Audit quality, audit pricing, and information asymmetry

MISSING IMAGE

Material Information

Title:
Audit quality, audit pricing, and information asymmetry the implications of peer review
Physical Description:
vii, 156 leaves : ill. ; 29 cm.
Language:
English
Creator:
Payne, Jeffery Lynn, 1959-
Publication Date:

Subjects

Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1995.
Bibliography:
Includes bibliographical references (leaves 151-155).
Statement of Responsibility:
by Jeffery Lynn Payne.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002057558
notis - AKP5584
oclc - 33832529
sobekcm - AA00004718_00001
System ID:
AA00004718:00001

Full Text















AUDIT QUALITY, AUDIT PRICING, AND INFORMATION ASYMMETRY:
THE IMPLICATIONS OF PEER REVIEW














By

JEFFERY LYNN PAYNE


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA













ACKNOWLEDGEMENTS

I am indebted to several individuals whose contributions facilitated the

completion of this research. Professor Robert Knechel, my dissertation committee

chairperson, provided guidance, counsel, and encouragement. His commitment to the

completion of my degree is greatly appreciated. Professor Karl Hackenbrack provided

detailed critique of my design and analysis which substantially improved this research.

I sincerely appreciate his assistance. I was fortunate to receive direction from

Professors John Lynch and David Sappington, through coursework and throughout the

development and completion of this research. I was greatly impressed by the

willingness of my committee members to share their time and expertise with me. This

research has been greatly improved by their collective contributions.

Several other individuals and institutions were instrumental in the completion

of this research. I acknowledge the assistance of the entire faculty of the Fisher

School of Accounting for allowing me to recruit subjects from their classes and for

providing insightful comments during the workshop presentations of this research.

The completion of the experimental market would not have been possible without the

market program received from Professor Ron King at Washington University which

was modified for this research. The Bureau of Economic and Business Research at

the University of Florida provided the experimental lab and the assistance of Clinton

Collins. Financial support for this research was provided by the Deloitte & Touche







Foundation. Lastly, I am grateful to my fellow Ph. D. students for their friendship,

encouragement and participation in my pilot studies. I look forward to collaborating

with them in the future.

Finally, I am forever indebted to my wife, Teena, for her allowing me to

pursue my ambitions. Her support and encouragement were instrumental to the

completion of this research and along with my children, Zach and Garrett, she helped

me to remember what is really important in life.














TABLE OF CONTENTS

ACKNOWLEDGEMENTS ....................................... ii

ABSTRACT ................................................. vi

CHAPTERS

1 INTRODUCTION AND BACKGROUND ....................... 1

Introduction ............................................. 1
Purpose of the Research ..................................... 4
Organization of Remaining Chapters ........................... 5

2 REVIEW OF RELEVANT RESEARCH ......................... 6

Agency Theory .......................................... 6
The Role of Auditing ...................................... 7
Empirical Results ......................................... 8
Experimental Economics ................................... 10
Audit Quality and Peer Review .............................. 14
The Present Research ..................................... 20

3 DEVELOPMENT OF MODELS AND HYPOTHESES ............. 22

Introduction ............................................ 22
Model Settings and Definitions .............................. 22
Analysis of Player Strategies ................................ 34
H ypotheses ............................................ 40

4 RESEARCH DESIGN AND EXPERIMENTAL METHODS .......... 44

Introduction ............................................ 44
Laboratory M markets ...................................... 45


1








5 RESEARCH RESULTS ................................... 57

Introduction ............................ ............... 57
R results ............................................... 58
Summary ............................................. 66

6 SUMMARY AND CONCLUSIONS .......................... 81

Summary of the Study .................................... 81
Summary and Discussion of Results ........................... 81
Lim stations ............................................ 83
Future Research ......................................... 84

APPENDICES

A EXPERIMENTAL MATERIALS ............................. 87

B INTERNAL VALIDITY CHECKS ........................... 136

C PLAYER'S STRATEGIES ................................. 140

D MARKET RESULTS ..................................... 145

REFERENCE LIST ............................................ 150

BIOGRAPHICAL SKETCH ..................................... 156













Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

AUDIT QUALITY, AUDIT PRICING, AND INFORMATION ASYMMETRY:
THE IMPLICATIONS OF PEER REVIEW

By

Jeffery Lynn Payne

August 1995

Chairperson: W. Robert Knechel
Major Department: Accounting

Historically, the accounting profession, through various oversight committees

and boards, has promulgated standards to regulate the quality of the auditing function.

One such regulation adopted by the AICPA was the creation of the peer review

process. The purpose of the present study is to address the implications of the peer

review process on audit demand, audit quality provision, and audit pricing within the

auditing market.

Based on the economic incentives of market participants, hypotheses were

developed to examine these implications. This study was conducted utilizing the

experimental markets methodology to examine various institutional settings by

manipulating the demand for audit quality and the availability and timing of the peer

review process.







The results of this study demonstrate that the information provided by the peer

review process is useful to buyers in assessing the quality of services they are

receiving. The results indicate that the heterogeneous demand for audit quality creates

a disparate valuation of the peer review process. In markets where audit quality was

not important to buyers the ability to identify high quality providers by utilizing peer

review was of minimal value. In markets where high quality audit services were

demanded, the information provided by a 1-yr peer review process increased the

quality of audits provided in the market. The 3-year review process currently utilized

within the profession did not significantly alter the market results from the non-peer

reviewed markets.

