Audit Fees Non Audit Fees and Auditor

ABACUS, Vol. 44, No. 3, 2008

doi: 10.1111/j.1467-6281.2008.00263.x

284

© 2008 The AuthorsJournal compilation © 2008 Accounting Foundation, The University of Sydney

Blackwell Publishing AsiaMelbourne, AustraliaABACAbacus0001-30721467-6281© 2008 Accounting Foundation, Unviersity of SydneyXXX

ORIGINAL ARTICLES

U.K. AUDITOR REMUNERATION AND GOING-CONCERN REPORTSABACUS

ILIAS G. BASIOUDIS, EVANGELOS PAPAKONSTANTINOU AND MARSHALL A. GEIGER

Audit Fees, Non-Audit Fees and Auditor Going-Concern Reporting Decisions in the

United Kingdom

The accounting profession has come under increased scrutiny over recentyears about the growing number of non-audit fees received from audit clientsand the possible negative impact of such fees on auditor independence.The argument advanced is that providing substantial amounts of non-auditservices to clients may make it more likely that auditors concede to thewishes of the client management when difficult judgments are made.Such concerns are particularly salient in the case of reporting decisionsrelated to going-concern uncertainties for financially stressed clients.

This study empirically examines audit reports provided to financiallystressed companies in the United Kingdom and the magnitude of auditand non-audit service fees paid to the company’s auditors. We find thatthe magnitude of both audit fees and non-audit fees are significantlyassociated with the issuance of a going-concern modified audit opinion.In particular, financially stressed companies with high audit fees are

more

likely to receive a going-concern modified audit opinion, whereascompanies with high non-audit fees are

less

likely to receive a going-concern modified audit opinion. Additional analyses indicate that theresults are generally robust across alternative model and variable specifica-tions. Overall, evidence supports the contention that high non-audit feeshave a detrimental effect on going-concern reporting judgments forfinancially stressed U.K. companies.

Key words:

Auditing: audit fees, opinion; Going concern; Non-auditfees; U.K.

In this study we examine the association between audit service fees and non-auditservice (NAS) fees in the U.K. and the audit firm’s final decision regarding thetype of opinion provided to a financially stressed client. Specifically, evidence fromprior research has found that the amount of non-audit fees are not significantly

I

lias

G. B

asioudis

([email protected]) is a Senior Lecturer and

Evangelos Papakonstantinou

agraduate of the Aston Business School, University of Aston.

Marshall A. Geiger

([email protected])occupies the Joseph A. Jennings Chair in Business in The Robins School of Business, University ofRichmond.We would like to thank Samuel Agyei-Ampomah, Stuart Turley and workshop participants at theAnnual Congress of the 2006 European Accounting Association, Dublin, the 2006 National AuditingConference, Manchester, and workshop participants at RMIT (Australia) and Aston University fortheir comments on earlier versions of the article. We would also like to thank Graeme Dean and theanonymous reviewers for their insightful comments and recommendations.

U.K. AUDITOR REMUNERATION AND GOING-CONCERN REPORTS

285


related to financially stressed companies receiving or not receiving a going-concernmodified audit opinion in the U.K. (Lennox, 1999) or in the U.S. (DeFond

et al

.,2002; Geiger and Rama, 2003). Yet other research on U.K. audit firms has founda significant association of NAS fees and all types of qualified audit reports (Firth,2002), and that audit services are priced differently in the U.K. compared withAustralia and the U.S. (Basioudis and Francis, 2007). In order to extend this lineof NAS fee research and provide a more robust examination of the possiblereporting association in the U.K., we examine audit reports for companies infinancial stress and test for the association between the type of audit opinionissued (going-concern modified

1

or not) and the level of both audit fees and NASfees paid by the client company to the audit firm.

Payment of audit and NAS fees directly from the client to the audit firm haslong been a contentious issue (c.f., Mautz and Sharaf, 1961; AICPA, 1978; Simunic,1984; Barkess and Simnett, 1994; Firth, 1997b; Levitt, 1998; DeFond and Francis,2005), and the public accounting profession has come under increased scrutinyabout the large and growing amount of NAS fees received from audit clients andthe potential adverse effect such fees could have on auditor independence(Ezzamel

et al.

, 1996; Lennox, 1999; Weil, 2001; Carcello, 2005). The economicbond of clients paying audit and non-audit fees directly to their external audit firmhas been argued by many to be a substantial impediment to audit independence(Mautz and Sharaf, 1961; Levitt, 1996; Ezzamel

et al.

, 1996; ISB, 2001). Still othershave argued that merely the growth in NAS fees in recent years has strengthenedthis economic bond and substantially increased the threat of impaired auditorindependence (Firth, 1997a, 2002; Ezzamel

et al.

, 1999; Levitt, 2000). That is, auditorswho perform significant NAS for audit clients have an

additional

economicincentive to retain the client, possibly at the risk of deciding difficult issues in theclient’s favour (Kida, 1980; Citron and Taffler, 1992; Geiger

et al.

, 1998; Levitt,2000; Carcello and Neal, 2003). In fact, the legislatures in the U.K. and U.S., aswell as the European Union (Buijink

et al.

, 1996; Federation of EuropeanAccountants, 1996; Department of Trade and Industry, 2003; SEC, 2000a), haveargued that NAS fees have the potential to adversely impact auditor independenceand impair auditor decision-making, especially when those decisions involve asubstantial amount of professional judgment. Such concerns have led to thedisclosure of these fees for public companies in the U.K. beginning in 1991 and inthe U.S. beginning in 2001 (

Companies Act

, 1989; SEC, 2000a, 2000b).This study empirically examines the assertion implied by these legislative

initiatives that U.K. auditors may act more favourably toward those clients fromwhom they receive higher NAS fees. Following prior research focusing on auditfirm reporting decisions, and in order more to properly assess the possible effectof NAS fee influence on audit firms, we examine the audit reports rendered tofinancially stressed companies and the magnitude of NAS fees and audit service

1

In this study we refer to ‘going-concern modified’ audit opinions and intend it to include all typesof going-concern audit opinions (e.g., unqualified, qualified, adverse and disclaimer) allowable inthe U.K.

ABACUS

286


fees paid by such companies to their auditors. Prior researchers (c.f. DeFond

et al.

, 2002; Firth, 2002; Abbott

et al.

, 2003) contend that a thorough examinationof fees paid to audit firms and firm decision-making must jointly include auditservice fees as well as NAS fees. Hence, our study examines if there is an associationbetween the magnitude of both NAS fees and audit fees on audit opinions forfinancially stressed U.K. companies.

If there is a substantial incentive for audit firms to appease clients that payhigher NAS fees, we would expect a negative relationship between NAS fees paidand the issuance of a going-concern modified audit opinion from the auditor.Conversely, several prior audit fee studies find that increased audit service feesare positively associated with auditors rendering modified audit reports (Simunic,1980; Francis, 1984; Francis and Simon, 1987; Davis

et al.

, 1993). These earlierstudies have concluded that in cases where the auditor modifies their report, theytypically do more substantive testing and documentation, and engage in lengthydiscussions and negotiations with the client in order to support their final opinion.This additional work leads to increased audit service fees and results in the positiveassociation of audit fees and report modifications. Accordingly, we expect a similarpositive relation in our study of stressed U.K. companies.

Unfortunately, prior research examining fees paid by U.K. companies and auditfirms’ reporting decisions has either not separately examined going-concernreport modifications, or has not appropriately examined the

concurrent

effect ofboth NAS and audit service fees on reporting decisions. Accordingly, we presentthe first focused study of both NAS and audit service fees and going-concernmodified reporting decisions in the U.K.

As hypothesized, this analysis of financially stressed U.K. companies findsa significant negative association between receiving a going-concern modifiedaudit report and the magnitude of NAS fees, which suggests NAS fees may impairauditor decision-making. The findings are consistent with those of Firth (2002)in his examination of all types of audit report qualifications in the U.K. We alsofind a positive association between audit service fees and going-concern modi-fied opinions, consistent with audit pricing studies that have found modifiedaudit opinions require additional audit work and therefore are associated withincreased audit fees (Bell

et al.

, 2001; Firth, 2002). We then perform further testsand demonstrate that our sample is representative of samples used in earlier U.K.studies and that results are generally robust across alternative model and variablespecifications.

This study answers the calls for additional research in this area (c.f. Lennox,1999; Butler

et al.

, 2004; Carcello, 2005; DeFond and Francis, 2005; Basioudis andFrancis, 2007) by extending the literature in several important ways. First, itdirectly assesses the relationship between U.K. auditors’ going-concern modifica-tion decisions and NAS fees only for financially stressed companies for which thegoing-concern modified reporting decision is most salient (Mutchler

et al.

, 1997;Geiger and Rama, 2003). Second, it extends the analysis of going-concern reportingdecisions in the U.K. and fees paid to the external audit firm to include thesimultaneous examination of audit service fees as well as NAS fees (DeFond


287


et al.

, 2002; Firth, 2002; Geiger and Rama, 2003). Third, it controls for company-specific factors such as default status, cost reduction plans, and issuing new debtor equity securities in the examination of fees and going-concern reportingdecisions (Chen and Church, 1992; Behn

et al.

