Information content of Going Concern Audit Reports and the provision of Non-audit services in Credit Decisions:

A Lending Knowledge Structures Approach*

Andrés Guiral**

University of Alcalá (Spain) and Yonsei University (South Korea)

Office phone: +34 91 885-4299 Mobile: +34 629 23 58 19

Fax: +34 91 885-4294 e-mail: andres.guiral@uah.es

Emiliano Ruiz

University of Cádiz, Spain

e-mail: emiliano.ruiz@uca.es

and

Hiu Lam Choy

Drexel University, US

e-mail: hc349@drexel.edu

* We are also grateful to previous comments and suggestions of Michael Bamber, Ken Trotman, Chris Agoglia, Doocheol Moon, Manuel Illueca and Michael Paz on a previous version of this manuscript. The authors are grateful to participants at the 33rd Annual Congress of the European Accounting Association, Istanbul, 2010. We gratefully acknowledge funding received from the Centro Internacional de Formación Financiera (CIFF), Foundation, the Spanish National R+D+I Plan through the research projects SEJ2007-62215/ECON, SEJ2004-00791ECON, SEJ 2006-14021 and the postdoctoral/Fulbright grant EX2004-0294. ** Corresponding author

1

Information content of Going Concern Audit Reports and the provision of Non-audit services in Credit Decisions:

A Lending Knowledge Structures Approach

Abstract:

This study examines whether auditor economic dependence, as proxied by the provision of non-audit services (NAS), affects the audit quality as interpreted by loan officers. We conduct an experiment where eighty experienced loan officers from one of the largest European commercial banks review a loan request under two lending scenarios: 1) a potential borrower receiving an unqualified but modified going concern opinion, and 2) one receiving a qualified going concern opinion. Economic dependence in terms of the provision of NAS is manipulated (low vs. high economic dependence) in both lending scenarios. Using the loan knowledge structures framework, we find that that the provision of NAS only has a negative impact on loan officers’ assessment of the borrower’s character in the unqualified but modified going concern report scenario. We also find that the auditor’s economic dependence has a more significant impact on the male loan officers’ evaluation of the borrower’s character. Keywords Audit report information content; Auditor independence; Non-audit fees; Going concern opinions; Lending knowledge structures; Gender differences.

JEL Classification

M42; G30; G21

2

1 Introduction

The major task of a loan officer is to evaluate a potential borrower’s application,

assess its credit risk and then make a loan recommendation. One significant input to the loan

officers’ decision is the borrower’s financial statements, which allow them to evaluate its

current cash flows, predict its future cash flows, and assess its ability to serve the debt when

due. Also, financial statements are a critical source of information on the assets that can be

pledged as collateral. Because financial statements play such a critical role in the loan

evaluation process, the credibility of financial statements is of major concern to the loan

officers. To this end, the loan officers rely on the audit reports. The audit reports inform the

loan officers on whether the statements are prepared according to GAAP, whether there is any

uncertainty regarding the firm’s survival in the following year, whether there is any constraint

on the audit performed, etc. If the auditor indicates there is a departure from GAAP in its

report, the loan officers will discount the credibility of the financial statements. If the auditor

issues an unqualified report, it suggests that, based on the results of all the audit steps

performed, the financial statements fairly present the client’s financial condition.

The degree of reliance a loan officer placed on an audit report depends on the officer’s

perception of the auditor’s ability to detect material misstatements or omissions in the

financial statements and his/her incentive to report those errors when found. Previous

experimental studies investigate whether lenders are less likely to grant loans to a company

with a qualified audit opinion than to a company with an unqualified audit report. The results

are mixed. While several authors believe that the audit report has no effect on financial

statement users because it solely confirms the users’ expectations of the audit opinion (e.g.,

Libby 1979b; LaSalle and Anandarajan 1997; Elias and Johnston, 2001; Bessell et al 2003),

other authors consider that the audit report provides valuable information because it makes

information contained in financial statements more salient to users (e.g., Firth 1979, 1980;

3

Gul 1987; Miller et al. 1993; Bamber and Stratton 1997). Further, more recent research

suggests that the audit report is only informative to loan officers when it suggests that the

potential borrower firm’s financial condition is not as robust as projected by the financial

statements. Otherwise, the audit report seems to lack information content (Guiral-Contreras et

al. 2007).

In this paper, we examine the issue from a different perspective. We integrate the

research on loan officers’ perception of auditor independence with research on the

information content of audit report. Using this approach and the loan knowledge structures

framework (Guiral 2012; Beaulieu 1994;; Kelly 1985; McDonald and McKinley 1981) as our

basis of analysis, we hope to shed new light on how and under what circumstances the audit

report provides useful information for lending decisions. Specifically, we conduct an

experiment to explore whether the relative importance loan officers assign to the going

concern opinions varies with their perception of auditor independence in terms of the

provision of non-audit services (NAS). In addition to influencing auditor’s independence in

appearance, the provision of NAS can also affect the loan officers’ perception of the

borrower’s character.

An auditor’s independence is viewed as one of the critical determinants of the value of

audit service. Auditing Standard (AU) Section 150 requires that auditor has to maintain “an

independence in mental attitude”. An auditor’s independence in fact enhances the quality of

his/her clients’ financial statements (Church and Zhang 2011). However, because an

auditor’s independence in fact cannot be observed, an auditor’s independence in appearance

becomes critical for the public’s evaluation of the audit quality and hence the perceived value

of the audit service. The provision of NAS is perceived as potentially compromising an

auditor’s independence. Schneider et al. (2006) suggest that the provision of NAS negatively

4

affects the public’s perception of auditor independence.1 With a series of accounting scandals,

the public’s confidence in the audit profession was shattered. The Congress passed the

Sarbanes-Oxley Act in year 2002. One of the major objectives of the Sarbanes-Oxley Act is to

improve the audit profession’s independence.2 The Congress hoped that with the restriction

on NAS provision by auditors, it can improve the perceived independence of auditors and

hence audit quality.

According to the loan knowledge structures framework, we expect that the qualified

audit report signals not only the financial condition of the borrower (i.e., the capacity-capital

knowledge) but also potential problems with the credibility of the financial statements and

hence the borrower’s integrity and honesty (i.e., character knowledge).3 For the unqualified

but modified audit report, while it also informs the loan officers about the financial condition

of the borrower, it does not provide any hint on the borrower’s character. Information on the

auditor’s provision of NAS can fill in this gap and help the loan officers in evaluating the

borrower’s integrity. Thus, this information helps loan officers in updating their knowledge

about the borrower’s character.

In order to investigate whether the auditor’s independence has any impact on the

information content provided by the audit report, we create two lending scenarios: one where

the borrower receives an unqualified but modified going concern opinion and the other one 1 In fact, the independence of Arthur Andresen was in the core of the debate on Enron’s failure and eventually led to the collapse of Andersen. The media and the public criticized that Andersen derived a significant portion of its revenues from providing NAS to Enron and in order to sustain this stream of income from NAS, Andersen compromised its independence and audit quality. 2 Section 201 of the Act prohibits an auditor from providing eight types of services to its clients: (i) bookkeeping, (ii) financial information systems design and implementation, (iii) appraisals or valuation, (iv) actuarial, (v) internal audit outsourcing, (vi) management and human resources, (vii) broker/dealer and investment banking, and (viii) legal or expert services unrelated to audit services (Section 201 identifies these services as outside the scope of practice of auditors). For non-audit services other than those listed above (e.g. tax services (TAX)), an approval by the audit committee is required. 3 The qualified going-concern audit report provides information on both the capacity-capital and character knowledge structures of the borrower in the following way: the going-concern portion of the report informs the loan officers about the potential difficulty the borrower may have in serving the debt when it is due. The qualified portion of the report informs the officers about the deviation of the financial statements from GAAP and signals about the characteristics of the management of the borrower.

5

where the borrower receives a qualified going concern opinion. We then manipulate the

auditor’s independence in appearance in both scenarios by introducing whether the auditor

also provides NAS to the borrower. This manipulation allows us to investigate whether the

provision of NAS undermines the credibility of the audit report and hence that of the financial

statements in each lending scenario.

Results suggest that the information content of going concern audit report as

interpreted by the loan officers is affected by the provision of NAS. When the auditor issues

an unqualified but modified going concern opinion, the provision of NAS has a significantly

negative effect on the loan officers’ assessment of the credit risk of the borrower and hence

the final lending recommendation. On the other hand, if the auditor issues a qualified going

concern opinion, signaling the auditor’s concern on both the borrower’s ability to continue

operation in the following year and the departure of the financial statements from GAAP, then

the provision of NAS has no significant effect on the loan officers’ assessment.

We interpret these results as suggesting that the qualified audit report has already led

the loan officer to revise downward his/her evaluation of the character knowledge structure of

the borrower, in addition to the capacity and capital knowledge structures; hence, the

provision of NAS does not provide any incremental information on the character of the

borrower. Furthermore, the issuance of a qualified opinion signals that the auditor does not

compromise his/her independence in performing the audit task. On the other hand, upon

receiving the unqualified audit report, the loan officers only update their beliefs regarding the

borrower’s capacity and capital knowledge structure to serve the debt. The unqualified report

does not signal any negative aspect regarding the character of the borrower. However, with

the additional information that the borrower also purchases its NAS from the auditor, the loan

officers are concerned about the independence of the auditor and the audit quality. They

discount the credibility of the audit report and that of the financial statements. Even though

6

the borrower could have purchased the NAS from the auditor purely out of cost and/or

efficiency consideration, the loan officers can view the purchase as a way the borrower tries

to “induce” the auditor to provide favorable opinion on its financial statements. As such, the

loan officers likely revise downward their interpretation regarding the borrower’s integrity

and honesty (i.e., the character knowledge structure). That is, while the information on the

provision of NAS by the auditor is considered redundant in the case of qualified going

concern opinion, it provides incremental information in the case of unqualified but modified

opinion. Consequently, we observe that the impact of the provision of NAS on loan officers’

decisions varies with the types of audit opinions the borrower receives.

Our additional analyses also show that the gender of loan officers affects their lending

recommendations. Female loan officers tend to be more conservative and are less likely to

make a favorable loan recommendation than male loan officers. In the unqualified but

modified audit opinion scenario, gender of loan officers also moderates the effect of auditor’s

economic dependence on the lending decision. Auditor’s economic independence has a more

positive impact on the lending recommendation on male loan officers than on female ones.

That is, male loan officers tend to interpret the provision of NAS to be more informative

about the character of the borrower than female loan officers.

Our paper contributes to the auditing literature by taking a new approach to investigate

the information content of the audit report and the effect of NAS on perceived auditor

independence (Glezen and Miller 1985; Frankel et al. 2002; Ashbaugh et al. 2003). We

observe that the information content of audit report and the effect of NAS on loan decisions

are interdependent. In our setting, we observe that loan officers view the provision of NAS as

a supplemental source of information when the audit report does not provide any information

on the character of the borrower. Hence, its effect on lending recommendation is limited to a

situation where the primary source of information—the audit report—does not provide the

7

information necessary for the loan officers’ assessment. Future studies on the impact of the

provision of NAS on perceived auditor independence should consider how its impact may

vary with the research settings (in our case, unqualified versus qualified audit opinions) and

partition the analyses accordingly.

Also, our results suggest that the provision of NAS by a firm’s auditor not only

impairs auditor’s independence in appearance but also has a negative impact on the client’s

perceived honesty and integrity. Thus, client firms should account for this opportunity cost

when considering whether to purchase the NAS from their auditors.

The remainder of the paper is organized as follows. Section 2 develops our

hypotheses. Section 3 describes the research method. Section 4 provides the results of our

analyses. Section 5 concludes the paper.

2 Hypotheses development

2.1 Loan knowledge structures of lending decisions The loan application evaluation and lending decision can be characterized as a

sequential process as depicted in Figure 1 (Wright and Davidson 2000; Guiral-Contreras et al.

