Specialist Project

Order Instructions/Description
the Article that needs critique is below.

The Effects of the Level of Assurance,
Accounting Firm, Capital Structure, and
Bank Size on Bank Lending Decisions
This study examines the effect of the type of accounting service (audit,
review, compilation, and no report), the accounting firm providing the
service, the capital structure of the borrower, and the bank size on loan
officers’ lending decisions. The findings show the level of service had an
impact on the amount of the loan, but not on the interest rate awarded
nor on the decision whether or not to grant a Une of credit. The accounting
firm (large Intemational with good reputation versus local with unknown
reputation) providing the accounting service, however, did not significantly
affect the bank loan decision. Capital structure of the borrower had
a significant impact on the decision to grant a line of credit, the loan size,
and the interest rate. In addition, the loan size and interest rate were
significantly infiuenced by the size of the bank granting the loan.
1. Introduction
Reporting on the credibility of financial information has been an important
concern of the accounting profession. Over the last several decades, the U.S. accounting
profession increased the number of attestation levels available to nonpublic
businesses and has actively tried to educate the users of financial statements
about these assurance levels (Boyd, Boyd, and Boyd, 2000-2001).
The primary purpose of this paper is to examine the effect of various levels
of an independent accountant’s association with the financial statements on bank
loan officers’ perceptions and decisions. The levels of assurance examined in this
paper were an audit, a review, a compilation, and no accountant association.
This paper is organized as follows. The research questions and related literature
are presented first. Then the contributions of this study and the research methods
*College of Business Administration, Augusta State University
**Aeeounting Department, Mays College of Business, Texas A&M University
The authors gratefully acknowledge the valuable guidance and insight of Richard A. White. In
addition, the authors are grateful for the valuable comments and insights of an associate editor and
used are addressed. The remaining sections of the paper are the statistical analysis,
discussion of the results, and conclusions.
2. Literature Review and Research Questions
The chief aim of the accountant’s report is to add credibility to the accompanying
financial information. In other words, the attest function assists the user
in evaluating the quality of information that is being received (cf., Committee on
Basic Auditing Concepts [1973]). Financial statements accompanied by an audit
should be less biased and finer (i.e., provide more information) than unaudited
financial statements (Ng [1978]). In addition, a review should result in financial
statements that contain less bias and that are finer than a compilation. Thus, the
amount of risk to the bank should decrease as the level of accounting service
increases, assuming all other factors remain constant. This study’s main research
question is the following:
Does the type of accounting service, and consequently the type of report, affect
a line of credit decision made by commercial bank lending officers?
Specifically, this research will examine whether the level of accounting attestation
will affect the loan officers’ decision to grant a line of credit. Then, for those cases
where a line of credit decision is granted, the effect on the loan size and interest
rate will be analyzed.
After extensive published research, there remains debate, and ongoing research
efforts, regarding the impact of the accountant’s report on bank lending decisions.
Blackwell, Noland, and Winters (1998) affirmed the conventional wisdom, and ofttested
hypothesis, that audited firms pay their lenders (i.e., banks) a lower interest
rate (in their sample, on average, 25 basis points lower) than unaudited firms.
Wright and Davidson (2000) examined the effect of auditor attestation and
tolerance for ambiguity on a commercial lending decision. The subjects were from
six financial institutions in Canada. The results indicate that the risk assessment
judgment was affected by tolerance for ambiguity, but risk assessment was not
affected by auditor attestation (audit, review, and no accountant association). Consequently,
the levels of attestation did not affect the loan approvals. The researchers
also asked the bankers to make an interest rate recommendation on the loan regardless
of whether the loan was approved or not. They found no differences
between a review and an audit in the interest rate recommendations. However, the
low-tolerance subjects charged a lower interest rate for some attestation (i.e., audit
or review) than in the no attestation treatment. The opposite occurred for the hightolerance
subjects. This group charged a lower interest rate for no accountant’s
assurance than for some assurance (i.e., audit or review). These results appear
Bandyopadhyay and Francis (1995) investigated the effect of differing levels
of auditor assurance on bankers’ lending decisions and found that the rate of loan
approval increased and the interest rate decreased, as the level of assurance increased.