, 2001). Last, it extends the literatureon audit market differences between countries and provides further evidence thatthe U.K. audit market may not be the same as audit service markets of evenseemingly similar countries like Australia or the U.S. (Simunic, 1980, 1984; Simunicand Stein, 1987, 1996; Chen

et al.

, 2005; Basioudis and Francis, 2007).

AUDITOR REPORTING DECISIONS AND NAS FEES

Auditing literature has long defined auditor independence as the joint probabilityof auditors detecting and reporting errors (DeAngelo, 1981; Watts and Zimmerman,1986). As further discussed by Craswell (1999), the first element of the definitionrelates more to auditor technical competence, and the second more specifically tothe issue of auditor

independence

. Thus, auditor independence depends not onlyon the ability of the auditor to identify problems (i.e., their professional ability),but also their willingness to report those problems appropriately (i.e., theirprofessional independence from the client). Since prior research has found thatauditors typically do not have difficulty identifying cases involving going-concernissues in their audit clients (Kida, 1980; Mutchler, 1985, 1986; Simnett and Trotman,1989), the independence issue is more directly related to auditor reporting decisionson these financial difficulties than the ability of the audit firm identifying theirexistence. Accordingly, numerous researchers have argued that a direct test of theeffects of NAS fees on auditor independence is the examination of NAS fee levelsand auditor reporting decisions (c.f. Craswell, 1999; Reynolds and Francis, 2000;DeFond

et al.

, 2002; Firth, 2002; Geiger and Rama, 2003; DeFond and Francis, 2005).Audit reporting standards (e.g., SAS No. 600 in the U.K.; SAS No. 59 in the

U.S.) require auditors to assess the continued viability of their clients in everyaudit engagement. If there remains substantial doubt in the auditor’s mindregarding the ability of the client to continue in its business as a going concern,even after considering management’s plans and mitigating circumstances, then theauditor must disclose this doubt in their audit opinion. The going-concern assessment,the evaluation of the appropriateness and probability of the success of manage-ment’s plans,

2

and the final decision to issue a modified opinion, all involve highlysubjective judgments on the part of the auditor which could be intentionally orunintentionally affected by the level of fees obtained from the client.

In the U.K. a modified audit report is defined as any audit report other than astandard unqualified (clean) opinion, including any unqualified reports containingexplanatory paragraphs. An unqualified audit report may contain explanatoryparagraphs that provide additional information and that draw attention to issues

2

Mutchler

et al.

(1997) and Behn

et al.

(2001) find that decisions about going-concern modifiedopinions in the U.S. are significantly related to the auditor’s evaluation of management’s plans andmitigating circumstances.

ABACUS

288


affecting the financial statements. In this reporting context, one type of unqualifiedaudit report modification is referred to as the going-concern modified report whichincludes additional explanatory paragraphs relating to fundamental uncertaintiesaffecting the ability of the organization to continue as a going concern in thefuture.

3

Additionally, an auditor might determine that the going-concern uncer-tainties of the client are so pervasive that an objective opinion on the fairness ofthe financial statements is not possible at present, and that they must disclaim anopinion on the current financial statements.

With respect to these final audit reporting decisions, negotiations between theauditor and client regarding the type of audit opinion to be issued are very sensitive,and are made all the more problematic in the case of a financially stressed client(Hopwood

et al.

, 1994; Mutchler

et al.

, 1997). These discussions become particularlycontentious if the client believes that the receipt of a going-concern modifiedaudit opinion will increase their financial difficulties and become a self-fulfillingprophesy regarding their inability to continue in existence (Altman, 1982; Geiger

et al.

, 1998; Pryor and Terza, 2002).

PRIOR RESEARCH

In a large sample study of NAS fees and audit reporting decisions in the U.S.,DeFond

et al.

(2002) conclude that there is no significant association betweenboth audit fees and NAS fees and audit opinions on financially stressed com-panies. However, given their large sample size, DeFond

et al.

(2002) do notinclude important mitigating factors such as client default status, managementplans, or obtaining additional funding that have been identified in prior researchto be associated with audit reporting decisions on stressed companies (Chen andChurch, 1992; Behn

et al.

, 2001). A study by Geiger and Rama (2003) examines asample of sixty-six going-concern modified companies in the U.S. and a matchedsample of financially stressed, non-going-concern modified companies andincludes controls for default status, management plans and additional fundingsources. Using this small-sample, matched-pairs methodology, Geiger and Ramaconclude that audit fees were positively associated with going-concern modifica-tions, but that NAS fees were not significantly associated with going-concernmodification decisions. Thus, prior research in the U.S. has been unable todemonstrate a significant association between NAS fees and auditor’s going-concernmodification decisions.

The relationship between NAS fees and auditor reporting behaviour has alsobeen previously examined in Australia, where audit fee data have been publicly

3

There are several other possible non-going-concern related audit report modifications in the U.K.that auditors may issue in addition to those related to going-concern. Other modified auditreports are issued for reasons including limitations on scope of the examination (lack of auditevidence-disclaimer of opinion), and disagreements regarding the accounting treatment or disclosureof specific items (adverse opinion). These reports may include explanatory paragraphs relatingto fundamental uncertainties not directly affecting the ability of the organization to continue as agoing concern.


289


disclosed since 1980. However, the results of that research are somewhat mixed.For example, Wines (1994) finds evidence of a negative association between NASfees and the issuance of any type of qualified report in a sample of seventy-sixAustralian companies over the period 1980–89. Additionally, Sharma (2001) andSharma and Sidhu (2001) examine a sample of forty-nine bankrupt Australiancompanies and conclude that higher NAS fees were associated with a lowerlikelihood of receiving going-concern modified reports. Yet, Barkess and Simnett(1994) and Craswell (1999) in two large sample studies find no significant associationbetween the level of NAS fees and Australian audit report qualifications.

However, these non-U.K. results are not necessarily unilaterally transferable tothe market for audit services in the U.K. (Simunic and Stein, 1987, 1996; Ezzamel

et al.

, 1999; Basioudis and Francis, 2007). Prior researchers have argued that thereis less audit litigation in the U.K., and more of a principles-based approach tofinancial reporting compared to the U.S. or Australia (Ball

et al

., 2000; Gul

et al

.,2002; Khurana and Raman, 2004). As a result, U.K. auditors may respond differentlywhen faced with similar NAS fee and reporting issues for their financially stressedclients. Hence, direct testing is required to determine whether U.K. auditorsmight respond differently to litigation and reputation concerns than their U.S. andAustralian colleagues.

In the first published study of NAS fees and audit reporting decisions on U.K.firms, Lennox (1999) examined NAS fee disclosures during the period 1988 to1994 surrounding the initial fee disclosure requirements instituted in 1991.

4

Whilethe focus of his study was primarily on the early/on-time/late NAS fee disclosurepatterns of U.K. companies and audit firms, he did test for a relationship betweenNAS fees and audit opinions (which he used as a surrogate for audit firm quality).His results suggest that after controlling for prior year going-concern modifiedaudit reports, NAS fees were not significantly associated with auditors render-ing a going-concern modified opinion to their clients. Since the main focus of theLennox study was not on the relationship between NAS fees and audit opinionsper se, he did not concurrently assess the effect of the level of audit service feeson opinions (DeFond

et al.

, 2002), or control for firm-specific factors found tobe significantly related to audit reporting decisions (Chen and Church, 1992;Behn

et al.

, 2001).A study by Firth (2002) has examined auditor decisions to issue qualified opinions

in the U.K. and the relationship between all qualification decisions in the year1996 and relative level of audit and NAS fees (i.e., fees scaled by total assets). Inthis study, Firth examines all quoted non-financial companies and, as hypothesized,finds a significant negative association between the proportion of NAS fee levelsand receiving a qualified opinion, but a non-significant association between theproportions of audit service fee levels and receiving a qualified opinion. However,while Firth tests for differences in reporting decisions relating to fundamental

4

Earlier studies by Chan

et al.

(1993), Brinn

et al.

(1994), Ezzamel

et al.

(1996, 1999) and Firth(1997a, 1997b) have examined audit fee determinants in the U.K. audit market. However, thesepricing studies have focused on modelling the level of audit fees or the proportion of NAS fees inrelation to audit fees and not the relationship between fees and audit reporting decisions.

ABACUS

290


qualifications and material (but not fundamental) qualifications, he does notspecifically examine the effect of fee levels on auditor going-concern modifiedreporting decisions. Butler

et al.

(2004), however, conclude that the level offinancial stress exhibited by companies receiving going-concern modified auditopinions is significantly greater than by companies receiving other types of auditopinion qualifications, and these financial stress differences necessitate theseparate evaluation of going-concern modified reports from other types of auditopinions. Accordingly, it is unclear whether the results presented in Firth aregeneralizeable to the association of NAS fees and audit firm decisions regardinggoing-concern modified opinions, and in particular, the audit reporting decisionsof financially stressed audit clients.

Further, as noted by prior researchers, results of studies performed on settingsin one country may not be universally transferable to other, even seeminglysimilar countries (c.f., Simunic, 1980, 1984; Simunic and Stein, 1987, 1996; Ezzamel

et al.

