemeasurement and evaluation in building organizational capabilitiesand performance. Accounting, Organizations and Society, 35(7), 689706.

    Gul, F., Tsui, J., & Dhaliwal, D. (2006). Non-audit services, auditor qualityand the value relevance of earnings. Accounting and Finance, 46,797817.

    Hair, J., Anderson, R., Tatham, R., & Black, W. (1995). Multivariate dataanalysis. Englewood Cliffs, NJ: Prentice Hall.

    Hardwig, J. (1994). Toward an ethics of expertise. In D. Wueste (Ed.),Professional ethics and social responsibility. Lanham, MD: Rowman &Littleeld.

    Holder-Webb, L., & Sharma, D. S. (2010). The effect of governance oncredit decision and perceptions of reporting reliability. BehavioralResearch in Accounting, 22, 120.

    Hooper, D., Coughlan, J., & Mullen, M. (2008). Structural equationmodeling: Guidelines for determining model t. The ElectronicJournal of Business Research Methods, 6(1), 5360.

    Hu, L., & Bentler, P. (1998). Cutoff criteria for t indexes in covariancestructure analysis: Conventional criteria versus new alternatives.Structural Equation Modeling, 6(1), 155.

    Iacobucci, D. (2010). Structural equation modeling: Fit indices, samplesize, and advanced topics. Journal of Consumer Psychology, 20, 9098.

    International Federation of Accountants (IFAC) (2010). Handbook of thecode of ethics for professional accountants. New York, NY: IFAC.

    Jamal, K., & Sunder, S. (2011). Is mandated independence necessary foraudit quality? Accounting, Organizations and Society, 36(45), 284292.

    Jeppesen, K. K. (1998). Reinventing auditing, redening consulting andindependence. European Accounting Review, 7(3), 517539.

    Joe, J., & Vandervelde, S. (2007). Do auditor provided non-audit servicesimprove audit effectiveness? Contemporary Accounting Research,24(2), 467487.expertise. Accounting, Organizations and Society, 19(8), 701716.Lindberg, D. L., & Beck, F. B. (2004). Before and after Enron: CPAs views on

    auditor independence. The CPA Journal, 74(11), 3640.Mautz, R. K., & Sharaf, H. A. (1961). The philosophy of auditing. Sarasota, FL:

    American Accounting Association.Minnis, M. (2011). The value of nancial statement verication in debt

    nancing: Evidence from private US rms. Journal of AccountingResearch, 49(2), 457506.

    Montgomery, R. H. (1940). Auditing theory and practice (6th ed.). NewYork, NY: The Ronald Press Company.

    Nelson, M. (2006). Ameliorating conicts of interest in auditing: Effects ofrecent reforms on auditors and their clients. Academy of ManagementReview, 31(1), 3042.

    Reinstein, A., & Leibowitz, M. (2010). New standards for CPA compilationsand reviews: Some implications for bankers. The RMA Journal, 93(3),3032.

    Stanley, J., & DeZoort, T. (2007). Audit rm tenure and nancialrestatements: An analysis of industry specialization and fee effects.Journal of Accounting and Public Policy, 26, 131159.

    Simunic, D. (1984). Auditing, consulting, and auditor independence.Journal of Accounting Research, 22, 679702.

    Sweeney, J., & Roberts, R. (1997). Cognitive moral development andaccountant independence. Accounting, Organizations, and Society, 22,337352.

    Tan, H.-T., & Jamal, K. (2001). Do auditors objectively evaluate theirsubordinates work? The Accounting Review, 76(1), 99110.

    Taylor, M., DeZoort, T., Thomas, M., & Munn, E. (2003). A proposedframework emphasizing auditor reliability over auditorindependence. Accounting Horizons (September), 257266.

    Watts, R., & Zimmerman, J. (1986). Positive accounting theory. EnglewoodCliffs, NJ: Prentice-Hall.

    Yardley, J. (1989). Lenders and CPAs perceptions of the assuranceprovided by prescribed procedures. Auditing: A Journal of Practice andTheory, 9, 4156.

    A test of the auditor reliability framework using lenders judgmentsIntroductionTheory and hypothesesIntegrity effectsExpertise effectsIndependence effectsObjectivity effects on auditor reliabilityAuditor reliability effects on financial statement reliabilityNonaudit service effects

    MethodParticipantsResearch instrumentIndependent variablesDependent variablesManipulation checks and other questions

    ResultsDescriptive results and measure validationSEM resultsNonaudit service effectsSupplemental analysis

    Discussion and conclusionAcknowledgementsAppendix A Experimental materialsPanel A. Background informationPanel B. Experimental treatmentsPanel C. Reliability framework questions

    Appendix B Supplementary materialReferences