2007). After receiving a loan application, the loan officers evaluate both the financial and

non-financial information in order to form their credit risk judgments and lending decisions

regarding the potential borrower (Guiral 2012). According to Beaulieu (1994, 1996, 2001)

and Wright and Davidson (2000), loan officers are generally trained to search for and

organize information, both financial and non-financial, using a framework known as the “five

C’s of credit”—character, capacity, capital, conditions, and collateral (Beaulieu 1994; Kelly

1985; McDonald and McKinley 1981). This framework provides loan officers with certain

standards to follow and ensures that the basic information and loan granting criteria are

followed in the loan application processing. As capacity and capital assessments are all based

8

on the borrower’s financial statements, they can be grouped together as accounting

information (Beaulieu 1996), denoted as the capacity-capital knowledge structure in this

paper. Thus, the five C’s of credit can be consolidated into four loan knowledge structures as

depicted in Figure 1 (Beaulieu 1996). Information collected by loan officers is analyzed with

respect to each of these four dimensions. Each dimension of these knowledge structures is

tied to each other and has influence on one another, as denoted by the two-way arrows in

Figure 1. Together they form a cohesive knowledge structures framework on which the loan

officers based their decisions.

(Insert Figure 1 about here)

Character is the most complex lending knowledge structure among the five C’s of

credit. It appears at the center of the loan structures because it affects the presentation of and

responses to all other information. That is, it provides the loan officers with a “summary

measure” of management’s determination to repay debt (i.e., credibility judgments).

Assessing the quality of a borrower’s character includes evaluating her integrity, stability, and

honesty; all of which can affect the borrower’s intention to repay the debt.4

Credible financial information can be important to lenders as a means of evaluating

the company’s ability to meet their debt commitments (Senkow et al. 2001). That is, financial

information helps in evaluating the other four C’s of the borrower. Loan officers then

combine their judgment regarding the borrower’s character with those on the other four C’s of

credit in order to form credit judgments and make their lending decisions. Capacity and

Capital together form the knowledge structure dealing primarily with the financial

performance of the borrower. Both of which can be judged primarily using information

provided in the financial statements (Beaulieu 1996). Capacity is defined as the

4 For example, the fact that a borrower did not keep a promise to provide certain financial information would be classified as a character fact and interpreted as a negative signal (Beaulieu, 1994).

9

management’s ability to operate a business capable of repaying debt in the past, at present,

and in the future. It is assessed as the probability of a successful repayment of loan mainly

through the analysis of financial statements. Capital, on the other hand, refers to the funds

available to operate a business. Financial statements (balance sheet, income statements,

statements of cash flows, and notes to financial statements) are also the primary source of

information about capital. For instance, profitability analysis, in terms of ROA and ROE, is a

component of capital analysis.

The remaining two lending knowledge structures are conditions and collateral.

Conditions refer to the prevailing economic conditions, such as the key risks facing the

borrower (e.g., recession, risks of the industry, business cycle) and how well those risks can

be mitigated. Finally, collateral is defined as assets available to use as an alternative source of

repayment, such an explicit pledge is required when loan officers detect weaknesses in the

borrower’s other C’s (Beaulieu 1996).5 Once loan officers assess and integrate financial and

non-financial information using these lending knowledge structures, they form a perception of

the borrower’s level of credit risk. This perceived level of risk forms the basis for the loan

approval decision. Using this lending knowledge structures framework, we discuss the

potential impact of the auditors’ going concern reports, their provision of NAS, and the

interaction between the two on loan officers’ credit judgments and lending decisions in the

next subsections.

(Insert Table 1 about here)

5 See Guiral (2012) as a recent example of how loan officers’ knowledge structures are affected in terms of the borrower’s innovation intensity (i.e., capacity–capital and conditions) and its corporate social performance (i.e., character). Holder-Webb and Sharma (2010) is another behavioral study that may illustrate the loan knowledge structures. These authors manipulated Governance strength (i.e., character) and financial performance (i.e., capacity and capital).

10

2.2 Going concern reports

The role played by the audit report in the financial statement users’ decision-making

process has become the subject of interest for both regulators and academics. Since the main

objective of an audit report is to assist financial statement users in assessing the credibility of

financial statements, it helps statement users to make more informative decisions. While a

clean audit report signals that the statements show a true and fair view of the financial

position of the company, a qualified audit report conveys the auditor’s doubt regarding the

credibility of the financial statements (Firth 1979, 1980).6 This information is particularly

helpful in the case of a going-concern opinion, where the auditor sends a message to the

financial statement users expressing its doubts about the ability of the client firm to continue

in existence in the next year (Asare 1990).

Several behavioral studies have also explored the effect of going concern audit reports

on loan officers’ decision process (Firth 1979, 1980; Bamber and Stratton 1997; for a review

also see Asare 1990 and Strawser 1991). The findings of these previous experimental studies

are not conclusive (Bessell et al. 2003). Several papers find that qualified audit reports do not

provide any specific information useful for financial statement users beyond that contained in

the financial statements themselves (Libby 1979b; Houghton 1983; Abdel-Khalik et al. 1986;

LaSalle and Anandarajan 1997; Elias and Johnston 2001; Lin et al. 2003; and Bessell et al.

2003). It has been argued that the impact of a qualified audit report on financial statement

users’ decision-making is solely based on the events contained in the audit report rather than

6 Recent research of Gray et al. (2011) shows a general consensus among financial statement users that a going-concern opinion provides important information for their decisions. Indeed, Gray et al. (2011) indicate that users assume that if the auditor’s report does not include a going-concern comment, this implies that the auditor has performed a going-concern analysis and concluded that going-concern was not an issue. Asare and Wright (2012) conclude that lenders’ assessments on the message conveyed by the going concern audit reports are not significantly different from the assessments of auditors. This suggests that an unqualified audit report provides loan officers with a high level of confidence in the going-concern status of a company.

11

on the audit report itself.7 Conversely, other research papers have provided empirical support

for the hypothesis that the audit report can act as an independent second opinion that makes

the information contained in financial statements more salient to users (Gul 1987, 1991). Firth

(1979, 1980), Gul (1987), Miller et al. (1993), Bamber and Stratton (1997), and Guiral-

Contreras et al. (2007) all find evidence supporting the hypothesis that loan officers perceive

companies receiving qualified audit reports as being more risky.8

While the aforementioned studies have made significant contribution to the audit

literature, the mixed results suggest that the role played by the audit report in credit decisions

deserves further investigation.9 Following Bamber and Stratton (1997), we assume that a

going concern opinion should be informative to loan officers to the extent that such an

opinion provides additional information about the borrower’s future cash flows or if it directs

7 A summary of the findings of previous studies that do not find audit reports provide any incremental information is as follows. Libby (1979b) find no evidence that qualified audit reports (general uncertainty) affect lending decisions of U.S. loan officers. Specifically, he finds that a qualified audit report did not accentuate the message if disclosures have already been made in the notes of the financial statements. In a similar study, Abdel-Khalik et al. (1986) conclude that uncertainty qualifications do not affect Canadian lenders’ decisions. Houghton (1983) found, in an Australian setting, that, although loan officers believe the audit report provides useful information, uncertainty qualifications do not affect lenders’ decisions. In an experiment using U.S. loan officers, LaSalle and Anandarajan (1997) show no difference in loan granting decisions between a qualified audit report and a disclaimer of opinion. Elias and Johnston (2001) conclude that, once a going-concern contingency is disclosed in the notes to the financial statements, an explanatory paragraph describing a going-concern uncertainty appeared to convey no additional information. More recent research by Lin et al. (2003) and Bessell et al. (2003) also do not find audit report has any information content in the Chinese and Australian auditing environment, respectively. The results of Lin et al. (2003) indicate that, although loan officers perceived the qualified audit opinion as having a negative impact on the credibility of the financial statements, it does not have any significant impact on their lending decisions. Finally, Bessell et al. (2003) show that, for a financially distressed firm, the issuance of a going-concern opinion does not affect loan officers’ credit decisions. 8 Firth (1979, 1980) find that British loan officers make less favorable decisions for companies receiving serious qualifications than for those receiving an unqualified opinion. Gul (1987) concludes that an uncertainty-modified audit report significantly increased Singapore loan officers’ perceptions of risk for their clients. In the U.S. auditing setting, Bamber and Stratton (1997) find that, when a loan application included a modified audit report, bank loan officers give the application a higher risk assessment and a higher interest rate premium. These officers are also more reluctant to grant the loan. More recently, Guiral-Contreras et al. (2007) design an experiment manipulating both the sign of the audit report and other annual financial information in a series of sequential evidence. These authors find evidence that the audit report has information content but only when it is contrary to favorable financial expectations. 9 According to Bamber and Stratton (1997), the inconclusive results are possibly attributable to differences in the amounts of information provided to subjects, different cultural/auditing environments, and differences in experimental designs. However, in the current study, we suggest that an alternative explanation for the inconclusive results could be due to the omission of variables which may potentially interact with the audit report in loan officers’ decisions.

12

loan officers’ attention to the impact of this uncertainty on future cash flows. Using the

knowledge structures framework, we discuss below how a going concern opinion affects both

the capacity-capital and character knowledge structures.

In this paper we rely on the International Standard on Auditing (ISA) 570

(International Auditing and Assurance Standards Board [IAASB], 2008) to create two

different lending scenarios. ISA 570 specifies that auditors have the responsibility to evaluate

“the appropriateness of management’s use of going concern assumption in the preparation of

financial statement and to conclude whether there is a material uncertainty about the entity’s

ability to continue as a going concern” (paragraph 6 of ISA 570). ISA 570 further specifies

the kind of opinions the auditor should provide. It states that if the management has made

adequate disclosure in the financial statements regarding the uncertainty, then the auditor can

“express an unmodified opinion and include an Emphasis of Matter paragraph in the auditor’s

report” (paragraph 19 of ISA 570). If, on the other hand, adequate disclosure is not made in

the financial statements, the auditor “shall express a qualified opinion or adverse opinion”

(paragraph 20 of ISA 570). We construct our two lending scenarios based on these auditor’s

opinion guidelines in ISA 570.

In the first scenario, loan officers receive an unqualified but modified going concern

report on the borrower, where the auditor adds an emphasis of matter paragraph. In the

paragraph, the auditor discusses the existence of a material uncertainty—the loss of two main

patents—that can cast significant doubt on the client firm’s ability to continue as a going

concern. Also, the auditor draws the financial statement users’ attention to the management’s

disclosure in the footnote of financial reports. We expect that an unqualified but modified

going concern report has a negative impact on the capacity-capital knowledge structure (see

Table 1, direct effects section). The report arouses the loan officers’ doubts regarding the

13

management’s ability to operate a business capable of repaying debt on time and, what is even

more important, the ability of the firm to continue in existence (Bamber and Stratton 1997).

In the second lending scenario, loan officers face a qualified going concern opinion.

According to ISA 570, the qualified going concern report captures the auditor’s concern

regarding not only the firm’s ability to continue as a going concern but also an adequate

disclosure of this uncertainty has not been made in the financial statements. That is, whereas

the unqualified but modified report alerts financial statement users regarding the risk that the

firm may not survive in the following period, the qualified report further warns the users

about the management’s potential lack of integrity and truthfulness in financial reporting.

For the second scenario, we expect that a qualified going concern opinion will

negatively affect not only the capacity and capital (i.e., for the reasons indicated in the first

scenario) but also the character lending knowledge structure. Because in addition to the going

concern uncertainty, loan officers are alerted that the management has not adequately

disclosed a going concern uncertainty (i.e., a departure from GAAP) and hence doubt about

the integrity and honesty of the management (see Table 1, direct effects).