, 1999; Carey

et al

., 2008). In fact, recent audit fee research by Basioudis andFrancis (2007) examining the audit fee market in the U.K. concludes that ‘theU.K. results provide a sharp contrast to U.S. and Australia . . . The U.K. resultsreinforce that there are important differences in the pricing of Big 4 industryexpertise across countries, even among a group of countries that is viewed asbeing relatively similar in nature.’ Accordingly, without direct testing, it is unclearwhether the U.K. auditors might respond differently to litigation and reputationconcerns than their U.S. and Australian colleagues causing U.K. auditors toreport differently when faced with similar NAS fee and reporting issues. Thus, weprovide a more robust examination of the U.K. audit market in an effort to bothextend the research on the U.K. audit market, as well as determine if the resultsof NAS fee and reporting studies in other countries also hold in the U.K.

RESEARCH HYPOTHESES

As noted earlier, audit decision-making related to going-concern uncertainties isan area that involves considerable auditor judgment. However, to date there hasnot been a directed, thorough examination of the potential relationship betweenlevels of NAS and audit service fees and auditor going-concern reporting decisionson financially stressed clients in the U.K. Due to the fact that audit opinions forfinancially stressed companies have been an ongoing issue of concern to standard-setters, legislators, the public, and the business press (Ezzamel

et al.

, 1999; Weil,2001; Coffee, 2002; Dietz, 2002), and have been the topic of several governmentalhearings over the years (e.g., U.K. Department of Trade and Industry, 2003), it isimportant to examine empirically whether higher fees received for NAS by U.K.audit firms might impair independence and lead to fewer going-concern modificationdecisions for companies in financial stress. Thus, our first research hypothesis (inalternative form):

H

1

: There is a negative association between the magnitude of non-audit fees and auditopinions modified for going concern.


291


Since clients pay all fees directly to their audit firms, examining one type of auditfee necessitates the examination of all fees paid to the audit firm. Accordingly,prior researchers and regulators (Abdel-Khalik, 1990; SEC, 2000b; DeFond

et al.

,2002; Firth, 2002; Abbott

et al.

, 2003) have argued that examining the magnitudeor impact of NAS fees cannot be properly accomplished without also concurrentlyassessing the level of audit service fees paid by clients. A systematic examinationof fees paid to the audit firm and their effect on reporting decisions should includeboth the level of non-audit fees as well as the level of audit service fees. There-fore, we also examine the association between audit service fees and audit firmreporting decisions. Based on prior audit fee and audit reporting research, weexpect a positive association between audit service fees and going-concernmodified reporting decisions (Simunic, 1980; Francis, 1984; Geiger and Rama,2003). This leads to our second hypothesis (in the alternative form):

H

2

: There is a positive association between the magnitude of audit service fees and auditopinions modified for going concern.

SAMPLE AND DATA

To identify our sample of going-concern modified companies, we begin with allcompanies listed on the London Stock Exchange for the 2003 financial reportingyear. Global Access and FAME databases provide financial and other data.

5

Asreported in Table 1, the initial sample includes 2,200 listed companies that are

5

Global Access is a product by Thomson Publishing Group, and FAME is an acronym for FinancialAnalysis Made Easy, a comprehensive database for U.K. private and publicly listed companiesmaintained by Bureau Van Dijk.

Table 1

SAMPLE SELECTION

No. %

Companies listed in the London Stock Exchange (LSE) and included in FAME database

2,200 100.00

Financial services companies (i.e., banks, trusts, etc.) −1,365 −62.05

Non-financial companies listed in LSE and included in FAME 835 37.95

LSE companies with missing or incomplete data −192 −8.73

Non-financial companies with all available data 643 29.23

Companies received a going-concern qualified audit opinion(GCM sample)

−29 −1.32

Companies with clean audit opinion 614 27.91

Non-stressed companies −222 −10.09

Stressed non-financial companies available for matching 392 17.82

ABACUS

292


also included in the FAME database. Following prior research, we exclude 1,365financial and services companies due to their differing financial statement propertiesthat are not well captured in available financial stress models. Then we eliminateanother 192 companies with missing data,

6

and arrive at a sample of 643 publiclylisted non-financial companies with all data available.

We then examine annual report filings and auditor reports for all 643 availablefirms using the Global Access database. We undertook the laborious examinationof the actual audit reports in order to eliminate any database coding errors andto ensure that we were able to identify properly all companies receiving agoing-concern modified opinion from their auditor for the year.

7

This processidentified twenty-nine companies that received a going-concern modified opinionfor the year.

8

These twenty-nine companies comprise our test sample of all thenon-financial listed companies that received a going-concern modified opinion(i.e., the GCM sample) that also had all the control variable information neededfor our models.

9

Prior researchers have argued that proper comparison of any GCM sampleshould be made only to other financially stressed non-going-concern modifiedcompanies (McKeown

et al.

, 1991; Hopwood

et al.

, 1994). Therefore, selection ofour control sample begins with the 614 remaining listed companies that did notreceive a going-concern modified opinion. Following Mutchler

et al.

(1997) andGeiger and Rama (2003), we consider a company as being in financial stress if itmeets one of the following criteria: (a) negative working capital at the end of theyear, or (b) bottom line loss for the year. This screening yielded 392 companiesthat exhibited at least one of the financial stress criteria.

Having identified the financially stressed companies that did not receive agoing-concern modified audit opinion, we then calculated the probability ofbankruptcy for all 392 companies in this sample according to the model used inHopwood

et al.

(1994). Following Geiger and Rama (2003), the non-going-concern modified companies were matched to our test sample companies based

6

Companies were eliminated due to missing or incomplete audit report data, audit and non-audit feedata, missing financial statement data needed to calculate the measure of financial distress, or formissing or incomplete information on any of the control variables discussed in the next section.

7

After checking a sample of the above 643 companies, we found many auditor opinions to beincorrectly classified by FAME. This is similar to Butler

et al.

(2004), who found the Compustatdatabase to contain significant amounts of error with respect to correctly classifying auditoropinions. Accordingly, we used the Global Access database to identify the firm’s correct auditopinion. Global Access acts as a comprehensive repository of published company information andincludes scanned annual reports for companies in numerous countries.

8

Of these twenty-nine companies, twenty-seven received modified unqualified opinions and tworeceived disclaimers of an opinion for going-concern related uncertainties.

9

Arguably, we could have restricted our analysis to only those companies receiving a first-timegoing-concern modified opinion, but the sample size is reduced substantially (by fourteen companies).Accordingly, following Lennox (1999) and Reynolds and Francis (2000) we include a prior-yeargoing-concern modified report indicator variable in our regression model to control for first-timeversus continuing modifications.


293


on probability of bankruptcy, net sales and U.K. SIC code, in the order mentioned.These twenty-nine non-going-concern modified companies comprise our financiallystressed control sample (i.e., the NGCM sample). We perform our analyses onthe twenty-nine GCM firms and the matched twenty-nine NGCM firms. Thisprocedure ensured that similar companies with respect to financial stress, companysize, and type of industry were included in the experimental and control samplesfor analysis.

RESEARCH METHOD

Following prior research (Lennox, 1999; Geiger and Rama, 2003), we employ alogistic regression to examine the association between the magnitude of auditand non-audit fees and the probability that the auditor modifies the report for going-concern uncertainties for our GCM and matched NGCM companies. Accord-ingly, we estimate the coefficients in the following model:

GCM

=

b

0

+ b1* SIZE + b2* PROB + b3* DFT + b4* COSTRED + b5* MITIGATE + b6* REPORTLAG + b7* BIG4 + b8* PRIORGCM + b9* NASFEE + b10* AUDFEE + standard error (1)

The variables are defined as follows:

GCM = coded 1 if audit opinion was modified for going-concern, else 0,SIZE = natural log of total assets (in millions of pounds sterling),PROB = probability of bankruptcy from Hopwood et al.’s (1994) model,DFT = coded 1 if the company was in default, else 0,COSTRED = coded 1 if company entered into significant cost reduction

program, else 0,MITIGATE = coded 1 if company announced sales of significant assets, issued

new debt or new equity during the year, else 0,REPORTLAG = number of days from the end of the year to the audit report date,BIG4 = coded 1 if company was audited by a Big 4 audit firm, else 0,PRIORGCM = coded 1 if company received a going-concern modified report

in the prior year, else 0,NASFEE = natural log of fees paid for non-audit services, andAUDFEE = natural log of fees paid for audit services.

In the model above, NASFEE and AUDFEE are the fee variables of interestand including them both in the logistic regression enables us to simultaneouslyassess H1 and H2 with respect to both types of fees and audit reporting decisions.As in prior research, we use natural log of audit and non-audit fees in ourlogistic regression as our audit fee measures (DeFond et al., 2002; Geiger andRama, 2003).10 Based on prior research, we also include the following types ofcontrol factors in our analysis: (a) financial variables, (b) management plans

10 Using raw NAS and audit fees produces results substantively the same as those presented.

ABACUS

294© 2008 The AuthorsJournal compilation © 2008 Accounting Foundation, The University of Sydney

and mitigating factors, (c) audit reporting lag, (d) audit firm size, and (e) prioryear audit report.