A major flaw to this study, however, is that bank loan offlcers evaluated
all 12 scenarios, which allowed them to compare the different levels of assurance
before making their loan recommendations. This within-subjects design for such a
study has been criticized for the potential “demand characteristics.” That is, the
subjects know that they are involved with an experiment, and when presented with
all the potential scenarios, there is “a tendency to ‘help’ the experimenter by replying
in the manner which the subject perceives that the experimenter desires”
(Pany and Reckers [1987, p. 41]). Pany and Reckers conducted an experiment comparing
the within-subjects design with a between-subjects design. They found that
the within-subjects design led to strong demand effects that significantly influenced
the responses of the participants.
The effect of the level of accounting service on bank loan officers’ decisions
was also examined by Baker (1990). Baker did not examine a compilation, but did
look at an audit and a review for a generally accepted accounting principles
(GAAP) basis statement and an income tax basis statement. Consistent with previous
studies, the level of accounting service did not appear to affect the bank
lending decision.’
Johnson, Pany, and White (1983) manipulated the independent variable (accounting
service) at four levels: an audit, a review, a compilation, and no accountant’s
association. This study found the type of accountant’s report did not affect
the bank lending decision. Specifically, neither the decision to loan the company
money nor the decision pertaining to the interest rate to be charged was affected.
The results of this and other studies indicate that either the level of accounting
service does not affect a bank lending decision or the signals from accounting
reports are not sufficiently strong to overcome the signals emitting from the other
financial information (Houghton [1983]).
Although not dealing directly with compilation and review services, studies in
related areas have examined the effects of differing assurance levels on creditors’
decisions. For example, Johnson and Pany (1984) found the level of auditor assurance
on forecasted information did not affect a bank lending decision. Specifically,
whether auditors reviewed the forecasted information did not affect loan
officers’ interest rate recommendations. Carter (1984) also found that bank loan
officers’ lending decisions were not affected by the various levels of assurance on
forecasted information.
Another factor that may influence a bank loan officer’s perceptions regarding
an independent accountants’ report is the independent accounting firm conducting
1. In Baker’s study, the loan officers were also asked to assume that if they (or a competitor)
had approved the loan, what interest rate should be charged. Therefore, the loan officer’s analysis
included interest rates on loans that were not approved. Baker’s findings showed a 0.3 percent (statistically
significant, p-value = 0.03) lower interest rate recommendation for the audit than for the review.
It would be difficult to make any generalization about this finding, however, since interest rates on
unapproved loans were included.
the service. A survey conducted hy Strupeck and Figlewicz (1984) found that
bankers indicated the accounting firm issuing the report was an important factor in
making a lending decision. A second research question is presented here:
Does the perceived quality of the firm issuing the independent accountant’s
report affect the bank loan officers’ decision?
Researchers (e.g., DeAngelo [1981]; Dopuch and Simunic [1980]; Nichols and
Smith [1983]) have presented arguments that large CPA firms provide audits that
are perceived to be of higher quality than audits performed by smaller firms. Empirical
studies pertaining to differentiation in auditor quality have not been conclusive.
Results from studies by Ettredge, Shane, and Smith (1988); Nichols and Smith
(1983); Wallace (1978); and Wilson (1982) suggest, however, that audits by large
accounting firms are associated with a favorable impact.^ Nonetheless, an experiment
performed by McKinley, Pany, and Reckers (1985) showed that the type
(size) of CPA firm did not affect a bank loan decision based on audited financial
statements. They examined the impact of an unqualified opinion, but did not examine
any other type of audit report or level of accounting service.
Other factors are also considered in this study to provide a richer context.