The above discussion relates to the direct effect of a going concern report, either

unqualified but modified or qualified, on loan officers’ knowledge structures. Next, we

discuss the direct effect of the degree of auditor’s economic dependence, as proxied by the

provision of NAS, on loan officers’ knowledge structures.

2.3 Auditor economic independence and the provision of non-audit services

Auditors are expected to act independently, without any influence from their clients’

management, when forming audit opinions. AICPA (1997) defines auditor independence as

the absence of actual or perceived interests or threats that create an unacceptable risk of bias

with respect to the quality or content of information that is the subject of an audit

14

engagement. It has been argued that the joint provision of audit and NAS to the same client

may pose an unacceptable threat to auditor independence (Krishnan et al. 2005). However, the

independence implications of auditors providing NAS to audit clients have long been a source

of controversy in the auditing literature. Certain empirical studies suggest that the auditor

independence is not negatively affected by the provision of NAS. To the contrary, they argue

that (i) the provision of NAS enhances the knowledge of the auditor regarding non-routine

activities of the client and increases auditors’ independence (Wallman 1996; Goldman and

Barlev 1974), and (ii) the increased revenues generated by auditors from the provision of

NAS lead to greater reputational capital that contributes to improve auditor independence

(DeAngelo 1981; Arruñada 1999; Dopuch et al. 2003).

In contrast, other studies question whether auditors are able to maintain independence

when, at the same time, a significant portion of their revenues come from the fees of non-

audit services (Simunic 1984; Schneider et al. 2006). If the provision of NAS creates an

economic bond between the auditor and the client firm, then the auditor would become

economically dependent on the client. Auditors may opportunistically adjust their audit

opinions in order to retain such non-audit fees in the future (Geiger and Rama 2003). That is,

contingent economic rents can heighten the auditor’s desire to maintain a long-term

relationship with the client and potentially impair auditor independence. Such desire can

trigger a favorable pre-decisional distortion of client-related information and bias audit

judgment in favor of the client (Beeler and Hunton 2002, Nelson 2009).

Results of archival studies on the effect of NAS on auditor independence are

inconclusive. While Sharma (2001), Sharma and Sidhu (2001), and Basioudis et al. (2008)

observe a negative association between NAS and the propensity to issue qualified going

concern opinions, Lennox (1999), DeFond et al. (2002), and Geiger and Rama (2003) find no

significant relationship between the issuance of going-concern opinions and NAS. Studies on

15

the relation between NAS and earnings management also have mixed results. Frankel et al.

(2002) find a positive relationship between NAS and various surrogates of earnings

management. Krishnan et al. (2005), using earnings response coefficient (ERC) as a proxy for

investors’ perception of earnings quality, provides some evidence of a negative association

between NAS purchases and perceived auditor independence. Ghosh et al. (2009), however,

find an insignificant association between the non-audit fee ratio and ERC after controlling for

client importance. Furthermore, Dhaliwal et al. (2008), Brandon et al. (2004), and Abbott et

al. (2011) find that firms seeking external financing decrease NAS purchases from their

auditors.

Contrary to the mixed results of the empirical research, the majority of behavioral

papers have found that the provision of NAS affects not only investors but also loan

officers.10 Davis and Hollie (2008) show that the disclosure of NAS provision by auditors

negatively affects investors’ decisions and leads to asset pricing inefficiencies in experimental

markets. Shockley (1981) and Teoh and Lim (1996) show that the provision of NAS is

detrimental to auditors’ independence in appearance. Gul (1991) and Knapp (1985) observe

that management is perceived as more likely to obtain its preferred resolution when auditors

also provide NAS. Firth (1980), Lowe and Pany (1995 and 1996), Lowe et al. (1999), and

Thornton (2004) find that the provision of NAS has a significantly negative impact on

lenders’ perception of auditor independence. Therefore, behavioral research provides

evidence that loan officers believe that auditor independence is compromised by the provision

of NAS (Nelson and Tan 2005).

10 Two exceptions are the studies of McKinley et al. (1985) and Bandyopadhyay and Francis (1995). McKinley et al. (1985) find that the provision of NAS does not significantly affect loan decision, financial statement reliability perception, or auditor independence perception. Bandyopadhyay and Francis (1995) conclude that the comparison between the provision of NAS and the absence of NAS does not to appear to be economically significant in lending decisions.

16

According to the knowledge structure framework, we expect that the provision of

NAS can be perceived negatively by loan officers and affects both the capacity–capital and

character loan knowledge structures (see Table 1, direct effects). Loan officers can perceive

that the provision of NAS impairs auditor independence (in appearance) and adversely

impacts the audit quality. This will lead the loan officers to discount the credibility of the

unqualified audit report and that of the financial statements. Although management’s decision

to hire its auditor for NAS can be based on cost or efficiency considerations, it could be

perceived as providing an incentive for the auditor to “collude” with management and hold

the firm to less stringent audit standards. As a consequence, the loan officers can perceive the

purchase of NAS from the auditor as a negative reflection of the management’s integrity and

honesty (i.e., management’s character).11

2.4 Going concern audit report and the provision of NAS

After discussing the potential direct effects of going concern reports and auditor

independence on the lending knowledge structures, we can now investigate the combined

effects of the information content provided by both the audit report and the provision of NAS

on lending decisions12

As described in the prior section, we construct two lending scenarios according to ISA

570 guidelines on audit reports for going concern opinions: (i) an unqualified but modified

going concern report; (ii) a qualified/adverse going concern report. For the first lending

scenario where a borrower has received an unqualified but modified going concern report, we

expect that the provision of NAS has a negative impact on the lending decisions (see Table 1,

11 Mitra and Hossain (2007) demonstrate that institutional shareholders, who monitor corporate decision-making process, induce managers to reduce the level of NAS purchase when they perceive that such action has the effect of impairing auditor objectivity in auditing the financial reports. 12 Beaulieu (1994) states that the combined relationship between character and accounting information (i.e., capacity–capital) is especially important because the presentation of accounting information is most clearly under the control of borrowers.

17

combined effects). The unqualified report does not provide any information on the borrower’s

character. The information on the provision of NAS by the borrower’s auditor fills in this gap.

The purchase of NAS from its auditor signals the potential impairment of the auditor’s

independence and serves as a warning to loan officers regarding the management’ honesty

(i.e., character). That is, in addition to the signaling effect of the going concern opinion on

the firm’s potential incapability to pay back the debt and continue in existence (i.e., capacity–

capital), firms acquiring NAS from their auditors also face potential scrutiny on their

management’s character. This discussion leads us to our first hypothesis:

H1: According to the knowledge structures framework, while a modified but unqualified

going concern report does not affect the loan officers’ assessment of the borrower's

character, the provision of NAS has a negative impact on it and leads to more

skeptical lending decisions.

For the second lending scenario, where the loan officer receives a loan request from a

borrower with a qualified going concern report, we expect there to be no significant

differences in the loan officer’s perception of audit quality between borrowers that acquire

NAS from their auditors and other borrowers (see Table 1, combined effects). The provision

of NAS does not act as a warning signal in this case because the qualified going concern audit

report communicate information not only on the capacity and capital of the borrower but also

on its character. As the loan officers already obtain the knowledge on the borrower’s

character, the NAS provision does not provide any incremental information and hence has no

impact on the loan officers’ perception or decision. Further, the issuance of a qualified

opinion signals the independence of the auditor. This leads to our second hypothesis:

H2: According to the knowledge structures framework, since a qualified going concern report already has a negatively effect on the loan officers’ assessment of the borrower's character, the provision of NAS does not cause any negative impact on the lending decisions.

18

3 Research method

3.1 Participants

The subjects of our experiment were 80 experienced loan officers (19 females and 61

males) from the second largest commercial bank (in terms of market capitalization) in Europe.

The participants range in age from 29 to 59, with an average age of 42 years old. They had an

average experience of 15 years as loan officers. Subjects were informed of the importance of

working independently on the questionnaires. We also guaranteed that their responses will

remain anonymous. One of the researchers personally administered the experiment during

different advance accounting training sessions in the headquarters of the bank.

3.2 Experimental design and research instrument

Loan officers were asked to respond to a questionnaire, which describes a request for

credit, and make their loan decisions based on the situation described. The prospective

borrower requested a loan of EUR 20 million for expanding its production capacity. The loan

term would be payable monthly over sixty months. The borrower had no prior loan history

with the subject’s bank.

We provide participants with a summary of the audit report. As mentioned before, we

create two different lending scenarios according to ISA 570. The first lending scenario

illustrates a loan request of a potential borrower who has received an unqualified but modified

going concern report with an emphasis of matter paragraph that highlights the existence of a

material uncertainty and draws attention to the footnote in the financial statements that

discloses the going concern issue. The second lending scenario presents a loan request from a

potential borrower who has received a qualified going concern report alerting statement users

that there is a material going concern uncertainty and that this issue has not been adequately

disclosed in the financial statements (i.e., the existence of a going concern uncertainty plus a

19

departure from GAAP)13. A t-test (untabulated) shows no significant difference in the lending

experience of loan officers assigned to these two lending scenarios (an experience of 15.2

years vs. 14.9 years, t-statistic = 0.127, p-value = 0.899).

In addition to the auditor’s information, we provide loan officers with descriptive

information on both the company and the industry it is in: four years of key consolidated

financial information, eleven common financial ratios for the company, and three footnotes to

the financial statements. These case materials were built from a European listed firm in the

pharmaceutical industry.14 Similar to the research design of Bamber and Stratton (1997), the

first footnote explains the firm’s significant accounting policies and the second one discusses

a long-term debt. In case of the unqualified but modified audit report (i.e., lending scenario 1),

a third footnote explaining the reasons underlying the going-concern uncertainty is included.15

Finally, we manipulate auditor economic dependence by providing loan officers with a

summary of audit and non-audit fees paid to the auditor for the last two years. In the low

auditor economic dependence scenario, the auditor does not receive any fee for the NAS; in

the high auditor economic dependence scenario, the firm acquires both audit and NAS from

the same audit firm. Additionally, in the high auditor economic dependence scenario, we

show that the firm increased its purchase of non-audit services from the auditor. The non-

audit fee ratio (non-audit fees/total audit fees) increased from 20.8 percent in t-2 to 41.1

percent in t-1, where year t is the year of the audit report.16

13 Research questionnaires are shown in appendixes 1 and 2. 14 The cases materials were reviewed by experienced loan officers from the subject bank and other financial institutions in order to ensure their relevance to the participants. 15 The going-concern uncertainty was due to the loss of the two main patents of the client company. 16 The manipulation of the degree of auditor economic dependence variable is consistent with previous research of McKinley et al. (1985), Lowe and Pany (1995) and Ghosh et al. (2009). For instance, for a sample of 21,797 firm-years observations for the period 2001-2006, Ghosh et al. (2009) find a mean for the non-audit fee ratio of 35.3%, with a first quartile of 16.1% and a third quartile of 52.6%.

20

For each lending scenario, we manipulate the auditor’s economic dependence (i.e.,

low economic dependence vs. high economic dependence) (see Table 1, combined effects).

Loan officers are randomly assigned to one of the four groups: (i) unqualified but modified

report and low economic dependence; (ii) unqualified but modified report and high economic

dependence; (iii) qualified report and low economic dependence; (iv) qualified report and

high economic dependence.17

The case first asks the loan officers to provide some demographic information.

Second, these loan officers are presented with the different lending scenarios and loan

applications. Third, they are asked to make a preliminary assessment of the borrowers’ ability

to improve their financial performance by increasing profitability and lowering leverage.

Loan officers are asked to indicate their assessment of the client’s ability to generate profits

on a 1-to-5-point scale anchored “Very low” and “Very high.” Furthermore, they are asked to

provide their assessment about the client’s ability to reduce their financial leverage on a 1-to-

5-point scale anchored “Very low” and “Very high”. These preliminary assessments on

profitability and leverage also serve as manipulation checks.