Auditors have financial and other incentives (i.e., to maintain reputation) toretain large clients. Further, large companies may be able to mitigate their financialstress more easily than small companies. In addition, prior research has found anegative relation between company size and the receipt of a going-concern modifiedaudit opinion (Mutchler et al., 1997; Geiger and Raghunandan, 2001; Geiger andRama, 2003). Thus, a negative association between the likelihood of a going-concernmodified audit opinion and company size is expected. Hence, we include a companysize (SIZE) control variable measured using log of total assets (in millions ofpounds sterling).

Prior research suggests that there is a positive association between the likelihoodof a going-concern modified audit opinion and the company being in default ondebt obligations (Hopwood et al., 1994; Chen and Church, 1992; Mutchler et al.,1997). Hence, we include default status (DFT) as a control factor. We classify acompany as in default (DFT) if it is either in payment default or technical defaultof loan covenants, as identified in the company’s annual report.

Prior research also suggests that there is generally a positive associationbetween the likelihood of a going-concern modified audit opinion and financialstress (Mutchler, 1985; Hopwood et al., 1989, 1994; Behn et al., 2001). However,the relationship between level of financial stress and audit reporting for our samplesof highly stressed companies that have already been matched based on levels ofstress is problematic. In fact, using the same research design, Geiger and Rama(2003) find a negative association between levels of financial stress and receivinga going-concern modified report in their examination of reporting on stressedcompanies. However, to maintain consistency with prior research, we includelevel of financial stress (PROB) as a control factor. We measure financial stress(PROB) using the coefficients of the probability of bankruptcy model given inHopwood et al. (1994).11 Based on prior research, we expect also to find a negativerelationship between PROB and going-concern modifications for our alreadystressed companies.

Professional standards require auditors to evaluate client management planswhen there is uncertainty regarding the client’s ability to continue as a goingconcern. Behn et al. (2001) find that the ability of a company to raise capital andborrow funds to finance its operations is negatively associated with the likelihoodof receiving a going-concern opinion. Further, Reynolds and Francis (2000) findthat the combined actions of the company raising equity capital, borrowing funds,or selling significant assets was also negatively associated with the likelihood ofreceiving a going-concern opinion. Findings in Geiger and Rama (2003) furthersuggest that going-concern modified companies are more likely to have enteredinto significant cost reduction efforts. Hence, we take into consideration theimportance of management plans by examining company filings and annualreports in order to include two additional control factors. Following Reynolds and

11 We use the value of the intercept as corrected by Geiger and Raghunandan (2001).


295© 2008 The Authors

Journal compilation © 2008 Accounting Foundation, The University of Sydney

Francis we include an indicator variable for mitigating factors (MITIGATE) tocontrol for whether the company entered into the sale of significant assets, issuednew debt or new equity during the year. Second, following Geiger and Rama,we include an indicator variable for the company entering into a significant costreduction plan (COSTRED) to control for companies that announced a substantialcost reduction plan during the year. These additional control variables are gatheredmanually from the accounts, notes and management’s discussion provided in thecompany’s annual report.

Results from prior research have shown that the time between the company’sfiscal year-end and the audit report date (known as reporting lag) is associatedwith the type of the audit opinion (Francis, 1984; Chen and Church, 1992; Behnet al., 2001). Auditors typically take more time to audit financially stressedcompanies and to issue going-concern modified opinions than to audit financiallyhealthier companies and issue non-modified audit reports. As a consequence, weexamine the date the auditor signed each report for evidence of the time betweenthe end of the company’s fiscal year and the date the auditor issued their finalopinion on the financial statements. We include this time period of audit reportinglag as an additional variable (REPORTLAG) in our analysis to control for thetimeliness of audit opinions on stressed companies.

Auditor size may also be related to the type of audit opinion issued. Altman(1982), Citron and Taffler (1992), Geiger et al. (1998) and Pryor and Terza (2002)have investigated the possibility of the creation of the self-fulfilling prophecy afteran issuance of going-concern modified opinion. That is, the issuance of the financialstress warning in the going-concern modified report itself causes a higher likeli-hood of subsequent company failure. Further, audit firms also run the risk oflosing a client after rendering a going-concern modified opinion, because theaudit client is more likely to change an auditor when they receive a going-concernmodified opinion (Geiger et al., 1998; Carcello and Neal, 2003). This risk may beeven higher for the Big 4 audit firms, because they charge higher prices incomparison with small audit firms, as documented from prior audit pricingresearch (Francis, 1984; Basioudis and Francis, 2007). In addition, Big 4 auditorsmay have more to lose from litigation related to any one particular audit whichmay lead them to be more conservative in their reporting decisions compared tonon-Big 4 audit firms. This would suggest a more cautious and conservativereporting model would be used by the largest accounting firms, which would lead tothe issuance of more modified audit reports (Francis and Krishnan, 1999). Conversely,the Big 4 may derive a higher proportion of their revenues from non-audit servicesthan do the non-Big 4 firms. This would suggest that the largest audit firms mightbe more leery of upsetting and possibly losing an existing client by issuing them amodified opinion (Carcello and Neal, 2003). Hence, based on these prior studieswe have no strong a priori directional expectations. We empirically examine theassociation between audit firm size and audit opinions in our study by includingan audit-firm size indicator dummy variable (BIG4) in our regression.

Finally, an indicator variable for the prior year audit opinion (PRIORGCM) isincluded in our model in order to control for the effect of the report type in the

ABACUS


prior year. Previous research has found that audit firms are reluctant to issue aninitial report modification, as well as remove an existing report modification(Kida, 1980; Krishnan, 1994; Krishnan et al., 1996). Further, Lennox (1999) finds thatprior year going-concern report modifications in the U.K. are significantly positivelyassociated with current year going-concern report modifications. Accordingly, weexpect a similar positive relationship for our samples of financially stressed U.K.firms.

RESULTS

Table 2 provides descriptive data for the two samples of companies: companiesreceiving a going-concern modified opinion from their auditors (our GCM testsample), and the financially stressed companies that did not receive a going-concern modified opinion from their auditors (our NGCM control sample). Thenon-significant differences between the two samples with respect to the net salesand total assets, as well as the non-significant difference in the probability ofbankruptcy measure (at p > .05), are partial evidence of the effectiveness of ourmatching procedure used in the formation of the NGCM control sample.

In addition, the differences in REPORTLAG and PRIORGCM for the twosamples were found to be highly significant (p < .001). Specifically, consistent withprior audit reporting research, the GCM companies were more likely to have alonger reporting lag and more likely to have received a going-concern modifiedopinion in the prior year than the NGCM control sample companies. In fact, wefind that none of the twenty-nine NGCM control firms received a going-concernreport modification in the prior year. However, we also find that the GCMcompanies were no less likely to have been in default on loans or to have raisedfunds by issuing new debt, new equity or from assets sales (MITIGATE) duringthe year compared to the stressed NGCM sample companies. The results furtherindicate that the two groups are not significantly different with respect to enteringinto substantial cost reduction plans (COSTRED). We also find no evidence ofsignificant differences between the two samples for the Big 4 audit firm variable,indicating that the Big 4 firms audited similar proportions of both samples.

Interestingly, the average audit fees and non-audit fees, which are the mainvariables of interest in our study, although not specifically used in our matchingprocedures, were not significantly different between the two samples (p > .05). Yetwhile the average audit and non-audit fee levels across the two groups appearsimilar, it is interesting to note that 27.6 per cent of the GCM sample did not payany non-audit or consultancy fees to their audit firm for the year, yet the rate forthe NGCM companies is only 3.4 per cent (p < .05). These univariate fee resultsare consistent with Geiger and Rama (2003) in their examination of the U.S.market, and indicate that going-concern modified companies were more likely tohave not paid any non-audit fees to their auditor during the year. However, asnoted previously, these univariate fee results are more appropriately assessed in amultivariate setting where additional fee-related control variables are included inthe analysis.




Table 3 presents the correlation coefficients of our model variables. Of thevarious correlations, the highest correlation is between SIZE and AUDFEE at0.782. SIZE is also positively correlated with NASFEE at 0.304, and the two feemeasures are correlated at 0.349. These results are consistent with prior researchfinding fees and company size all positively correlated. As expected, GCM andDFT are both positively correlated with PRIORGCM (at 0.564 and 0.414, respec-tively). This positive relationship is also consistent with prior research. All othercorrelations are below 0.300. Judge et al. (1988) suggest that correlations below

Table 2

DESCRIPTIVE DATA: GOING-CONCERN MODIFIED COMPANIES (GSM) VS NON-GOING-CONCERN MODIFIED COMPANIES (NGSM)

Variable Going-concern modified sample (n = 29)

Non-going-concern modified sample (n = 29)

Mean (st. dev.)

% Range [median]

Mean (st. dev.)