Specifically, the strength of capital structure of the company applying for the loan
and the size of the lending institution were also examined. Firms with a strong
capital structure should be viewed by bank loan officers as being more credit worthy
than firms with a weak capital structure. Thus, the former should receive more
favorable borrowing terms (Melnik and Plaut [1986]). fn addition, a loan decision
may be the size of bank for which the commercial bank loan officer works. Miller,
Reed, and Strawser (1993) found that, in some instances, bank loan officers from
large banks had significantly different perceptions of certain audit reports than
bankers working for smaller banks. This study also reported that bank loan officers
from larger banks may rely more on the independent accountant’s report than lenders
from smaller banks.
3. Contribution of this Study
A motivating factor in conducting this study is that prior studies have not
clearly demonstrated that the accountant’s report affects a user’s decision, even
though the accounting profession intends a clear distinction in the assurance levels
2. Wallace (1978) showed that the size of the auditor (i.e., Big Eight versus non-Big Eight)
significantly improved the net interest cost and the bond rating for municipalities. Wilson (1982) also
demonstrated that a significant relationship existed between municipal bond ratings and size of auditor.
Nichols and Smith (1983) showed that a switch from a non-Big Eight auditor to a Big Eight auditor
resulted in a positive market reaction, and a change from a Big Eight to a non-Big Eight auditor had
a negative impact. However, the market reaction was not significant. Similar results occurred for Ettredge,
Shane, and Smith (1988), who found that Big Eight auditors had a favorable effect on the
correlations between eamings forecast errors and cumulative abnormal stock retums, although the significance
levels were low.
given by various accountants’ reports. An implication of these studies is that firms
may be paying for unnecessary services when, for example, the intent of these
firms is to infiuence creditors’ decisions through an accountant’s report. Previous
studies hypothesized that the interest rate would decrease as the level of assurance
(i.e., accounting service) increased. In other words, higher risk (less assurance)
would yield higher rates of return (interest rates). However, in a bank lending
decision, a bank’s exposure is also in the amount that it lends, not just the interest
rate it charges.
Our study adds to the research that examines the issue of whether the level of
assurance provided by an independent accountant’s report affects users’ decisions.
It differs from prior research in certain significant respects. First, this study considers
not only the effect of an accountant’s level of assurance on interest rates,
but also on the loan amount. Prior research on the effect of differing levels of
assurance on a user’s decision have concentrated on the interest rate recommendations
(such as Blackwell, Noland, and Winters [1998]). Second, this study introduces
variables not considered in previous studies on user’s decisions. Specifically,
the accounting firm issuing the report, the capital structure of the company, and
the size of the bank are considered in this study. In addition, these factors may
interact with the level of accounting service. Third, a larger and, in most cases, a
broader sample than previous studies was used. More than 500 bank loan officers
responded to the survey. In contrast, the samples for the three studies that most
closely pertain to this study were as follows: Wright and Davidson (2000) had 75
bankers respond from a total of 6 Canadian banks; Bandyopadhyay and Francis
(1995) had 67 bankers respond from 10 banks located in three cities; and Johnson,
Pany, and White (1983) had 98 bankers respond. Finally, a between-subjects design
was used, overcoming some of the limitations incurred from a within-subjects
4. Research Methods
An experiment using a questionnaire as the test instrument was chosen to
accomplish the objectives of this study. The variables used in this study are discussed
first. Next, the sample selection and experimental task are presented.
4.1 The Variables
Three independent variables were used in the initial model for this study. The
main independent variable, and the focal point, was the level of accounting service.
Four types of accountant’s association were used—an audit, a review, a compilation,
and no accountant’s report.
The accounting firm conducting the service was manipulated at two levels—
a large/international firm with a good reputation and a local firm whose reputation
was unknown to the bank lending officer. Therefore, tbe dichotomy is not just
between large and small accounting firms, but it is also between reputations because
the international accounting firm is described as having a good reputation while
the local accounting firm’s reputation is unknown to the bank loan officer. The
disparity between the two types of accounting firms was wide in order to test
whether the firm had any effect on the lending decision.