Once participants are done with the preliminary assessments, we ask them to make

their lending decisions. Loan officers first assess their risk perception of the borrower (i.e., an

overall rating on riskiness) in terms of its ability to serve the debt on a 1-to-5 scale, anchored

“Low ability” and “High ability.” Respondents are then asked to indicate their final

credit/loan recommendation on a 1-to-5 scale anchored “Definitively not recommended” and

“Definitively recommended.” We expect that our manipulation of the auditor’s economic

dependence can affect both the preliminary assessment and the lending decision of loan

officers.

17 The original sample consisted of 83 subjects (20, 22, 21, and 20 loan officers for the first, second, third and fourth treatment groups, respectively). Two incomplete responses were eliminated from the second group and one from the third group.

21

4 Results

4.1 Loan officers’ preliminary assessments–manipulation checks

Table 2 shows a summary of the loan officer’s preliminary assessment of the client’s

ability to improve its financial status in terms of both profitability and financial leverage for

our two lending scenarios. In the first lending scenario (see Table 2, Panel A, Sample 1),

where the potential borrower received an unqualified but modified audit report, loan officers

view clients receiving the report from an economically independent auditor as being more

profitable (i.e., group 1 vs. group 2: group 1 mean= 3.47, standard deviation =0.56 vs. group 2

mean= 3.16, standard deviation =0.56; t-statistic = 1.718; p-value = 0.094) and less leveraged

(i.e., group 1 vs. group 2: group 1 mean= 3.27, standard deviation =0.51 vs. group 2 mean=

2.77, standard deviation =0.68; t-statistic = 2.58; p-value = 0.014) than those getting the

report from an economically dependent auditor.

(Insert Table 2 about here)

On the contrary (see Table 2, Panel B, Sample 2), the loan officers’ profitability and

leverage assessments of a borrower receiving a qualified going-concern opinion are not

affected by the auditor’s economic dependence. In this lending scenario, the loan officers

report a mean (standard deviation) profitability assessment of 2.01 (0.94) for the low

economic dependence case compared to an assessment of 1.62 (0.72) for the high economic

dependence case (t-statistic = -1.46; p-value = 0.151). In terms of leverage assessment, the

loan officers give a mean (standard deviation) assessment of 1.83 (0.65) to a borrower

receiving the report from an economically independent auditor in comparison with the

leverage assessment of 1.62 (0.59) for an economically dependent auditor (t-statistic = -

1.025; p-value = 0.312). These preliminary results suggest that the loan officers’ assessment

22

is affected by the degree of the auditor’s economic dependence only in the unqualified but

modified going concern report lending scenario.

When we consider the case where no NAS is provided by the auditor, an unqualified

but modified going concern report does not have a negative effect on the character

knowledge structure. In fact, loan officers perceive potential borrower receiving this kind of

report as the most profitable one (i.e., group 1 vs. group 3, t-statistic = 5.92; p-value = 0.000;

group 1 vs. group 4, t-statistic = 9.01; p-value = 0.000) and the least leveraged case (i.e.,

group 1 vs. group 3, t-statistic = 7.74; p-value = 0.000; group 1 vs. group 4, t-statistic = 9.41;

p-value = 0.000). These results suggest that the unqualified audit report without NAS

provision is viewed positively by the loan officers.

Taken together, these results suggest that whether the provision of NAS negatively

affects a loan officer’s preliminary assessment of the client’s ability to generate profits and

improve its financial leverage depends on the type of audit report issued to the potential

borrower. The provision of NAS has a negative effect on the assessments when an unqualified

but modified audit report is issued but not when a qualified audit report is issued.

4.2 Lending decisions

4.2.1 Univariate analysis

Correlations for all variables used in this study are presented in Table 3. Panels A and

B show the correlation matrixes for the unqualified but modified going concern report

scenario and the qualified going concern lending scenario, respectively. Both panels illustrate

that preliminary assessments on the borrower’s ability to generate profits and reduce leverage

are positively and significantly correlated with the loan officers’ risk perceptions and lending

23

recommendations.18 Thus, a profitable and low-leveraged client is more likely to receive a

favorable risk perception and lending recommendation from the loan officers in both lending

scenarios. Furthermore, we observe a positive and significant correlation between risk

perception (i.e., credit judgment) and lending recommendation, indicating that risk perception

may in turn affect loan officers’ lending decisions (Libby 1979a; Wright and Davidson 2000).

(Insert Table 3 about here)

Table 4 Panel A shows the descriptive statistics for loan officers’ risk perceptions and

lending recommendations in the unqualified but modified audit report lending scenario (i.e.,

lending scenario 1). Similar to our findings in the preliminary assessments, a potential

borrower receiving an unqualified audit report from an economically independent accounting

firm (group 1) receives a more favorable risk-perception rating (mean = 3.41; standard

deviation = 0.43) and a better final lending recommendation (mean = 4.08; standard deviation

= 0.65) from the loan officers. A firm with high auditor economic dependence (group 2)

receives a less favorable risk-perception rating (mean = 3.15; standard deviation = 0.41) and a

less favorable lending recommendation (mean = 3.34; standard deviation = 1.01). T-tests

comparing group 1 versus group 2 shows that the differences between the two groups are

significant for both the risk-perception (t-statistic = -1.95; p-value = 0.058) and the lending

recommendation (t-statistic = -2.71; p-value = 0.010) in the first lending scenario.

(Insert Table 4 about here)

Also, similar to the results in the preliminary assessments of loans, Table 4 Panel B

shows that both loan officers’ risk-perceptions and loan decisions are not affected by the

degree of auditor’s economic dependence in the qualified going concern report lending

18 The correlation coefficients between profitability and financial leverage (i.e., the ability of the borrower to reduce their financial leverage) are also positive and statistically significant. As firms with high ability to generate profits are rated as 5 and those with high ability to reduce leverage are also rated as 5, the positive correlation between the two measures suggests that firms with higher profitability are also viewed as having low financial leverage.

24

scenario (i.e., lending scenario 2). A firm receiving a going-concern opinion is generally

assessed as a high risk borrower, regardless of whether the auditor also provides NAS (i.e.,

group 3 vs. 4, mean= 1.97, standard deviation = 0.57 vs. mean= 1.89, standard deviation =

0.57; t-statistic = -0.392; p-value = 0.698), and receives a poor credit recommendation (i.e.,

group 3 vs. 4, mean= 1.59, standard deviation = 0.82 vs. mean= 1.42, standard deviation =

0.86; t-statistic = -0.649; p-value = 0.520).

Once again, these results are consistent with H1 and H2. Our results suggest that,

according to the loan knowledge structures framework, the provision of NAS does have a

negative impact on the character knowledge structure in the case of an unqualified but

modified going concern report. However, its effect is insignificant in the case of qualified

going concern report. Further, if we compare all potential borrowers, those receiving

unqualified but modified reports from auditors not providing NAS are perceived as the ones

with the lowest risk by the loan officers19 (i.e., group 1 vs. group 3, t=7.16; p-value = .000;

group 1 vs. group 4, t-statistic = 9.50; p-value = 0.000) and receive the best lending

recommendation (i.e., group 1 vs. group 3, t-statistic = 10.57; p-value = 0.000; group 1 vs.

group 4, t-statistic = 11.01; p-value = 0.000).

4.2.2 Multivariate analysis

We examine the role played by the auditor’s economic dependence by the type of

audit report lending scenario through the 1 x 2 ANCOVAs, with loan officers’ credit

recommendation as the dependent variable, auditor’s economic dependence (AD) as the

independent variable, and the risk perception as a covariate for the two lending scenarios.

Table 5 Panel A presents the results for the unqualified but modified audit report scenario.

Both the main effect for AD (F = 4.00; p-value = 0.053) and the risk-perception covariate (F

19 As the borrowers’ riskiness is evaluated in terms of their ability to repay the loan, the lower the score, the lower the ability to repay, and hence the higher the risks.

25

= 6.30; p-value = 0.016) are statistically significant. With respect to the sample of qualified

going-concern opinions (Table 5, Panel B), the main effect for AD was not significant (F =

.26; p-value = 0.613). However, we do observe a significant effect for the risk-perception

covariate (F = 21.82; p-value = 0.000). These results provide further support for our H1 that

an auditor’s economic dependence has an impact on lending decisions only when the

borrower has received an unqualified but modified audit report. The unqualified audit report

does not set the loan officers on any alert regarding the borrower’s character. However, the

information that the borrower also purchases the NAS from its auditor prompts the loan

officers to exercise professional skepticism regarding the “credibility” of the audit report and

the financial statements. The loan officers also question the incentive of the borrower to

purchase NAS from its auditor and hence lead to a downward revision about the borrower’s

character. This update in character knowledge affects the loan officers’ risk perception about

the client and their lending decisions. In the case of a qualified going concern report, the

report itself has already caused the loan officers to revise their evaluations of the borrower’s

character downward. Hence, the knowledge of the provision of NAS does not trigger any

further revision of the loan officers’ character knowledge about the borrower. These results

support our H2.

(Insert Table 5 about here)

4.2.3 Mediation analysis

In the previous section, we examined the direct effect of auditor’s independence on

loan officers’ risk perception and lending decisions. In this section, we explore whether

auditor’s independence affects the officers’ lending decisions via credit risk perception. To

this end, we conduct a series of mediation analyses consistent with Baron and Kenney (1986)

in order to get a better understanding of how the provision of NAS (capacity-capital and the

26

character knowledge structures) influences the entire lending decision making process, i.e.,

preliminary assessments, credit judgments and lending outcomes. To perform this mediation

analysis, we estimate four regression models as defined below:

Credit Judgments = intercept + β1 Profitability + β2 Leverage

+ β3 Auditor Economic Dependence + ε (1)

Profitability = intercept + β4 Auditor Economic Dependence + ε (2)

Leverage = intercept + β5 Auditor Economic Dependence + ε (3) Lending Decisions = intercept + β6 Credit Judgments + β7 Auditor Economic Dependence + ε (4)

Equation 1 models the potential impact of both the preliminary assessments

(profitability and leverage) and the auditor’s economic dependence on loan officers’ credit

judgments. Equations 2 and 3 specify the direct impact of the provision of NAS on the

preliminary assessments of profitability and leverage, respectively. Finally, Equation 4

presents the impact of credit judgment and auditor’s economic dependence on the lending

decisions.

Figure 2 shows the regression results in the unqualified but modified going concern

opinion scenario. As expected, we observe that both preliminary assessments on profitability

(β1 = 0.293; p-value = 0.005) and leverage (β2 = 0.295; p-value = 0.003) significantly affect

credit judgments (adj. R2= 0.415) which, in turns, affects loan officers lending decisions (β6 =

0.779; p-value = 0.016, adj. R2= 0.247). We also find a direct and negative effect of auditor’s

economic dependence on loan officers’ lending decisions (β7 = -0.530; p-value = 0.053). On

the contrary, we did not find a direct impact of the provision of NAS on credit judgments (β3

= -0.025; p-value = 0.832). However, both preliminary assessments on profitability (β4 = -

0.308; p-value = 0.094) and leverage (β5 = -0.493; p-value = 0.014) are negatively affected by

auditor’s economic dependence, which suggests a potential mediation effect of the provision

27

of NAS on credit judgments.20 To this end we run the Sobel (1982) test statistic in order to

check whether the aforementioned mediated relationships are significant. We find the

presence of mediating effects on credit judgments not only through profitability (Z= -1.516, p-

value = 0.064, one-tailed) but also through leverage (Z= -1.988, p-value = 0.023, one-tailed).

Thus, the mediation analysis suggests that, apart from the direct and negative effect of the

provision of NAS on lending decisions, there is also a negative indirect influence of the

provision of NAS on credit judgments through loan officers’ preliminary assessments.