% Range [median]

Mean difference p-value

NET SALES (in millions)

19.5 0.02–166.6 17.6 0–248.5 0.862(36.3) [6.48] (45.9) [6.95]

TOTAL ASSETS (in millions)

18.2 0.04–109.36 38.6 2.03–667 0.386(26.5) [7.89] (122.4) [9.2]

PROB 0.45 0.2–1.00 0.52 0.44–0.95 0.076(0.18) [0.39] (0.12) [0.47]

DFT 10.3 0 0.078

COSTRED 37.9 24.1 0.264

MITIGATE 69.0 62.1 0.588

REPORTLAG (in days)

160 44–342 94.3 27–180 0.000(60.8) [168] (31.2) [91]

BIG4 51.7 51.7 0.797

PRIORGCM 48.3 0.00 0.000

AUDITFEE (£000)

68.4 7.2–228 51.7 9.8–400 0.376(70.5) [39] (71.5) [37]

NONAUDITFEE (£000)

58.1 0–430 47.5 0–300 0.653(98.2) [25] (78.7) [16.8]

COMPANIES W/NO NASFEES

27.6 3.4 0.011

PROB = probability of bankruptcy from Hopwood et al.’s (1994) modelDFT = coded 1 if the company was in default, else 0COSTRED = coded 1 if company entered into significant cost reduction program, else 0MITIGATE = coded 1 if company announced sales of significant assets, issued new debt or new

equity during the year, else 0REPORTLAG = number of days from the end of the year to the audit report dateBIG4 = coded 1 if company was audited by a Big 4 audit firm, else 0PRIORGCM = coded 1 if company received a going-concern report in the prior year, else 0AUDFEE = fees paid for audit servicesNONAUDITFEE = fees paid for non-audit services

AB

AC

US

298©

2008 The A

uthorsJournal com

pilation © 2008 A

ccounting Foundation, T

he University of Sydney

Table 3

CORRELATION MATRIX

GCM SIZE PROB DFT COSTRED MITIGATE PRIORGCM NASFEE AUDFEE

GCM 1.000

SIZE −0.175 1.000

PROB −0.235*** −0.271** 1.000

DFT 0.234*** 0.192 −0.131 1.000

COSTRED 0.149 0.052 0.001 0.180 1.000

MITIGATE 0.073 0.010 0.036 −0.158 0.173 1.000

PRIORGCM 0.564*** −0.210 −0.187 0.414*** 0.231*** 0.070 1.000

NASFEE −0.283*** 0.304*** 0.261*** 0.130 0.139 −0.116 −0.292*** 1.000

AUDFEE 0.098 0.782*** −0.136 0.219*** 0.229*** −0.006 0.022 0.349*** 1.000

*** Correlation significant at the 0.01 level (2-tailed)** Correlation significant at the 0.05 level (2-tailed)

GCM = coded 1 if audit opinion was modified for going-concern, else 0SIZE = natural log of total assets (in millions of pounds sterling)PROB = probability of bankruptcy from Hopwood et al.’s (1994) modelDFT = coded 1 if the company was in default, else 0COSTRED = coded 1 if company entered into significant cost reduction program, else 0MITIGATE = coded 1 if company announced sales of significant assets, issued new debt or new equity during the year, else 0REPORTLAG = number of days from the end of the year to the audit report dateBIG4 = coded 1 if company was audited by a Big 4 audit firm, else 0PRIORGCM = coded 1 if company received a going-concern report in the prior year, else 0NASFEE = natural log of fees paid for non-audit servicesAUDFEE = natural log of fees paid for audit services




0.8 are not likely to substantively increase multicollinearity. Additionally, varianceinflation factors for all models reported in the article do not exceed ten, the typicalthreshold for indicating multicollinearity might be an impediment to modelinterpretation. Based on these analyses, it does not appear that multicollinearityis a substantial deterrent with respect to analysing and interpreting our data andmodel results.

Results from our multivariate logistic regression are presented in Table 4. Theoverall model is significant (χ2 = 64.659, p < .0001) and adequately fits the data(Hosmer-Lemeshow: 3.058, p = .880). From the control variables, and consistentwith prior research, the SIZE variable indicates a continued company size bias infavour of larger companies not receiving a going-concern modified opinion fromtheir auditor (p = .020). Consistent with Geiger and Rama (2003), the coefficient

Table 4

LOGISTIC REGRESSION RESULTS (n = 58)

GCM = b0 + b1* SIZE + b2* PROB + b3* DFT + b4* COSTRED + b5* MITIGATE + b6* REPORTLAG + b7* BIG4 + b8* PRIORGCM + b9* NASFEE + b10* AUDFEE + standard error

Variables Exp. sign Estimate Wald χ2 p-valuea

Constant −166.1411 4.3344 0.037

SIZE − −6.7051 4.2454 0.020

PROB − −20.0856 2.9380 0.043

DFT + 5.0826 0.0000 0.500

COSTRED + −0.8510 0.1256 0.361

MITIGATE − −3.5984 2.3251 0.064

REPORTLAG + 0.2573 4.4404 0.018

BIG4 ? 0.7181 0.0822 0.774

PRIORGCM + 23.5261 0.0000 0.499

NASFEE − −1.0292 4.9579 0.013

AUDFEE + 16.1955 4.3832 0.018

Model χ2: 64.659, p < 0.00001; Nagelkerke R2 = 0.896; Hosmer-Lemeshow: 3.058, p = 0.880a p-values are one-tail, except for the intercept and BIG4 term

GCM = coded 1 if audit opinion was modified for going-concern, else 0SIZE = natural log of total assets (in millions of pounds sterling)PROB = probability of bankruptcy from Hopwood et al.’s (1994) modelDFT = coded 1 if the company was in default, else 0COSTRED = coded 1 if company entered into significant cost reduction program, else 0MITIGATE = coded 1 if company announced sales of significant assets, issued new debt or new

equity during the year, else 0REPORTLAG = number of days from the end of the year to the audit report dateBIG4 = coded 1 if company was audited by a Big 4 audit firm, else 0PRIORGCM = coded 1 if company received a going-concern report in the prior year, else 0NASFEE = natural log of fees paid for non-audit servicesAUDFEE = natural log of fees paid for audit services

ABACUS


for the PROB control factor is negative and significant (p = .043) in our multivariateanalysis, even for these highly stressed companies. The MITIGATE control variableis negative and weakly significant (p = .064), indicating that those companies ableto raise additional capital by executing management’s plans to mitigate theirfinancial stress were marginally less likely to receive a going-concern modifiedaudit opinion. Further, companies that have longer audit report lags (REPORTLAG)were significantly (p = .018) more likely to receive a going-concern modifiedopinion, which is consistent with prior audit reporting literature indicating companiesexhibiting going-concern uncertainties need more time to be audited.

All other control variables are insignificant (at p > .10). Consistent with Geigerand Rama (2003) we also find that the DFT variable is insignificant (p = .500),indicating default status for our samples of financially stressed firms is notsignificantly related to auditor reporting decisions. However, none of our NGCMfirms was in default on their debt, limiting the variability of DFT variable and theability of the variable to discriminate the two sets of firms. The COSTRED variableis not significant (p = .361). The BIG4 variable is positive but not significant (p =.774), indicating a lack of a significant Big 4 audit firm bias when deciding whetheror not to issue a going-concern modified audit opinion to the financially stressedcompanies in our sample. The PRIORGCM indicator variable is also positive butnot significant (p = .499) implying that the prior years audit report did not have asignificant effect on the probability of the stressed company receiving a going-concern modified report in the current year. However, again, none of the NGCMfirms received a going-concern modified report in the prior year, thereby limitingthe variability of PRIORGC variable and the ability of this indicator variable todiscriminate GCM from NGCM firms.

The results for our first variable of interest, NASFEE, indicate that the magnitudeof NAS fees is significantly negatively (p = .013) associated with companies receivinga going-concern modified audit report from their audit firm. Our logistic regressionresults suggest that financially stressed U.K. companies that pay their externalaudit firms higher amounts of NAS fees were significantly less likely to receive agoing-concern modified audit report. Our results are in contrast to the non-significant association of NAS fees and going-concern modified audit opinionsreported by Geiger and Rama (2003) and DeFond et al. (2002) for U.S. firms, andthe non-significant results reported by Lennox (1999) studying U.K. reportingdecisions. However, they are consistent with the more recent U.K. findings ofFirth (2002) in his examination of all types of report qualifications and NAS fees,and with those of Wines (1994), Sharma (2001) and Sharma and Sidhu (2001) forAustralian firms. They provide clear evidence in support of our first hypothesis.The negative association between NAS fees and going-concern modifications suggestthat there may be an adverse effect of the magnitude of NAS fees on auditor’s finalgoing-concern reporting judgments for financially stressed U.K. companies.

The findings regarding our second variable of interest, AUDFEE, indicate thatthe magnitude of the audit fee paid is positively and significantly (p = .018) relatedto receiving a going-concern modified audit opinion. They are consistent with thesecond hypothesis and provide additional support to prior research on audit pricing




that finds a positive association between audit fees and report modifications(Francis and Simon, 1987; Bell et al., 2001). They are also consistent with those ofGeiger and Rama (2003) in their examination of going-concern modified auditreport decisions and levels of audit fees in the U.S.