The capital structure was set at two levels—strong and weak. Specifically, the
company with the strong capital structure had a debt/equity ratio equal to the upper
quartile ratio given by Robert Morris Associates (RMA). The company with the
weak capital structure had a debt/equity ratio equal to the lower quartile ratio given
by RMA. The companies used in this study were office supply and equipment
retailing firms.
To help isolate the effect of the capital structure, the income statements for
the two companies were identical from sales through operating income, which was
$122,500 in the current period. Interest expense differed because of the difference
in the debt/equity ratio. The company with the strong capital structure had less
debt. Therefore, it had less interest expense than the company with the weak capital
structure. Net income was $72,000 for the company with a strong capital structure
and $55,000 for the company with a weak capital structure. Assets, which totaled
$890,700, were the same for both companies.
The decision variables were the commercial bank loan officer’s loan amount
(line of credit) and interest rate recommendations. Line of credit contracts play a
significant role in bank credit allocations in the United States (Melnik and Plaut
[1986]). A binomial decision of whether or not to grant a company a specified
amount would be unlikely to provide the information necessary to measure the
effect of the independent accountant’s report. In fact, previous studies that requested
a particular loan amount and used a between-subjects design (i.e., Johnson
et al. [1983]; Hicks [1982]; Baker [1990]) found that the level of assurance had
no effect on the loan decision. While the loan amount is typically specified in a
loan request, a commercial loan officer should be able to estimate a line of credit
amount, as was requested in this study. Without such a controlled experiment, the
informational value of the accountants’ reports would be difficult to measure. The
loan officers were asked to state the maximum amount of the line of credit they
would be willing to grant the company and the minimum interest rate, stated as an
amount above prime, they would charge.
The interest rate is usually specified when a line of credit is granted. Normally,
the borrower will ask for a particular line of credit amount. In many loan decisions,
however, the amount loaned is often negotiated, particularly in line of credit decisions.
Therefore, a bank lending officer should be able to estimate a maximum
line of credit amount, even though it is customary for the customer to make a
request to borrow a particular amount. In other words, a bank lending officer could
compute a maximum line of credit amount, although it would not be revealed to
the customer unless the amount requested by the customer exceeded that maximum
amount. Thus, the realism of the task is not impaired by asking loan officers to
specify a maximum line of credit amount.
A line of credit decision gives the banker some flexibility in determining the
amount to lend. Prior to drawing the sample, the questionnaire was examined by
four experienced commercial loan officers. All felt that the questionnaire was realistic
and provided adequate information to allow a bank lending officer to make
a recommendation on the loan amount. Often a customer may request less than the
amount needed. A potential mistake in a lending decision is not giving the customer
enough money to carry on their operations. Thus, a good commercial loan officer
should have the ability to advise a customer on the amount of funds needed by a
loan applicant, not just to make a decision on whether to accept or reject a loan.
The average commercial lending experience of the respondents in this study was
approximately nine years with a standard deviation of about seven years. Therefore,
the respondents were highly capable of making the lending decision called for in
the study.
4.2 Sample Selection and Experimental Task
Participants in the experiment were selected from two sources: a list of bank
loan officers obtained from a banking school and a list of Federal Reserve Banks.
Three restrictions were placed on the sample selected from the list of Federal
Reserve Banks. First, the sample was restricted to 13 states, located in the United
States of America.^ This restriction was made because over 94 percent of the commercial
bank lending officers from the banking school list worked in these states.
This restriction was made also because the geographical location may have an
impact on the lending decision (Arnold and Diamond [1981]). Second, any bank
name appearing on the list of banks obtained from the banking school was deleted
from the Federal Reserve list to avoid the possibility of a bank officer responding
to more than one questionnaire. Third, banks that had less than $50,000,000 in
assets were eliminated from the Federal Reserve list because lending officers from
smaller banks normally have limited experience with audited financial statements
and with large international accounting firms (Libby [1979]). After making these
eliminations, a systematic sample was used to select 490 Federal Reserve Banks.”