(Insert Figure 2 about here)

We repeat this analysis for the qualified audit report lending scenario (see Figure 3)

but we only find a significant direct effect of preliminary assessments (β1 = 0.281; p-value =

0.004, β2 = 0.642; p-value = 0.000, for profitability and leverage, respectively) on credit

judgments (adj. R2= 0.600), which in turn influences the lending decisions (β6 = 0.746, p-

value = 0.000, adj. R2 = 0.344). These results suggest that the auditor’s economic dependence

has neither direct nor mediating effect on loan officers’ perceptions or decisions in the

qualified audit report lending scenario.

(Insert Figure 3 about here)

4.2.4 The role played by gender, lending experience and educational background

Psychological research on judgment and decision making suggests that females tend to

be more risk averse in their judgments and behaviors than males, and that men and women

adopt different strategies in financial-decision environments (Byrnes et al. 1999). Both

accounting and finance research have also found evidence that supports the “gender

differences” argument (Barua et al. 2010; Gold et al. 2009; Chung and Monroe 2001; Dwyer

et al. 2002; Powell and Ansic 1997). For example, Barua et al. (2010) find that firms with

20 For more details, see Baron and Kenny (1986).

28

female CFOs have lower performance-matched absolute discretionary accruals and lower

absolute accrual estimation errors; Chung and Monroe (2001) obtain evidence that females

are more accurate decision makers in complex accounting decision tasks.

We test whether our results pertaining to credit judgments and lending decisions may

be affected by the fact that females and males may adopt different strategies in their role as

loan officers. Table 6 presents the comparison results for the case of unqualified but modified

audit opinion. Results of the Wilcoxon matched-pair tests (Table 6 Panel A) suggest that

female loan officers were significantly more conservative than male loan officers in the

preliminary assessments of both profitability (Z = -1.91, p-value = 0.055) and leverage (Z = -

2.81, p-value = 0.004). Further, Panel B of Table 6 shows that female loan officers’ credit

judgments (Z = -2.23, p-value = 0.026) and lending decisions (Z = -3.84, p-value = 0.000)

were also more risk averse in comparison with those of male officers.

Table 7 illustrates gender differences in the qualified audit opinion scenario. In this

case, we find significant differences between female and male loan officers only in the

preliminary leverage assessments (Z = -1.83, p-value = 0.070, Panel A) and lending

recommendations (Z = -3.62, p-value = 0.000, Panel B).

For both the unqualified but modified opinion scenario and the qualified opinion

scenario, we observe that female loan officers have higher educational background than male

officers (unqualified but modified opinion scenario: Female= 3.72 vs. Male= 3.07, Z = -3.40,

p-value = 0.001; and qualified opinion scenario: Female= 3.25 vs. Male= 2.96, Z = -1.74, p-

value = 0.082). However, there are no gender differences in terms of lending experiences in

either the unqualified (Female= 14.5 years vs. Male= 15.3 years) or the qualified lending

scenario (Female= 17.7 vs. Male= 14.3).

(Insert Tables 6 and 7 about here)

29

In addition, we carry out two-way ANCOVAs with the auditor’s economic

dependence (AD) and gender (female vs. male) as the independent variables and loan officers’

lending decisions as the dependent variable. We also add credit judgment, lending experience,

and educational background as covariates. Results are reported in Table 8. Table 8 Panel A

shows that in the unqualified audit opinion scenario both gender (F = 8.394; p-value = 0.007)

and the auditor’s economic dependence (AD) (F = 4.937; p-value = 0.033) have significant

main effects on lending decisions. These results suggest that male loan officers are more

likely to make a favorable loan recommendation and clients whose auditors do not provide

NAS are more likely to receive favorable loan recommendations. Further, their interaction

effect (AD*Gender) is also significant (F = 3.271; p-value = 0.080; R2= 0.55). This result

suggests that auditor economic dependence has a significantly more positive impact on the

male loan officers’ evaluation than on that of female officers. Credit judgment (i.e., Risk

Perception) is the only significant covariate (F = 3.769; p-value = 0.016) in this scenario.

We repeat the ANCOVA analysis for the qualified lending scenario in Table 8 Panel

B. For this scenario, we only find a main effect of Gender (F = 4.831; p-value = 0.035) on

lending decisions. Neither the auditor’s economic dependence (AD) nor its interaction with

gender is significant. Credit judgment remains as the only significant covariate (F = 22.433;

p-value = 0.000).

(Insert Table 8 about here)

4.2.5 Other robustness checks

To check the robustness of our results, instead of the five-point scale of lending

recommendation used in analyses in the prior sections, we partition the lending

recommendation variable at the mean of the full sample (i.e., mean lending recommendation

30

of all 80 observations) and use a dichotomous variable to represent the recommendation.21 It

takes on a value of 1 if the lending recommendation scale point is greater than the sample

mean and 0 otherwise.

Table 9 Panel A presents this dichotomous lending recommendation by the types of

audit reports and the auditor’s economic dependence. In the unqualified but modified audit

report scenario, results suggest that loan officers recommend ninety-five percent of the loan

requests from borrowers not purchasing NAS from their auditors, in comparison with a forty-

five percent recommendation rate for those purchasing NAS from their auditors. The

difference is statistically significant (χ2 = 3.137, p-value = 0.077; Fisher exact test = 0.182, p-

value = 0.091). This result provides further support for H1. Next, we compare the percentage

of recommendation for the qualified report lending scenario. Results show that the percentage

of recommendation stays at 5% regardless of the auditor’s economic dependence. That is, the

auditor’s economic dependence does not affect the lending recommendation rate. These

results provide additional support for H2.

Finally, we examine the relationship between auditor’s economic dependence, credit

judgment, and lending decisions using an ordered logistic regression. Table 9 Panel B shows

the results for the unqualified but modified audit report scenario while Panel C presents those

for the qualified audit report scenario. Column 1 of each panel shows the results when the

lending recommendations are regressed on the auditor’s economic dependence (AD), the loan

officers’ education, gender and experience as a loan officer. Column 2 reports the regression

results of risk perception with the same set of independent variables. Column 3 displays

results of the mediation analysis of auditor’s economic dependence on the lending

recommendation via credit judgment (i.e., risk perception).

21 Results using a dichotomous variable constructed by splitting the lending recommendation variable at the median are qualitatively the same as those reported.

31

Results in Table 9 Panel B show that auditor’s economic dependence, AD, has a

significantly negative direct effect on both the risk perception and the lending

recommendation decision in the unqualified but modified audit report scenario. In addition,

we observe that it has an indirect effect via risk perception. Results in Table 9 Panel C show

that AD does not have any significant direct effect on either the credit judgment or the lending

recommendation decision. It also does not have any indirect effect via credit judgment.

Results on the characteristics of the loan officers indicate that male loan officers are

more likely to have a positive perception regarding the borrower’s ability to repay the loan

(coefficient of Gender = 2.594, p-value = 0.001) and hence make a favorable lending

recommendation (coefficient of Gender = 1.753, p-value = 0.011) in the unqualified but

modified audit report lending scenario. However, Gender only has a significant impact on the

lending recommendation (coefficient of Gender = 2.073, p-value = 0.015) but not on the

credit judgment (coefficient of Gender = 0.263, p-value > 0.10), in the qualified audit report

scenario.

In the qualified audit report scenario, College graduate has a significantly negative

effect on the lending decision (coefficient of College Graduate = -3.235, p-value = 0.013) but

no significant effect on officers’ credit judgment regarding the client. This result suggests that

loan officers who have a minimum of college education are less likely to make a favorable

loan recommendation than those without a college degree in the qualified audit report

scenario.

(Insert Table 9 about here)

5 Conclusion

This paper investigates whether the provision of NAS affects the quality and

information content of the audit report in lending decisions using the loan knowledge

32

structures framework. Based on the response of eighty loan officers from the second largest

European commercial bank to a questionnaire, we observe that the impact of auditor’s

economic dependence on lending recommendation varies with the audit opinion the borrower

receives. When the borrower receives an unqualified but modified opinion that reflects the

auditor’s concern regarding the borrower’s ability to survive, the auditor’s economic

dependence has a negative effect on the loan officers’ perception regarding the borrower’s

risk of repayment. This in turn affects the officers’ lending recommendations. On the other

hand, if the borrower receives a qualified opinion that reflects the auditor’s concern not only

on the firm’s ability to survive but also on the financial statements’ departure from GAAP, the

auditor’s economic dependence, in terms of the provision of NAS, has no significant impact

on the loan officers’ risk evaluation and lending recommendation. The qualified audit report

signals not only the financial condition of the borrower (i.e., capacity-capital knowledge) but

also potential problems with the credibility of the financial statements and hence the

borrower’s integrity and honesty (i.e., character knowledge). While the unqualified but

modified audit report also informs the loan officers regarding the financial condition of the

borrower, it does not provide any hint on the borrower’s character. Hence, information on the

auditor’s provision of NAS fills in this gap. The purchase of NAS from the borrower’s auditor

can cause loan officers to discount the credibility of the audit report and the financial

statements. Further, loan officers may become concerned about the incentives of the

borrower’s purchase of NAS from its auditor; thus leading them to revise downward their

knowledge about the borrower’s character.

We are aware of potential alternative interpretations of our results. One can interpret

the results from a default risk approach. Default risk is the major risk lenders are exposed to.

Similar to bondholders, when lenders evaluate whether to grant loans to a firm, their major

concern is whether the firm can make the fixed contractual payments (Elliot et al. 2010;

33

Easton 2009). In the situation where the borrower received a qualified or a modified going

concern opinion, loan officers may want to gather more information about the firm’s ability to

pay its contractual obligations. The provision of NAS can affect lending decisions by

conveying signals about the firm’s future cash flow (i.e., default risk). Hence, while the

provision of NAS could proxy for audit quality, we cannot eliminate the possibility that its

impact on the lending decision is via signaling about a firm’s default risk. There are certain

limitations to our study. First, we retrieve loan officers’ credit decisions from a single banking

institution. Hence, the results may not be generalized to other institutions or legal settings.

Second, participants in the experiment do not have the opportunity to seek any additional

information on their own. Hence, their evaluations and decisions are based solely on the

information provided by us. In the real world lending considerations, loan officers likely

would acquire other information before making a decision (Kida 1984; Firth 1979). Third,

contrary to other studies (e.g., Bamber and Stratton 1997; Ellias and Johnston 2001), we do

not ask the loan officers to provide an assessment of the interest rate they would charge.

Fourth, in this experiment we employ a 1-to-5-point scale to capture the dependent variables.

Using a larger scale might be able to provide additional information or insights into loan

officers’ lending recommendation decision.22 Fifth, in the high auditor’s economic

dependence case, we have manipulated the non-audit fee ratio to increase from 20.8 percent to

41.1 percent. However, we do not ask participants whether and how the magnitude of the

change in ratio affects their credit judgment or lending recommendations in the current

experiment. This effect of change in non-audit fee ratio on the lending decision can be

explored in future studies. Finally, we only use one measure (i.e., the provision of NAS) to

capture auditor’s economic dependence and do not provide any other information on the

22 Some examples of behavioral research studies using 5 point likert scales are those of Bamber and Iyer (2002), Hermanson (2000), Hartmann and Maas (2010), Weber et al. (2002), and Maas and Mateˇjka (2009), among others.

34

auditor’s relationship with the client. For example, we do not provide the loan officers

information on the importance of the audit client to the auditor in our experimentation (Chung

and Kallapur 2003; Ghosh et al. 2009). We left this for future experimental studies.

References

Abbott, L., S. Parker, G., & Peters, G. (2011). Does mandated disclosure induce a structural change in

the determinants of nonaudit service purchases? Auditing: A Journal of Practice and Theory,

30 (2): 51–76.

Abdel-Khalik, A. R., Graul, P. R., & Newton, J. D. (1986). Reporting uncertainty and assessment of

risk: Replication and extension in a Canadian setting. Journal of Accounting Research, 24:

372–382.