In sum, higher NAS fees paid to the auditor are found to be associated with alower probability that the company would receive a going-concern modified auditopinion, while higher audit fees are associated with a higher probability that thecompany would receive a going-concern modified audit opinion. Both of our mainresults are consistent with the expectations of the effect of audit and NAS fees onauditor reporting decisions. They also reinforce the necessity to concurrently analyseboth types of fees paid to the audit firm (i.e., audit and non-audit related) whenexamining the association of fees and audit reporting decisions.

Further TestsFirst, because there are a relatively small number of sample companies receivinga going-concern modified report in this study, we wanted to ascertain that thosecompanies are similar with respect to fees and audit opinions compared to thelarger number of U.K. companies examined in Lennox (1999). As noted previously,this earlier large-sample study focused on NAS fees exclusively, as opposed toboth NAS fees and audit fees, and did not include an audit service fee measure inthe model estimations. Therefore, the regression model used in this study wasmodified to be similar to the model used in the Lennox study by deleting ourAUDFEE measure and re-estimating the regression model presented in Table 4.The results of this modified regression are consistent with those reported inLennox. Specifically, when we remove our AUDFEE measure and include onlythe NASFEE in the model, as in Lennox, the coefficient for our NASFEE vari-able becomes insignificant (p > .25) and the control variables remain essentiallyunchanged. The similar results of this modified regression compared to those ofLennox both help confirm the sample representativeness to those of the largersample of U.K. listed firms, and reinforce the necessity to assess both audit feesand NAS fees simultaneously when examining the association of fees and auditorreporting decisions (DeFond et al., 2002; Firth, 2002; Geiger and Rama, 2003).Without concurrently examining both types of fees paid to the auditors, researcherscannot properly identify the effect of different types of fees paid to audit firmsand their impact on difficult audit reporting decisions.

To ascertain whether our results are driven by unusual sample observations, weevaluated our sample for outliers. Tabachnick and Fidell (2001) define outliers asthose with standardized residual values greater than +/− about 3.3. None of thestandardized residuals in our sample exceed these thresholds. However, oneGCM observation had a standardized residual value of −3.15, which we consideredclose enough to the cut-off to warrant further investigation. Eliminating thisobservation together with its NGCM matched firm does not affect the results ofour analyses. Specifically, when we re-estimate the model presented in Table 4, theNASFEE variable remained negative and significant (p < .05) and the AUDFEEvariable remained positive and significant (p < .05) in our re-estimated regression.

ABACUS


Accordingly, we do not find any strong evidence that our results are driven byoutliers or unusually influential observations.

The business press and legislators have also expressed concern that the relativelevel of NAS fees compared to audit fees derived from the same company is apotential contributor to the impairment of audit firm independence. If auditorsreceive proportionately more NAS fees than audit fees, auditors may be moreswayed to adopt the client’s perspective in order to preserve the more lucrativeNAS fees (Levitt, 2000). Accordingly, Firth (1997b) and Geiger and Rama (2003)have examined the ratio of NAS fees to audit fees, while Lennox (1999) andDeFond et al. (2002) have examined the ratio of NAS fees to total fees as a measureof relative NAS fee importance. Replacing the NASFEE and AUDFEE variablesin our logistic regression with either the ratio of NAS to audit fees, or the ratio ofNAS fees to total fees, produces coefficients on the ratio variables that are positivebut not significant (p > .20) in both regressions, while the remaining variablecoefficients are substantively unaltered. These non-significant NAS fee ratioresults are consistent with earlier reporting studies and, most notably, are alsoconsistent with the findings in Lennox, providing further support for the repre-sentativeness of our sample, as well as the need to separate NAS fee and auditservice fee measures for analyses.

We initially follow Reynolds and Francis (2000) in specifying the MITIGATEcontrol variable. However, Geiger and Rama (2003) include separate indicatorvariables to capture these mitigating factors. In order to ascertain whether theaggregate MITIGATE measure is better specified as separate indicator variables,we replace MITIGATE with separate indicator variables for: (a) whether thecompany raised funds by engaging in significant asset sales (ASSETSALE),(b) issuing new equity securities (NEWEQUITY), or (c) issuing new debt secur-ities (NEWDEBT) during the year. Replacing the MITIGATE variable in ourmodel with the separate ASSETSALE, NEWEQUITY and NEWDEBT variablesreveals that all three separate indicator variables are not significant (p > .10), andour variables of interest—NASFEE and AUDFEE—have the same direction asthe main results and remain significant at p = .044 and .058, respectively.

Reynolds and Francis (2000) have argued that the office-level influence ofaudit clients is an important factor in the audit firm’s decision-making process.Accordingly, we add a client influence variable (INFLUENCE) to the regressionmodel in Table 4. This attempts to capture the size of the client relative to the sizeof the auditor’s office. The accounting firm’s lead engagement office is identifiedfrom the office-specific letterhead used for the audit report and supplied by theGlobal Access database.

These data were then hand-collected for all listed companies in order to measureaccurately the client influence for each specific office. Following Reynolds andFrancis, client-specific influence is measured as the ratio of the log of that client’ssales divided by the log of sales for all public companies that that office of theaudit firm renders opinions on (e.g., log company sales / ∑log of sales of all publicclients of the office issuing the audit opinion). The ratio of company sales to allpublic company sales for the office issuing the audit opinion is intended to capture




the relative value or worth of that particular company to the specific office of theaudit firm making the reporting decision. We would expect that clients with largerinfluence (i.e., higher INFLUENCE ratios) would be able to negotiate for a morefavourable audit opinion and would be less likely to receive a going-concernmodified opinion from their auditor. Including the INFLUENCE measure in themodel reveals that its coefficient is not significant (p > .25); however, the NASFEEvariable remains negative and significant (p < .05) and the AUDFEE variableremains positive and significant (p < .05). These results imply that the potentialinfluence of a relatively larger client on a particular audit firm office is notsignificantly related to going-concern reporting decisions. However, as seen inTable 2, the companies in our sample are comparatively small and, hence, theirrelative influence to the specific office of the incumbent auditor making the auditreporting decision may not be as important for the firms in our study.

We also perform additional tests to ensure that the size of the audit firm (Big 4or not) does not significantly interact with NAS fees to exert additional influenceon the audit firm’s reporting decision if a large audit firm receives high NAS fees.When we add a BIG4*NASFEE interaction variable to our model the coefficienton the interaction term is negative but not significant (p > .25) and the NASFEEvariable remains negative and significant (p < .05), the AUDFEE variableremains positive and significant (p < .05), and the other model coefficients remainsubstantively the same as reported in Table 4. If we further expand the model byadding both the INLUENCE variable, discussed in the preceding paragraph, andBIG4*NASFEE interaction variable, both new variable coefficients are found tobe insignificant (p > .25) and similar results to those presented in Table 4 occur.Specifically, the NASFEE variable remains negative and significant (p < .05), andthe AUDFEE variable remains positive and significant (p < .05).

These additional tests provide support that the sample used in the study is rep-resentative of the larger U.K. samples from earlier studies, and that the results arerobust to alternative model and control variable specifications.

DISCUSSION AND CONCLUSION

After the recent, and seemingly wide-spread, high-profile corporate failures, criticsof the accounting profession have again expressed concerns that when auditorsreceive NAS fees from their audit clients audit independence may be at risk andaudit decisions may be adversely affected. In order to test empirically this concernin the U.K., this study examined the association between NAS fees, audit servicefees, and audit reporting decisions for financially stressed companies. We examineaudits and reporting decisions of financially stressed companies because these areexactly the types of engagements in which auditors must exercise a significantamount of independent professional judgment (Kida, 1980; DeFond et al., 2002).

The analysis here of audit opinions on U.K. companies in financial stress providesevidence of a significant negative association between NAS fees and the likelihoodof a company receiving a going-concern modified audit report. Specifically, theevidence suggests that financially stressed companies that paid high NAS fees are

ABACUS


less likely to receive a going-concern modified audit report from their U.K. auditfirm. This contrasts with the evidence from similar studies of NAS fees and auditreporting decisions in the U.S., as well as an earlier finding in Lennox (1999) whoexamined NAS disclosure patterns and U.K. reporting decisions. However, ourresults are consistent with several studies in the Australian market (Wines, 1994;Sharma, 2001; Sharma and Sidhu, 2001) and with the U.K. evidence presented inFirth (2002) regarding NAS fees and all types of qualified audit reportingdecisions. These collective findings are also not inconsistent with those of Basioudisand Francis (2007) who find that audit fee structures in the U.K. are different fromthose found in the U.S. Basioudis and Francis’s (2007) findings lead them to arguethat audit markets are not uniform across even what appear to be fairly homogeneouscountries. Our findings are consistent with theirs and reinforce the need toseparately examine markets and settings of different countries and contest theassumption that findings in one setting would be the same across all, even seeminglysimilar, countries (Simunic and Stein, 1987; Ball et al., 2000; Gul et al., 2002;Basioudis and Francis, 2007).