A total of 636 bank loan officers was obtained from the banking school list. Thus,
a total of 1,126 bank loan officers was selected for inclusion in the study.
The questionnaires contained one level of accounting service (i.e., either audit,
review, compilation, or no accountant’s report), the reasons for making an application
for a line of credit, financial statements for the last two years, footnotes to
the financial statements, some key financial ratios of the company along with industrial
averages computed by RMA, and a brief description of the company and
the principal owner. The questionnaire that did not have an accountant’s report
3. These states were Alabama, Arkansas, Florida, Georgia, Illinois, Kentucky, Louisiana, Missouri,
Mississippi, Oklahoma, South Carolina, Tennessee, and Texas.
4. In systematic random sampling, the first item between 1 and NIn (where N is the population
size and n is the sample size) is selected randomly, then each (N/n)th item is selected from the population
(Roberts [1978]). The questionnaires sent to the selected banks were addressed to the “Chief
Commercial Credit Officer.”
clearly stated that the financial statements had not been audited, reviewed, or compiled
by an accounting firm. Based on these data, the commercial bank loan officers
were asked to make two decisions on the line of credit application: (1) the amount
to lend and (2) the interest rate to charge (expressed as points above prime). In
addition, the loan officers were also asked several questions regarding their perceptions
of the accountant’s report. The responses to these questions should provide
insight into the bank loan officers’ lending decisions. The appendix shows these
perception questions and their scale endpoints.
5. Results
Of the 1,126 commercial bank loan officers asked to complete the questionnaire,
512 responded. This represents a response rate  of 45.5 percent (49.1 percent
from the banking school and 40.8 percent from the list of Federal Reserve Banks).’
The responses obtained from the commercial bank loan officers who were associated
with the banking school did not differ significantly from the lending officers
who were randomly selected from the list of Federal Reserve Banks. The average
commercial lending experience of the respondents was 9.2 years (7.8 years for the
sample from the banking school and 11.5 years for the sample selected from the
list of Federal Reserve Banks). In addition, nearly 78 percent of the subjects held
a college degree, and more than 96 percent had at least some college education.
The loan officers were asked to make two credit decisions: the amount to lend
and the interest rate to be charged. First, a chi-square test was used to see if the
level of accounting service infiuenced the decision to grant a line of credit. Then
for the loans on which a line of credit was granted, three-way ANOVAs were used
to see if either the loan size or interest rate recommendations were affected. The
three factors in the ANOVA model were the type of accountant’s report (audit,
review, and compilation), the type (size/reputation) of the accounting firm (large
international firm with a known reputation and local firm with an unknown reputation),
and capital structure (strong and weak).’^ The number of observations for
each cell varied in size from 25 to 40.
5. To test for nonresponse bias, an analysis of early responses with later responses found no
differences as to when the questionnaires were completed (Oppenheim [1966]).
6. Multivariate analysis of variance (MANOVA) was used to determine if any of the independent
variables had a significant effect on the bankers’ loan decision. The results based on the Wilks criterion
showed that capital structure was significant at the 0.0001 level and the report effect had a p-value of
0.1064. The CPA firm main effect and the interaction effects were not significant. In MANOVA, the
effects of the treatments on the dependent variables are examined simultaneously. Thus, for analysis
purposes, a response must be recorded for each dependent variable. Thirty observations could not be
used in the MANOVA model because the respondents did not state either an interest rate or a loan
amount. Twenty-eight of these 30 observations were due to the fact that the participant recommended
that no line of credit be given to the company, and thus they did not state an interest rate. A 3 X 2 X
2 MANOVA was run using an interest rate for the zero loan recommendations. The mean interest rate
for each combination of factors was used. By including these zero loan amounts In the MANOVA
model, the p-values for the report effect decreased to 0.0252. However, the results concerning the CPA
firm and capital structure effect were not changed. The CPA firm effect was still not significant (p =
0.2548), and the capital structure effect remained significant (p = 0.0001).