Arruñada, B. (1999). The provision of non-audit services by auditors: Let the market evolve and

decide. International Review of Law and Economics, 19: 513–531.

Asare, S. K. (1990). The auditor’s going-concern decision: A review and implications for future

research. Journal of Accounting Literature, 9: 39–64.

Asare, S. K., & Wright, A. (2012). Investors’, auditors’, and lenders’ understanding of the message

conveyed by the standard audit report on the financial statements. Accounting Horizons,

forthcoming.

Bamber, E. M., & Iyer, V. M. (2002). Big 5 Auditors' Professional and Organizational Identification:

Consistency or Conflict? Auditing: A Journal of Practice and Theory, 21(2): 21–38.

Bamber, M., & Stratton, R. A. (1997). The information content of the uncertainty-modified audit

report: Evidence from bank loan officers. Accounting Horizons, 11(2): 1–11.

Bandyopadhyay, S. P, & Francis, J. R. (1995). The economic effect of differing levels of auditor

assurance on bankers’ lending decisions. Canadian Journal of Administrative Sciences, 12(3):

238–249.

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social

psychological research: Conceptual, strategic and statistical considerations. Journal of

Personality and Social Psychology, 51: 1173–1182.

Barua, A., Davidson, L. F. Rama, D. V. , & Thiruvadi, S. (2010). CFO Gender and Accruals Quality.

Accounting Horizons, 24(1): 25–39.

Basioudis, I., Papanastasiou, E., & Geiger, M. (2008). Audit fees, non-audit fees and auditor going

concern reporting decisions in the United Kingdom. Abacus, 44(3): 284–309.

35

Beaulieu, P. (1994). Commercial lenders’ use of accounting information in interaction with source

credibility. Contemporary Accounting Research, 10: 557–585.

Beaulieu, P. (1996). A note on the role of memory in commercial loan officers’ use of accounting and

character information. Accounting, Organizations and Society, 21(6): 515–528.

Beaulieu, P. (2001). The effects of judgments of new clients' integrity upon risk judgments, audit

evidence, and fees. Auditing: A Journal of Practice and Theory, 20 (2), 85–99.

Beeler, J. D., & Hunton, J. E. (2002), Contingent economic rents: Insidious threats to audit

independence. Advances in Accounting Behavioral Research, 5, 21–50.

Bessell, M, Anandarajan, A., & Umar, A. (2003). Information content, audit reports and going-

concern: an Australian study. Accounting and Finance, 43: 261–282.

Brandon, D., Crabtree, A., & Maher, J. (2004). Nonaudit fees, auditor independence, and bond ratings.

Auditing: A Journal of Practice & Theory, 23 (2): 89–103.

Byrnes, J. P., Miller, D. C., & Schafer, W. D. (1999). Gender differences in risk raking: A meta-

analysis. Psychological Bulletin, 125 (3): 367–383.

Chung, H., & Kallapur, S. (2003). Client Importance, Nonaudit Services, and Abnormal Accruals. The

Accounting Review, 78(4): 931–955.

Chung, J., & Monroe, G. S. (1998). Gender differences in information processing: An empirical test of

the hypothesis-confirming strategy in an audit context. Accounting and Finance, 38(2): 265–

279.

Chung, J., & Monroe, G. S. (2001). A research note on the effects of gender and task complexity on

the audit judgment. Behavioral Research in Accounting, 13, 111–126

Church, B., & Zhang, P. (2011). Nonaudit services and independence in appearance: Decision context

matters. Behavioral Research in Accounting, 23, 2, 51–67.

Davis, S. M., & Hollie, D. Y. (2008). The impact of nonaudit service fee levels on investors’

perceptions of auditor independence. Behavioral Research in Accounting, 20: 31–44.

DeAngelo, L. E. (1981). Auditor size and audit quality. Journal of Accounting and Economics, 3

(December): 183–199.

Defond, M., Raghunandan, K., & Subramangam, K. R. (2002). Do non-audit services fees impair

auditor independence? Evidence from going concern audit opinions. Journal of Accounting

Research, 40(4) 1247–1273.

Dhaliwal, D., Gleason, C., Heitzman, S., & Melendrez, K. (2008). Auditor fees and cost of debt.

Journal of Accounting, Auditing and Finance, 23 (1): 1–22.

Dopuch, N., King, R., & Schwartz, R. (2003). Independence in appearance and in fact: An

experimental investigation. Contemporary Accounting Research, 20: 279–114.

36

Dwyer, P., Gilkeson, J., & List, J. A. (2002). Gender differences in revealed risk taking: Evidence

from mutual fund investors. Economics Letters, 76 (2): 151–158.

Easton, P. D., Monahan, S. J., & Vasvari, F. P. (2009). Initial evidence on the role of accounting

earnings in the bond market. Journal of Accounting Research, 47(3): 721–766.

Elias, R. Z., & Johnston, J. G. (2001). Is there incremental information content in the going concern

explanatory paragraph? Advances in Accounting, 18: 105–117.

Elliot J. A., Ghosh, A., & Moon, D. (2010). Asymmetric valuation of sustained growth by bond- and

equity-holders. Review of Accounting Studies, 15: 833–878

Firth, M. 1979. Qualified audit reports and bank lending decisions. Journal of Bank Research, 9: 237–

241.

Firth, M. 1980. Perceptions of auditor independence and official ethical guidelines. The Accounting

Review, (July): 451–466.

Frankel, R. M., Johnson, M. F., & Nelson, K. K. (2002). The relation between auditors’ fees for

nonaudit services and earnings management. The Accounting Review, 77 (4): 71–105.

Geiger, M., & Rama, V. (2003). Audit fees, non-audit fees, and auditor reporting on stressed

companies. Auditing: A Journal of Practice and Theory, 22: 53–69.

Ghosh, A., Kallapur, S., & Moon, D. 2009. Audit and non-audit fees and capital market perceptions of

auditor independence. Journal of Accounting and Public Policy, 28 (5): 369–385.

Glezen, G. W., & Millar, J. A. (1985). An empirical investigation of stockholder reaction to

disclosures required by ASR. No. 250. Journal of Accounting Research, 23 (Autumn) 859–

870.

Gold, A., Hunton, J. E., & Gomaa, M. I. (2009). The impact of client and auditor gender on auditors’

judgments. Accounting Horizons, 23(1): 1–18.

Goldman, A., & B. Barlev. 1974. The auditor–firm conflict of interests: Its implications for

independence. The Accounting Review, 49 (October): 707–718.

Gray, G. L., Turner, J. L. Coram, P. J., & Mock, T. J. (2011). Perceptions and misperceptions

regarding the unqualified auditor’s report by financial statement preparers, users, and auditors.

Accounting Horizons, 25(4): 659–684.

Guiral, A. 2012. Corporate social performance, innovation intensity and financial performance:

Evidence from lending decisions. Behavioral Research in Accounting, forthcoming.

Guiral-Contreras, A., Gonzalo-Angulo, J. A., & Rodgers, W. (2007). Information content and recency

effect of the audit report in loan rating decisions. Accounting and Finance, 47(2): 285–304.

Gul, F. A. (1987). The effects of uncertainty reporting on lending officers’ perception of risk and

additional information required. Abacus, 23(2): 172–79.

37

Gul, F. A. (1991). Size of audit fees and perceptions of auditors’ ability to resist management pressure

in audit conflict situations. Abacus, 27(2): 162–172.

Hartmann, F. G. H., & Maas, V. S. (2010). Why business unit controllers create budget slack:

involvement in management, social pressure, and machiavellianism. Behavioral Research in

Accounting, 22(2): 27–49.

Hermanson, H. M. (2000). An analysis of the demand for reporting on internal control. Accounting

Horizons, 14(3): 325–341.

Holder-Webb, L., & Sharma, D. S. (2010). The effect of governance on credit decisions and

perceptions of reporting reliability. Behavioral Research in Accounting 22(1): 1–20.

Houghton, K. A. (1983). Audit reports: Their Impact on the loan decision process and outcome: An

experiment. Accounting and Business Research, 66, 15–20.

International Auditing and Assurance Standards Board (IAASB). (2008). Going Concern. ISA No.

570. New York, NY: IFAC.

Kelly, A., (1985). Business credit skills. Credit and Financial Management, (December): 11–16.

Kida, T. (1984). The Impact of hypothesis-testing strategies on auditors’ use of judgment data. Journal

of Accounting Research, (Spring): 332–340.

Knapp, M. C. (1985). Audit conflict: An empirical study of the perceived ability of auditors to resist

management pressure. The Accounting Review, 60 (April): 202–211.

Krishnan, J., Sami, H., & Zhang, Y. (2005). Does the provision of non-audit services affect investor

perceptions of auditor independence? Auditing: A Journal of Practice and Theory, 24 (2):

111–135.

LaSalle, R. E., & Anandarajan, A. (1997). Bank loan officers’ reactions to audit reports issued to

entities with litigation and going-concern uncertainties. Accounting Horizons, 11(2): 33–40.

Lennox, C. S. (1999). Non-audit fees, disclosure and audit quality. European Accounting Review, 8(2):

239–252.

Libby, R. (1979a). The impact of uncertainty reporting on the loan decision. Journal of Accounting

Research, (supplement): 35–57.

Libby, R. (1979b). Bankers’ and auditors' perceptions of the message communicated by the audit

report. Journal of Accounting Research, 17 (1): 99–122.

Lowe, D. J., & Pany, K. (1995). CPA performance of consulting engagements with audit clients:

Effects on financial statement users’ perceptions and decisions. Auditing: A Journal of

Practice & Theory, 14(2): 35–53.

38

Lowe, D. J., & Pany, K. (1996). An examination of the effects of type of engagement, materiality and

structure on CPA consulting engagements with audit clients. Accounting Horizons, 10(4): 32–

51.

Lowe, D. J., Geiger, M. A., & Pany, K. (1999). The effects of internal audit outsourcing on perceived

external auditor independence. Auditing: A Journal of Practice & Theory, 18(Supplement): 7–

26.

Maas, V. S., & Mateˇjka, M. (2009). Balancing the dual responsibilities of business unit controllers:

field and survey evidence. The Accounting Review, 84 (4): 1233–1253.

McDonald, J., & McKinley, J. (1981). Corporate banking: A practical approach to lending.

Washington, D. C.: American Bankers Association.

McKinley, S., Pany, K., & Reckers, P. M. J. (1985). An examination of the influence of CPA firm

type, size, and MAS provision on loan officer decisions and perceptions. Journal of

Accounting Research, 23 (Autumn): 887–896.

Miller, J. R., Reed, S. A., & Strawser, R. H. (1993). Bank loan officers’ perception of the new audit

report. Accounting Horizons, (March): 39–52.

Mitra S., & Hossain, M. (2007). Ownership composition and non-audit service fees. Journal of

Business Research, 60: 348–356.

Nelson, M. W. (2009). A Model and literature review of professional skepticism in auditing Auditing:

A Journal of Practice and Theory, 28(2): 1–34.

Nelson, M. W., & Tan, H. T. (2005). Judgment and decision making research in auditing: a task,

person, and interpersonal interaction perspective. Auditing: A Journal of Practice and Theory,

24(Supplement): 41–71.

Powell, M., & Ansic, D. (1997). Gender differences in risk behaviour in financial decision-making:

An experimental analysis. Journal of Economic Psychology, 18(6): 605–628.

Schmidt, J. (2012). Perceived auditor independence and audit litigation: The role of nonaudit services

fees. The Accounting Review, (forthcoming).

Schneider, A., Church, B., & Ely, K. (2006). Non-audit services and auditor independence: a review of

the literature. Journal of Accounting Literature, 25: 169–211.

Senkow, D. W., Rennie, M. D. Rennie, R. D., & Wong, J. W. (2001). The audit retention decision in

the face of deregulation: Evidence from large private Canadian corporations. Auditing: A

Journal of Practice and Theory, 24(2): 101–113.