Although we find the expected negative relationship between level of NAS feesand going-concern modified audit reports, as discussed in Wines (1994) and Firth(2002), we cannot rule out an alternative explanation. While the existence of thisnegative relationship is consistent with impaired auditor independence, it doesnot prove it. Firth argues that on the one hand this NAS fee/reporting relationshipmay suggest that high NAS fees deter auditors from issuing going-concernmodified opinions, yet on the other hand it might well be that the additional servicesperformed by the audit firm in the process of doing the NAS work actually aid theclient and afford them the legitimate ability to avoid the going-concern modification.If the NAS work allows the client to resolve any remaining uncertainties in themind of the audit firm (e.g., resolve substantive questions regarding poor management,resolve questions regarding unreliable information systems, or resolve issues ofpossible fraud on the accounts), or the NAS enable the client to become moreprofitable during the year and potentially in future periods, these circumstances ofenhanced NAS fees would clearly improve the chances of the company receivingan unmodified opinion from the auditor. Finally, although our tests cannotdisentangle the competing reasons for the relationship, the impairment of auditindependence, and as a result the reduced incidence of going-concern modifiedopinions, can also not be ruled out.

Our audit service fee results are consistent with prior research that also finds apositive association between audit service fees and going-concern modifiedopinions. The positive audit service fee and modified opinion relationshipconfirms prior audit fee research that documents a positive association betweenaudit fees, audit effort and the likelihood of receiving a modified audit opinion(Francis, 1984; Francis and Simon, 1987; Beatty, 1993).

Importantly, our aggregate findings of a negative NAS fee association and apositive audit fee association with going-concern modified reporting decisionsreinforce the necessity to examine both types of audit fees simultaneously whenassessing the association of auditor fees and reporting decisions (DeFond et al.,




2002; Firth, 2002; Geiger and Rama, 2003). This is the first study to perform asimultaneous assessment of audit and non-audit fees in the U.K. audit reportingcontext. The inclusion in the model of both fee measures, along with examiningonly stressed companies for which the going-concern modification might be plaus-ible, enables us to examine more appropriately the effect of both fee types andtheir possible effect on audit reporting decisions for going concern.

While the analyses provide fairly strong and consistent evidence regarding theassociation of NAS and audit service fees with audit reporting decisions, ourstudy is subject to limitations. This study has been limited to the final auditreporting decision for financially stressed companies as documented in the annualreports. We have not examined the subsequent consequences of the reportingdecision, namely Type I Errors (a going-concern modified opinion issued for aclient that is subsequently viable) and Type II Errors (a clean opinion issued fora client that subsequently fails). A fruitful area for future research would be toexamine if the two types of audit reporting errors are also associated with auditand non-audit fees in the U.K audit market.

The sampled companies are fairly small and have relatively low levels of NASand audit fees, which could possibly limit the generalizability of our conclusions.However, the sample selection procedures captured all the relevant companieswith respect to the research questions, as we started with the entire population ofnon-financial companies listed on the London Stock Exchange and identified allcompanies receiving a going-concern modified audit report. Restricting ouranalyses to companies with data available on the FAME database, however,would also tend to restrict the sample to the larger companies on the exchange.The resultant sample, however, is probably more reflective of the size bias ingoing-concern reporting, in that prior research has consistently found that smallercompanies are more likely to get a going-concern modification than are largercompanies (Kida, 1980; Hopwood et al., 1994; Mutchler et al., 1997) than in anysample selection bias. We have also performed several supplemental analyses thatindicate our sample results are similar to earlier studies using larger samples, orincluding larger sized companies. Additionally, the overall effects found in thesample companies having relatively low levels of audit and NAS fees might likelybe stronger if larger firms with higher fees were included for analysis. Accordingly,we believe that the results provide sufficient evidence that they are reflective ofconditions found in the general population of U.K. companies and that thepossible firm size limitation is minimal.

The findings also raise additional issues surrounding the construction of, andrelationship between, NAS and audit service fees. Following Basioudis and Francis(2007), further examination of how audit and NAS fees are constructed in practice(as well as disclosed) in the U.K. may shed additional insight into the fee relation-ships noted here (Houghton and Jubb, 1999; Abdel-Khalik, 1990). Along with theNAS and audit fee determination, an additional avenue for future research is thepossible association of auditor decision-making and the separate components ofNAS fees. For companies reporting on fiscal years ending after 1 October 2005,the new Companies Regulations (2005) requires U.K. companies to disclose NAS

ABACUS


fees paid to their audit firms broken down into more descriptive categories (e.g.,tax compliance, internal audit, litigation services, actuarial services, remunerationservices, etc.).12 Examination of these components of NAS fees and their possibleassociation with auditor decision-making would provide additional evidence onthe effect of more specific non-audit service fees on auditor decisions in the U.K.,including going-concern reporting decisions.

Finally, assessing the impact of NAS fees (and different types of NAS fees) onother auditor judgments (e.g., client acceptance and retention decisions, pre-planningmateriality decisions, engagement risk assessments, internal control evaluations,the amount and types of substantive testing) would extend our knowledge regardingthe possible impairment of auditor independence in the face of NAS fees in theU.K. and elsewhere.

references

Abbott, L. J., S. Parker, G. Peters and K. Raghunandan, ‘An Empirical Investigation of Audit Fees,Non-Audit Fees and Audit Committees’, Contemporary Accounting Research, Summer 2003.

Abdel-Khalik, A. R., ‘The Jointness of Audit Fees and Demand for MAS: A Self-Selection Analysis’,Contemporary Accounting Research, Spring 1990.

Altman, E. I., ‘Accounting Implications of Failure Prediction Models’, Journal of Accounting,Auditing and Finance, Fall 1982.

American Institute of Certified Public Accountants, Commission on Auditors’ Responsibilities [TheCohen Report], Report, Conclusions and Recommendations, AICPA, 1978.

Ball, R., S. P. Kothari and A. Robin, ‘The Effect of International Institutional Factors on Propertiesof Accounting Earnings’, Journal of Accounting and Economics, February 2000.

Barkess, L., and R. Simnett, ‘The Provision of Other Services by Auditors: Independence and PricingIssues’, Accounting and Business Research, Spring 1994.

Basioudis, I. G., and J. R. Francis, ‘Big 4 Audit Fee Premiums for National and Office Level IndustryLeadership in the United Kingdom’, Auditing: A Journal of Practice and Theory, November2007.

Beatty, R. P., ‘The Economic Determinants of Auditor Compensation in the Initial Public OfferingsMarket’, Journal of Accounting Research, Autumn 1993.

Behn, B. K., S. E. Kaplan and K. R. Krumwiede, ‘Further Evidence on the Auditor’s Going-ConcernReport: The Influence of Management Plans’, Auditing: A Journal of Practice and Theory,March 2001.

Bell, T. B., W. R. Landsman and D. A. Shackelford, ‘Auditors’ Perceived Business Risk and AuditFees: Analysis and Evidence’, Journal of Accounting Research, June 2001.

Brinn, T., M. J. Peel and R. Roberts, ‘Audit Fee Determinants of Independent and SubsidiaryUnquoted Companies in the U.K.—an Exploratory Study’, British Accounting Review, Vol. 26,No. 2, 1994.

Buijink, W. F., S. Maijoor, R. Meuwissen and A. Van Witteloostuijn, The Role, Position and Liabilityof the Statutory Auditor Within the European Union, Maastricht Accounting and AuditingResearch Center, 1996.

Butler, M., A. J. Leone and M. Willenborg, ‘An Empirical Analysis of Auditor Reporting and itsAssociation With Abnormal Accruals’, Journal of Accounting and Economics, June 2004.

12 Of our fifty-eight sample companies from 2003, five (three GCM and two NGCM) made voluntarydisclosures of the types of NAS fees paid to their audit firms prior to the new reporting regulationstaking effect.




Carcello, J. V., ‘Discussion of: Audit Research After Sarbanes-Oxley’, Auditing: A Journal of Practiceand Theory, Supplement, 2005.

Carcello, J. V., and T. L. Neal, ‘Audit Committee Characteristics and Auditor Dismissals Following“New” Going-Concern Reports’, The Accounting Review, January 2003.

Carey, P. J., M. A. Geiger and B. T. O’Connell, ‘Costs Associated With Going-Concern-ModifiedAudit Opinions: An Analysis of the Australian Audit Market’, Abacus, March 2008.

Chan, P., M. Ezzamel and D. Gwilliam, ‘Determinants of Audit Fees for Quoted U.K. Companies’,Journal of Business, Finance and Accounting, November 1993.

Chen, K. C. W., and B. K. Church, ‘Default on Debt Obligations and the Issuance of Going-ConcernOpinions’, Auditing: A Journal of Practice and Theory, Fall 1992.

Chen, K. Y., R. J. Elder and J. Liu, ‘Auditor Independence, Audit Quality and Auditor–Client Nego-tiation Outcomes: Some Evidence from Taiwan’, Journal of Contemporary Accounting andEconomics, December 2005.

Citron, D. B., and R. J. Taffler, ‘The Audit Report Under Going Concern Uncertainties: An EmpiricalAnalysis’, Accounting and Business Research, Autumn 1992.

Coffee, J. C., ‘Guarding the Gatekeepers’, New York Times, 13 May 2002.

Companies Act, The Companies Act 1989 (Disclosure of Remuneration for Non-Audit Work)Regulations 1991, Statutory Instrument 1991/2128, 1989, U.K.

Companies Regulations, ‘The Companies (Disclosure of Auditor Remuneration) Regulations 2005’,Statutory Instrument 2005/2417, 2005, U.K.