Results for the Line of Credit (Loan) Decision
Response Variable
Line of credit (loan) amount
Loan size
Report effect
Accounting firm effect
Company (capital structure) effect
Report * firm effect
Report * company effect
Firm * company effect
Report * firm * company
Interest rate
Report effect
Accounting firm effect
Company (capital structure) effect
Report * firm effect
Report * company effect
Firm * company effect
Report * firm * company
F Value
P Value
5.1 Analysis of the Level of Assurance
Results of the chi-square test indicate that the level of assurance (audit, review,
or compilation) did not affect the loan officer’s decision on whether to grant a line
of credit. As expected, the number of nonapproved loans increased as the level of
assurance decreased. However, it was not statistically significant. Bank loan officers
rejected 17 loan applications accompanied by a compilation, 11 accompanied by a
review, and 10 accompanied by an audit.
Table 1 presents the results of the main effects of the ANOVA models for
each response variable where a line of credit was granted. Thus, the zero loan
amounts are not included in these ANOVAs. None of the interaction terms were
significant (at p = 0.05), although it appears that there may be a slight three-way
interaction effect. An examination of the mean scores for each cell indicates that
a company with a strong capital structure which received an audit by a local firm
received the highest loan recommendation, while reviews and compilations given
by a local firm for a company with a weak capital structure had the lowest loan
amount recommendations.
The accountant’s report (level of service) had a significant effect on the size
of the line of credit.” Since there were more than two levels of report, a multiple
7. A 3 X 2 X 2 ANOVA was also run including the zero loan amount recommendations. The
report effect (p = 0.0093) and the capital structure effect (p = 0.0001) were significant. No other
effects were significant at the 0.05 level. Thus, the results are similar to those shown in Table 1, but
Mean Scores of the Subjects’ Responses Regarding
Decision Variables
Capital structure
Loan Size
(Line of Credit)
Interest Rate
(Above Prime)
2.16% [
Note: Means linked by a common bracket do not differ significantly
from each other. Means not linked together differ significantly
at an alpha level of 0.05 (using Tukey-Kramer’s multiple comparison
comparison technique was used to determine which factor levels differed significantly.
The Tukey-Kramer method of multiple comparisons was selected because
it gives sufficient power while protecting against the experimentwise error rate.
When the sample sizes are not equal, Tukey-Kramer provides a conservative estimate
(Kleinbaum and Kupper [1978]). An alpha level of 0.05 was used. The line
of credit recommended for the company when the financial statements were audited
was significantly different from the line of credit granted when the financial statements
were compiled. An audit did not result in a statistically significant different
line of credit than a review; nor did a review result in a statistically significant
different line of credit than a compilation.
The mean scores of the subjects’ responses are presented in Table 2. These
averages do not include the zero loan recommendations. The mean loan size increased
as the level of accounting service increased. While the loan size recommendations
were affected by the level of assurance, the interest rate was not
significantly affected. The largest difference in the mean scores for the interest rate
was only one tenth of 1 percent.
The effect of no accountant’s report was also analyzed. This factor was not
included in the 3 X 2 X 2 ANOVA because there is no accounting firm involved
when a report is not issued. The percentage of rejected loans (i.e., no line of credit)
followed expectations. No accountant’s report resulted in a 16.2 percent rejection
the report effect shows up even stronger for it includes the zero loan amount recommendations. As
noted in the paper, more zero loan amounts were given for a compilation than a review, and more were
given for a review than an audit.
ANOVA Results for No Accountant’s Report
Response Variable
Loan Size
Report effect
Company (capital structure) effect
Interest rate
Report effect
Company (capital structure) effect
F Value
P Value
Local Accounting Firm
F Value
P Value