Sharma, D. S. (2001). The association between non-audit services and the propensity of going concern

qualifications: Implications for audit independence. Asia-Pacific Journal of Accounting and

Economics, 8: 127–142.

39

Sharma, D. S., & Sidhu, J. (2001). Professionalism vs. commercialism: The association between non

Audit Services and Audit Independence. Journal of Business Finance and Accounting, 28:

595–617

Shockley, R. A. (1981). Perceptions of auditors’ independence: An empirical analysis. The Accounting

Review, 56 (October): 785–800.

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

22: 679–702.

Sobel, M. E. (1982). Asymptotic confidence intervals for indirect effects in structural equation models.

In Sociological Methodology, edited by S. Leinhardt, 290–312. Washington, D.C.: American

Sociological Association.

Strawser, J. R. (1991). The role of accountant reports in users’ decision-making processes: A review

of empirical research. Journal of Accounting Literature, 10: 181–208.

Teoh, H. Y., & Lim, E. (1996). An empirical study of the effects of audit committees, disclosure of

non-audit fees and other issues on audit independence: Malaysian evidence. Journal of

International Accounting, Auditing and Taxation, 5(2): 231–48.

Thornton, J., Reinstein, A., & Miller, C. L. (2004). Non-audit services and perceived auditor

independence post-Enron. Working Paper, Washington State University, Richland.

Wallman, S. M. H. (1996). The future of accounting, part III: Reliability and auditor independence.

Accounting Horizons, 10 (December): 76–97.

Weber, E., Blais, A., & Betz, N. (2002). A domain-specific risk-attitude scale: Measuring risk

perceptions and risk behaviors. Journal of Behavioral Decision Making, 15(4): 263–290.

Wright, M., & Davidson, R. (2000). The effect of auditor attestation and tolerance for ambiguity on

commercial lending decisions. Auditing: A Journal of Practice and Theory, (Fall): 67–81.

Table 1* The Impact of Audit Going Concern Reporting and Auditor Economic Dependence in terms of the

Provision of NAS on Loan Knowledge Structures

Loan Knowledge Structures

Character Capacity and Capital Conditions Collateral

Dire

ct

Effe

cts

Unqualified but Modified Report no Negative no no

Qualified Going Concern Report Negative Negative no no

High Auditor Dependence (high non-audit fee ratio) Negative Negative no no

Low Auditor Independence (no provision of NAS) no no no no

Com

bine

d E

ffect

s by

S

cena

rio Unqualified but Modified Report + High Auditor Dependence (high non-audit fee ratio) Negative Negative no no

Unqualified but Modified Report + Low Auditor Independence (no provision of NAS) no Negative no no

Qualified Going Concern Report + High Auditor Dependence (high non-audit fee ratio) Negative Negative no no

Qualified Going Concern Report + Low Auditor Independence (no provision of NAS) Negative Negative no no

* “no” indicates a neutral effect on these structures.

1

Table 2*

Descriptive statistics for Loan Officers’ Preliminary Assessments

Panel A: Loan Officers’ Preliminary Assessments by Sample 1 (mean, minimum, maximum and standard deviation)

Sample 1 N=40

Unqualified but Modified Audit Report Lending Scenario Profitability Assessment Leverage Assessment

Mean Minimum Maximum S.D. Mean Minimum Maximum S.D.

Group 1 (n=20)

Unqualified Audit Report + Low Auditor Economic Dependence 3.47 3 5 .56 3.27 3 5 .51

Group 2 (n=20)

Unqualified Audit Report + High Auditor Economic Dependence 3.16 3 5 .56 2.77 2 5 .68

Total 3.32 3 5 .57 3.02 2 5 .64

Panel B: Loan Officers’ Preliminary Assessments by Sample 2 (mean, minimum, maximum and standard deviation)

Sample 2 N=40

Qualified Audit Report Lending Scenario Profitability Assessment Leverage Assessment

Mean Minimum Maximum S.D. Mean Minimum Maximum S.D.

Group 3 (n=20)

Qualified Audit Report + Low Auditor Economic Dependence 2.01 1 4 .94 1.83 2 4 .65

Group 4 (n=20)

Qualified Audit Report + High Auditor Economic Dependence 1.62 1 3 .72 1.62 1 3 .59

Total 1.82 1 4 .85 1.72 1 4 .62

* Unqualified Audit Report = The company received an unqualified but modified audit report for the last accounting period; Qualified Audit Report = The company received a going concern opinion for the last accounting period; Low Auditor Economic Dependence = The audit firm did not provide any non-audit services (NAS) for the last two years; High Auditor Economic Dependence = The firm purchased NAS from its auditor for the last two years and has experienced a significant increase in the non-audit fee ratio (non-audit fees/audit fees) last year; Profitability Assessment = Preliminary assessment about the client’s ability to generate profits on a 1-to-5-point scale anchored “Very Low” and “Very High”; Leverage Assessment = Preliminary assessment about the client’s ability to reduce its financial leverage on a 1-to-5-point scale anchored “Very Low” and “Very High”.

2

Table 3

Correlation Matrix for the Dependent Variables

Panel A: Correlation Matrix for the Unqualified but Modifi ed Audit Report

Lending Scenario (Sample 1)

Preliminary Assessments Risk

Perception Lending Decision Profitability Leverage

Profitability 1 .544*** .526*** .392**

Leverage - 1 .563*** .556***

Risk Perception - - 1 .457***

Lending Decision - - - 1

*** Significant at the 1% level (two-sided). ** Significant at the 5% level (two-sided).

Panel B: Correlation Matrix for the Qualified Audit Report Lending Scenario (Sample 2)

Preliminary Assessments Risk

Perception Lending Decision Profitability Leverage

Profitability 1 .447** .590*** .384**

Leverage - 1 .729*** .556***

Risk Perception - - 1 .611***

Lending Decision - - - 1

*** Significant at the 1% level (two-sided). ** Significant at the 5% level (two-sided).

3

Table 4*

Descriptive Statistics for Loan Officers’ Risk Perceptions and Credit Decisions

Panel A: Loan Officers’ Risk Perceptions and Credit Decisions by Sample 1 (mean, minimum, maximum and standard deviation)

Sample 1 N=40

Unqualified but Modified Audit Report Lending Scenario

Risk Perception (rating-riskiness)

Decision (Lending recommendation)

Mean Minimum Maximum S.D. Mean Minimum Maximum S.D.

1 (n=20) Unqualified Audit Report + Low Auditor Economic Dependence 3.41 2 5 .43 3.27 3 5 .51

2 (n=20) Unqualified Audit Report + High Auditor Economic Dependence 3.15 2 4 .41 2.77 2 5 .68

Total 3.28 2 5 .43 3.71 2 5 .90

Panel B: Loan Officers’ Risk Perceptions and Credit Decisions by Sample 2 (mean, minimum, maximum and standard deviation)

Sample 2 N=40

Qualified Audit Report Lending Scenario

Risk Perception (rating-riskiness)

Decision (Lending recommendation)

Mean Minimum Maximum S.D. Mean Minimum Maximum S.D.

3 (n=20) Qualified Audit Report + Low Auditor Economic Dependence 1.97 1 4 .78 1.59 1 4 .82

4 (n=20) Qualified Audit Report + High Auditor Economic Dependence 1.89 1 2 .57 1.42 1 4 .86

Total 1.93 1 4 .68 1.50 1 4 .83

* Unqualified Audit Report = The company received an unqualified but modified audit report for the last year accounting period; Qualified Audit Report = The company received a going concern opinion for the last accounting period; Low Auditor Economic Dependence = The audit firm did not provide any non-audit services for the last two years; High Auditor Economic Dependence = The client firm purchased NAS from its auditor for the last two years and has experienced a significant increase in the non-audit fee ratio (non-audit fees/audit fees) last year; Profitability Assessment = Preliminary assessment about the client’s ability to generate profits on a 1-to-5-point scale anchored “Very Low” and “Very High”; Leverage Assessment = Preliminary assessment about the client’s ability to reduce its financial leverage on a 1-to-5-point scale anchored “Very Low” and “Very High”.

Table 5* ANCOVA with Loan officers’ Decisions (Lending Recommendation)

as the Dependent Variable, Risk Perception as a covariate and Audit Economic Dependence as the Independent Variable by Lending Scenario

Source of Variation

Sum of Squares df Mean

Square F-statistic p-value

Panel A: Unqualified Audit Report Lending Scenario (Sample 1)

Covariate

Risk Perception (rating-riskiness) 4.074 1 4.074 6.386 .016

Main Effects

Auditor Economic Dependence (AD) 2.552 1 2.552 4.000 .053

Residual Error 23.603 37 .638

Total 583.978 40

Total (corrected) 33.042 39

ANOVA adj. R2 .30

Panel B: Qualified Audit Report Lending Scenario(Sample 2)

Covariate

Risk Perception (rating-riskiness) 9.966 1 9.966 21.822 .000

Main Effects

Auditor Economic Dependence (AD) .119 1 .119 .260 .613

Residual Error 16.897 37 .457

Total 117.913 40

Total (corrected) 27.161 39

ANOVA adj. R2 .38

* Auditor Economic Dependence (AD) = 1, when the audit firm did not provide any NAS for the last two years, 2 when the client firm purchased NAS from its auditor for the last two years and the non-audit fee ratios (non-audit fees/audit fees) were 20.8 and 41.1 percent for the last two years, respectively; Risk Perception (Rating-Riskiness) = Perception of the ability of the company to service the debt on a 1-to-5 scale anchored “Low ability” and “High ability”; Decision (credit recommendation) = Loan officers’ credit decisions were expressed in a 1-to-5-point scale from “definitely not recommended” to “definitely recommended”.

Table 6*

Descriptive Statistics for Loan Officers’ Preliminary Assessments, Risk Perceptions and Credit Decisions by Gender (Unqualified but Modified Lending Scenario)

Panel A: Loan Officers’ Preliminary Assessments by Gender, Sample 1 (mean, minimum, maximum and standard deviation)

Sample 1 N=40

Gender Profitability Assessment Leverage Assessment

Mean Minimum Maximum S.D. Wilcoxon Z-stat

Mean Minimum Maximum S.D. Wilcoxon Z-stat

1 (n=29) Female 3.01 2 5 .55 -1.91

(p= .055)

2.53 3 5 .51 -2.81

(p= .004) 2 (n=11) Male 3.42 3 4 .56 3.21 2 5 .68

Panel B: Loan Officers’ Risk Perceptions and Credit Decisions by Gender, Sample 1 (mean, minimum, maximum and standard deviation)

Sample 1 N=40

Gender

Risk Perception (rating-riskiness)

Decision (Lending recommendation)

Mean Minimum Maximum S.D. Wilcoxon

Z-stat Mean Minimum Maximum S.D. Wilcoxon

Z-stat

1 (n=29) Female 3.03 2 4 .41 -2.23

(p= .026)

2.77 2 4 .64 -3.84

(p= .000) 2 (n=11) Male 3.37 2 5 .40 4.06 2 5 .89

* Profitability Assessment = Preliminary assessment about the client’s ability to generate profits on a 1-to-5-point scale anchored “Very Low” and “Very High”; Leverage Assessment = Preliminary assessment about the client’s ability to reduce its financial leverage on a 1-to-5-point scale anchored “Very Low” and “Very High”. The z-statistic is for the non-parametric matched-pair Wilcoxon test. The corresponding p-value is for a two-sided test.