Craswell, A. T., ‘Does the Provision of Non-Audit Services Impair Auditor Independence?’,International Journal of Auditing, Vol. 3, No. 1, 1999.

Davis, L. R., N. Ricchiute and G. Trompeter, ‘Audit Effort, Audit Fees, and the Provision of Non-Audit Services to Audit Clients’, Accounting Review, January 1993.

DeAngelo, L., ‘Auditor Independence, “Low Balling”, and Disclosure Regulation’, Journal ofAccounting and Economics, August 1981.

DeFond, M. L., and J. R. Francis, ‘Audit Research After Sarbanes-Oxley’, Auditing: A Journal ofPractice and Theory, Supplement, 2005.

DeFond, M. L., K. Raghunandan and K. R. Subramanyam, ‘Do Nonaudit Service Fees Impair AuditorIndependence? Evidence From Going-Concern Audit Opinions’, Journal of AccountingResearch, September 2002.

Department of Trade and Industry, Final Report of the Coordinating Group on Audit and AccountingIssues to the Secretary of State and Industry and the Chancellor of the Exchequer, January 2003.

Dietz, D., ‘“Auditors Are Timid” They Failed to Warn in Most Big Firm Bankruptcies Since 1996’,Pittsburgh Post-Gazette, 26 April 2002.

Ezzamel, M., D. R. Gwilliam and K. M. Holland, ‘Some Empirical Evidence From Publicly QuotedU.K. Companies on the Relationship Between the Pricing of Audit and Non-Audit Services’,Accounting and Business Research, Winter 1996.

——, ‘The Relationship Between Categories of Non-Audit Services and Audit Fees: Evidence FromU.K. Companies’, International Journal of Auditing, March 1999.

Federation of European Accountants, The Role, Position and Liability of the Statutory Auditor Withinthe European Union, Federation des Experts Comptables Européens, 1996.

Firth, M., ‘The Provision of Non-Audit Services and the Pricing of Audit Fees’, Journal of Business,Finance and Accounting, April 1997a.

——, ‘The Provision of Non-Audit Services by Accounting Firms to Their Clients’, ContemporaryAccounting Research, Summer 1997b.

——, ‘Auditor-Provided Consultancy Services and Their Associations With Audit Fees and AuditOpinions’, Journal of Business, Finance and Accounting, June/July 2002.

Francis, J., ‘The Effect of Audit Firm Size on Audit Prices: A Study of the Australian Market’, Journalof Accounting and Economics, August 1984.

Francis, J., and J. Krishnan, ‘Accounting Accruals and Auditor Reporting Conservatism’, ContemporaryAccounting Research, Spring 1999.

ABACUS


Francis, J., and D. Simon, ‘A Test of Audit Pricing in the Small-Client Segment of the U.S. AuditMarket’, Accounting Review, January 1987.

Geiger, M. A., and K. Raghunandan, ‘Bankruptcies, Audit Reports, and the Reform Act’, Auditing:A Journal of Practice and Theory, March 2001.

Geiger, M. A., K. Raghunandan, and D. V. Rama, ‘Costs Associated With Going-Concern ModifiedAudit Opinions: An Analysis of Auditor Changes, Subsequent Opinions, and Client Failures’,Advances in Accounting, Vol. 16, 1998.

Geiger, M. A., and D. V. Rama, ‘Audit Fees, Non-Audit Fees, and Auditor Reporting on StressedCompanies’, Auditing: A Journal of Practice and Theory, September 2003.

Gul, F., S. Lyn and A. Seetharaman, ‘Litigation Risk and Audit Fees: Evidence From United KingdomFirms Cross-Listed on United States Markets’, Journal of Accounting and Economics, February2002.

Hopwood, W., J. McKeown and J. Mutchler, ‘A Test of the Incremental Explanatory Power ofOpinions Qualified for Consistency and Uncertainty’, The Accounting Review, January1989.

——, ‘A Reexamination of Auditor Versus Model Accuracy Within the Context of the Going-Concern Opinion Decision’, Contemporary Accounting Research, Spring 1994.

Houghton, K. A., and C. A. Jubb, ‘The Cost of Audit Qualifications: The Role of Non-Audit Services’,Journal of International Accounting, Auditing and Taxation, Vol. 8, No. 2, 1999.

Independence Standards Board, Staff Report, A Conceptual Framework for Auditor Independence,ISB 2001.

Judge, G. G., R. C. Hill, W. E. Griffiths, H. Lutkepohl and T. C. Lee, Introduction to the Theory andPractice of Econometrics, Wiley, 1988.

Khurana, I. K., and K. K. Raman, ‘Litigation Risk and the Financial Reporting Credibility of Big 4Versus non-Big 4 Audits: Evidence From Anglo-American Countries’, The Accounting Review,April 2004.

Kida, T., ‘An Investigation Into Auditors’ Continuity and Related Qualification Judgments’, Journalof Accounting Research, Autumn 1980.

Krishnan, J., ‘Auditor Switching and Reporting Conservatism’, The Accounting Review, January 1994.

Krishnan, J., J. Krishnan and R. G. Stevens, ‘The Simultaneous Relation Between Auditor Switchingand Audit Opinion: An Empirical Analysis’, Accounting and Business Research, Vol. 26, No. 3,1996.

Lennox, C. S., ‘Non-Audit Fees, Disclosure and Audit Quality’, European Accounting Review, July 1999.

Levitt, A., The Accountant’s Critical Eye, Remarks at the 24th Annual National Conference on Cur-rent SEC Developments, AICPA, December 1996.

——, The Numbers Game, Speech given at New York University Center for Law and Business, 28September 1998.

——, Renewing the Covenant With Investors, Remarks before the New York University Center forLaw and Business, May 2000.

Mautz, R. K., and H. A. Sharaf, The Philosophy of Auditing, American Accounting Association, 1961.

McKeown, J. C., J. F. Mutchler and W. Hopwood, ‘Towards an Explanation of Auditor Failure toModify the Audit Opinion of Bankrupt Companies’, Auditing: A Journal of Practice and The-ory, Supplement, 1991.

Mutchler, J. F., ‘A Multivariate Analysis of the Auditor’s Going-Concern Opinion Decision’, Journalof Accounting Research, Autumn 1985.

——, ‘Empirical Evidence Regarding the Auditors’ Going-Concern Opinion Decision’, Auditing: AJournal of Practice and Theory, Fall 1986.

Mutchler, J. F., W. Hopwood and J. C. McKeown, ‘The Influence of Contrary Information andMitigating Factors on Audit Opinion Decisions on Bankrupt Companies’, Journal of Account-ing Research, Autumn 1997.

Pryor, C., and J. V. Terza, ‘Are Going-Concern Audit Opinions a Self-Fulfilling Prophecy?’, Advancesin Quantitative Analysis of Finance and Accounting, Vol. 10, 2002.




Reynolds, J. K., and J. Francis, ‘Does Size Matter? The Influence of Large Clients on Office-LevelAuditor Reporting Decisions’, Journal of Accounting and Economics, December 2000.

Securities and Exchange Commission (SEC), Proposed Rule: Revision of the Commission’s AuditorIndependence Requirements, Release Nos 33-7870; 34-42994, 2000a.

——, Revision of the Commission’s Auditor Independence Requirements, Release Nos 33-7919; 34-43602, 2000b.

Sharma, D. S., ‘The Association Between Non-Audit Services and the Propensity of Going ConcernQualifications: Implications for Audit Independence’, Asia-Pacific Journal of Accounting andEconomics, December 2001.

Sharma, D. S., and J. Sidhu, ‘Professionalism vs Commercialism: The Association Between Non-AuditServices (NAS) and Audit Independence’, Journal of Business, Finance and Accounting, June/July 2001.

Simnett, R., and K. Trotman, ‘Auditor Versus Model: Information Choice and Information Process-ing’, The Accounting Review, July 1989.

Simunic, D., ‘The Pricing of Audit Services: Theory and Evidence’, Journal of Accounting Research,Spring 1980.

——, ‘Auditing, Consulting, and Auditor Independence’, Journal of Accounting Research, Fall 1984.

Simunic, D., and M. T. Stein, Product Differentiation in Auditing: Auditor Choice in the Market forUnseasoned New Issues, Research Monograph No. 13, The Canadian Certified GeneralAccountants’ Research Foundation, 1987.

——, ‘The Impact of Litigation Risk on Audit Pricing: A Review of the Economics and the Evidence’,Auditing: A Journal of Practice and Theory, Supplement, 1996.

Tabachnick, B. G., and L. S. Fidell, Using Multivariate Statistics, 4th ed., HarperCollins, 2001.

Watts, R. L., and J. L. Zimmerman, Positive Accounting Theory, Prentice-Hall, 1986.

Weil, J., ‘Going Concerns: Did Accountants Fail to Flag Problems at Dot-Com Casualties?’, WallStreet Journal, 9 February 2001.

Wines, G., ‘Auditor Independence, Audit Qualifications and the Provision of Non-Audit Services:A Note’, Accounting and Finance, May 1994.

Documents

Audit Fees Non Audit Fees and Auditor