1

Table 7*

Descriptive statistics for Loan Officers’ Preliminary Assessments, Risk Perceptions and Credit Decisions by Gender (Qualified Lending Scenario)

Panel A: Loan Officers’ Preliminary Assessments by Gender, Sample 2 (mean, minimum, maximum and standard deviation)

Sample 1 N=40

Gender Profitability Assessment Leverage Assessment

Mean Minimum Maximum S.D. Wilcoxon Z-stat

Mean Minimum Maximum S.D. Wilcoxon Z-stat

1 (n=8) Female 1.78 1 3 .93 -.18

(p= .852)

1.40 1 4 .43 -1.83

(p= .070) 2 (n=32) Male 1.83 1 4 .84 1.82 1 4 .64

Panel B: Loan Officers’ Risk Perceptions and Credit Decisions by Gender, Sample 2 (mean, minimum, maximum and standard deviation)

Sample 1 N=40

Gender

Risk Perception (rating-riskiness)

Decision (Lending recommendation)

Mean Minimum Maximum S.D. Wilcoxon

Z-stat Mean Minimum Maximum S.D. Wilcoxon

Z-stat

1 (n=8) Female 1.69 1 4 .41 -.93

(p= .351)

.66 2 4 .41 -3.62

(p= .000) 2 (n=32) Male 1.99 1 4 .77 1.71 2 5 .77

* Profitability Assessment = Preliminary assessment about the client’s ability to generate profits on a 1-to-5-point scale anchored “Very Low” and “Very High”; Leverage Assessment = Preliminary assessment about the client’s ability to reduce its financial leverage on a 1-to-5-point scale anchored “Very Low” and “Very High”. The z-statistic is for the non-parametric matched-pair Wilcoxon test. The corresponding p-value is for a two-sided test.

Table 8* ANCOVA with Loan officers’ Decisions (Lending Recommendation)

as the Dependent Variable, Risk Perception, Lending Experience and Educational Background as Covariates, Gender and Audit Economic Dependence as the Independent

Variable by Lending Scenario

Source of Variation

Sum of Squares df Mean

Square F-statistic p-value

Panel A: Unqualified Audit Report Lending Scenario (Sample 1)

Covariate Risk Perception (rating-riskiness)

Educational Background Lending Experience

1.707 .051 .052

1 1 1

1.707 .051 .052

3.769 .113 .114

.016

.739

.737

Main Effects

Gender Auditor Economic Dependence (AD)

Interaction AD x Gender

3.802 2.237

1.482

1 1 1

3.802 2.237

1.482

8.394 4.937

3.271

.007

.033

.080

Residual Error 14.949 33 .453

Total 583.978 40

Total (corrected) 33.042 39

ANOVA adj. R2 .55

Panel B: Qualified Audit Report Lending Scenario(Sample 2)

Covariate

Risk Perception (rating-riskiness) Educational Background Lending Experience

7.814 .104 .829

1 1 1

7.814 .104 .829

22.433 .299

2.379

.000

.588

.132

Main Effects

Gender Auditor Economic Dependence (AD)

Interaction AD x Gender

1.683 .329

.364

1 1 1

1.683 .329

.364

4.831 .944

1.045

.035

.338

.314

Residual Error 11.494 33 .348

Total 117.913 40

Total (corrected) 27.161 39

ANOVA adj. R2 .57

* Auditor Economic Dependence (AD) = 1, when the audit firm did not provide any NAS for the last two years, 2 when the client firm purchased non-audit service from its auditor and the non-audit fee ratios (non-audit fees/audit fees) were 20.8 and 41.1 percent for the last two years, respectively; Risk Perception (Rating-Riskiness) = Perception of the ability of the company to service the debt on a 1-to-5 scale anchored “Low ability” and “High ability”; Decision (credit recommendation) = Loan officers’ credit decisions were expressed in a 1-to-5-point scale from “definitely not recommended” to “definitely recommended”; Lending Experience = number of years of experience in lending decisions; Educational background = 1, when the participant has attended high school, 2 when he/she is a High school graduate , 3 when he/she has a Bachelor’s degree, 4 when he/she did graduate work, 5 when he/she has a Master’s degree, and 6 when he/she has a graduate degree beyond master’s degree.

Table 9

This table presents the distribution of loan officers lending’s decisions by lending scenario and economic dependence condition (Panel A). Panel B presents the results of the logistic regression of the lending recommendation. Panel C is similar to Panel B with the exception that we include the gender, education, and experience in our analyses.

Panel A: Descriptive Statistics for Loan Officers’ Recommendation Decisions by Lending scenarios and Auditor Economic Dependence Condition

Not Recommended (Recommended)

% Recommended

Lending Scenario

Unqualified but Modified

Qualified Overall

Aud

itor

Eco

nom

ic

Dep

ende

nce

Low

1 (19) 95% n=20

19 (1) 5%

n=20

20 (20) 40% n=40

High

5 (15) 45% n=20

19 (1) 5%

n=20

24 (16) 45% n=40

Overall

6 (34) 85% n=40

38 (2) 5%

n=40

44 (36) 45% n=80

* Auditor Economic Dependence = Low when the audit firm did not provide any NAS for the last two years; Auditor Economic Dependence = High when the audit firm provided NAS for the last two years and the non-audit fee ratios (non-audit fees/audit fees) were 20.8 and 41.1 percent for the last two years, respectively;

1

Panel B: Logistic Regression Analyses—Unqualified Audit Report Scenario

Lending Risk

Perception Lending

Effects Coefficient

(p-value) Coefficient (p-value)

Coefficient (p-value)

Auditor Economic Dependence (AD) -1.437** (0.012)

-0.931* (0.053)

-1.125** (0.043)

College Graduate -0.454 (0.289)

0.638 (0.211)

-0.783 (0.177)

Gender 2.594*** (0.001)

1.753** (0.011)

2.196*** (0.000)

Long Experience (i.e., >15 years) 0.079

(0.446) -0.472 (0.209)

0.223 (0.354)

Risk Perception 0.064* (0.052)

Pseudo-R2 0.097 0.037 0.108

N=40

Auditor Economic Dependence (AD) =1 when the auditor also provides NAS to the borrower and the non-audit fee ratios (non-audit fees/audit fees) were 20.8 and 41.1 percent for the last two years, respectively; 0 when the audit firm did not provide any NAS for the last two years. College Graduate = 1 when the loan officer is at least a college graduate and 0 otherwise. Gender = 1 if the loan officer is a male and 0 if female. Long Experience = 1 if the loan officer has an experience as a loan officer greater than 15 years and 0 otherwise.

2

Panel C: Logistic Regression Analyses—Qualified Audit Report Scenario

Lending Risk

Perception Lending

Effects Coefficient

(p-value) Coefficient (p-value)

Coefficient (p-value)

Auditor Economic Dependence (AD) -0.620 (0.146)

0.4457 (0.470)

-0.595 (0.159)

College Graduate -3.235**

(0.013) -0.706 (0.271)

-2.856* (0.028)

Gender 2.073** (0.015)

0.263 (0.377)

2.106** (0.014)

Long Experience (i.e., >15 years) -0.024 (0.485)

-0.020 (0.487)

0.176 (0.386)

Risk Perception 0.085*** (0.001)

Pseudo-R2 0.088 0.005 0.136

N=40

Auditor Economic Dependence (AD) =1 when the auditor also provides NAS to the borrower and the non-audit fee ratios (non-audit fees/audit fees) were 20.8 and 41.1 percent for the last two years, respectively; 0 when the audit firm did not provide any NAS for the last two years. College Graduate = 1 when the loan officer is at least a college graduate and 0 otherwise. Gender = 1 if the loan officer is a male and 0 if female. Long Experience = 1 if the loan officer has an experience as a loan officer greater than 15 years and 0 otherwise.

Figure 1

The Five C’s of Credit and the Four Loan Knowledge Structures (Beaulieu 1996)

CAPACITY & CAPITALFINANCIAL

PERFORMANCE

CHARACTERINTEGRITY, STABILITY,

HONESTY

CONDITIONSCONDITIONSINDUSTRY TYPE, ECONOMIC

CONDITIONS, BUSINESS CYCLE

COLLATERALCOLLATERALALTERNATIVE SOURCE

OF REPAYMENT

INFORMATION

EVALUATION(BOTH FINANCIAL AND

NONFINANCIAL)

LENDING DECISIONS

CREDIT

JUDGMENTS

Figure 2*

Mediation Analysis for the Unqualified but Modified

Going Concern Lending Scenario

* Profitability Assessment = Preliminary assessment about the client’s ability to generate profits on a 1-to-5-point scale anchored “Very Low” and “Very High”; Leverage Assessment = Preliminary assessment about the client’s ability to reduce its financial leverage on a 1-to-5-point scale anchored “Very Low” and “Very High”. Auditor Economic Dependence = 0, when the audit firm did not provide any NAS for the last two years, 1 when the client firm purchased NAS from its auditor for the last two years and the non-audit fee ratios (non-audit fees/audit fees) were 20.8 and 41.1 percent for the last two years, respectively; Credit Judgments = Perception of the ability of the company to service the debt on a 1-to-5 scale anchored “Low ability” and “High ability”; Lending Decisions = Loan officers’ credit decisions were expressed in a 1-to-5-point scale from “definitely not recommended” to “definitely recommended”. β1 through β7 are the estimated coefficients from the following regression equations:

Credit Judgments = intercept + β1 Profitability + β2 Leverage + β3 Auditor Economic Dependence + ε

Profitability = intercept + β4 Auditor Economic Dependence + ε

Leverage = intercept + β5 Auditor Economic Dependence + ε

Lending Decisions = intercept + β6 Credit Judgments + β7 Auditor Economic Dependence + ε

PRELIMINARY ASSESSMENTS

AUDITOR ECONOMIC

DEPENDENCE

CAPACITY & CAPITALFINANCIAL

INFORMATION

CHARACTERINTEGRITY, STABILITY,

HONESTY LENDING DECISIONS

CREDIT

JUDGMENTS

β3= -.025(p = .832)

β6= .779(p = .016)β7= -.530

(p = .053)

PROFITABILITY LEVERAGE

β2= .295(p = .003)

β1= .293(p = .005)

Adj. R2= .415

β4= -.308(p = .094)

β5= -.493(p = .014)

Adj. R2= .247

1

Figure 3

Mediation Analysis for the Qualified Going Concern Lending Scenario

* Profitability Assessment = Preliminary assessment about the client’s ability to generate profits on a 1-to-5-point scale anchored “Very Low” and “Very High”; Leverage Assessment = Preliminary assessment about the client’s ability to reduce its financial leverage on a 1-to-5-point scale anchored “Very Low” and “Very High”. Auditor Economic Dependence = 0, when the audit firm did not provide any NAS for the last two years, 1 when the client firm purchased NAS from its auditor for the last two years and the non-audit fee ratios (non-audit fees/audit fees) were 20.8 and 41.1 percent for the last two years, respectively; Credit Judgments = Perception of the ability of the company to service the debt on a 1-to-5 scale anchored “Low ability” and “High ability”; Lending Decisions = Loan officers’ credit decisions were expressed in a 1-to-5-point scale from “definitely not recommended” to “definitely recommended”. β1 through β7 are the estimated coefficients from the following regression equations:

Credit Judgments = intercept + β1 Profitability + β2 Leverage + β3 Auditor Economic Dependence + ε

Profitability = intercept + β4 Auditor Economic Dependence + ε

Leverage = intercept + β5 Auditor Economic Dependence + ε

Lending Decisions = intercept + β6 Credit Judgments + β7 Auditor Economic Dependence + ε

PRELIMINARY ASSESSMENTS

AUDITOR ECONOMIC

DEPENDENCE

CAPACITY & CAPITALFINANCIAL

INFORMATION

CHARACTERINTEGRITY, STABILITY,

HONESTY LENDING DECISIONS

CREDIT

JUDGMENTS

β3= -.154(p = .278)

β6= .746(p = .000)β7= -.109

(p = .613)

PROFITABILITY LEVERAGE

β2= .642(p = .000)

β1= .281(p = .004)

Adj. R2= .600

β4= -.203(p = .312)

β5= -.388(p = .151)

Adj. R2= .344