These results imply that the peer review process has the potential to address

regulators concerns regarding the quality of the auditing but that the timeliness with

which the information is received is critical. The results of this study support the

ability of the peer review process to provide quality differentiating information to

buyers of audit services that improves their ability to purchase high quality auditing

services, if desired.













CHAPTER 1
INTRODUCTION AND BACKGROUND

Introduction

The market for audit services has received broad attention in the extant

accounting literature. Initial investigations addressed concerns about the lack of

sufficient competition within the auditing market as expressed by governmental and

regulatory agencies (e.g., Metcalf Committee Report, U. S. Senate [1977]). In

contrast, practitioners (Public Accounting Report [1992]) and accounting oversight

boards (American Institute of Certified Public Accountants [1978]) have been

concerned that there is too much competition present in the auditing market. Research

findings (Simunic [1980], Dopuch and Simunic [1980], and Danos and Eichenseher

[1986]) support the view that extensive competition exists in the market for audit

services which may lead to deterioration of audit quality and a reduction of auditor

independence. While the concern about auditor independence has been reduced

somewhat by the findings of DeAngelo [1981a],' the concern about audit quality has

escalated, fueled by the increasing rate of litigation from audit failures, and





'DeAngelo [1981a] found that when future quasi-rents are available to auditing
firms, if the client companies incur transaction costs to switch auditors and there are
initial start-up costs for new auditors, the occurrence of low balling should not affect
independence, because the low balling itself does not create the future benefits. It is
just a competitive response to obtain the future benefits.









Congressional [1978] and GAO [1986] reports criticizing audit firm departures from

GAAS.

One necessary, but not sufficient, condition for the establishment and

maintenance of high quality audit services is the ability of audit fees to adequately

compensate firms that make the required investments in personnel, audit technologies,

and training to allow for the production of high quality service. Economic theory,

(e.g. Akerlof [1973]), indicates that for high quality services to be compensated

adequately, consumers must be able to differentiate between quality levels provided in

the marketplace. However, product differentiation is a perplexing area for CPA firms

because it may be difficult for clients to differentiate quality, ex ante or ex post,

among competing firms. Further confounding this asymmetry is the potential presence

of a heterogeneous demand for audit quality between prospective clients. These

factors may create a setting which encourages high quality producers to exit the

market as they may not be able to recover their costs of production.

The extant research on product differentiation has focused mainly on the

examination of audit firm size as a proxy for quality by examining

* fee differentials on continuing engagements (e.g., Simunic [1980]),

* litigation experience (Palmrose [1988]),

* stock market reaction to auditor switching (e.g., Nichols and Smith [1983]),

* IPO valuation (Menon and Williams [1991]),

* auditor switching (Chow and Rice [1982]), and







3

* the presence and magnitude of "low-balling" for initial audit engagements (e.g.,

Simon and Francis [1988]).

While the general result from this research is that audit firm size serves as a

quality surrogate, there are several inconsistencies present. No relationship has been

found between audit firm size and an auditing firm's propensity to issue qualified

opinions (Chow and Rice [1982]), the magnitude of "low-balling" on initial

engagements (Simon and Francis [1988]), or auditor switching before an initial public

offering (Menon and Williams [1991]). Additionally, while the extant literature on the

competitive nature of the auditing market indicates that the market is competitive with

a fee differential to large firms who participate in the small auditee market

(Palmrose[1986] and Francis and Simon[1987]), no direct evidence has been found

concerning what factors allow a firm to price their services above the competition and

still obtain an engagement.

Experimental economics has also been used to investigate the audit quality

issue. In this type of investigation, participants exchange experimental commodities or

services at negotiated prices or values in an experimental market. Theories are

developed, based on individual economic rationality, to predict the outcomes of each

market's transactions. Numerous studies have utilized this research approach to

examine

* auditors incentives and decisions when their quality choices cannot be directly

observed (e.g., Dejong, Forsythe, Lundholm, and Uecker [1985]),

the demand for strategic auditing, (Kachelmeier [1991]), and









* low-balling for initial engagements (Schatzberg [1990]).

Each of these studies has extended our understanding of how the individual

economic incentives of participants in the auditing market influence decision making.

However, extant laboratory market investigations of the auditing market have not been

able to isolate sufficient market conditions which could sustain the development of

high quality audit services.

Purpose of the Research

The concerns regarding audit quality have not been ignored by the accounting

profession. The American Institute of Certified Public Accountants (AICPA) issued

the Commission on Auditor Responsibilities: Report Conclusions and

Recommendations (AICPA, 1978) which contained a specific recommendation that

peer review be used to foster improvement in the performance of accounting firms.

Subsequently, the AICPA established the AICPA Division for Firms to organize,

conduct, and oversee the provision of peer review services. The objective of the peer

review process is to ensure that accounting firms are complying with generally

accepted auditing standards in their provision of auditing and accounting services.

Peer review is a costly process that provides a potential monitor of audit quality to

prospective purchasers of audit services and users of financial statements.

The purpose of the present study is to address the implications of the peer

review process on audit demand, audit quality provision, and audit pricing within the

auditing market. Based on the economic incentives of market participants, hypotheses







5

will be developed to examine these implications utilizing the experimental markets

methodology.

Organization of Remaining Chapters

Chapter 2 will review the contributions from various methodological and

theoretical paradigms which address the issues of product differentiation among

auditors and incorporate a discussion of the objectives of the present research.

Chapter 3 will provide the theoretical development and statement of hypotheses to be

tested. Chapter 4 will develop the research design and experimental methods.

Chapter 5 discusses the research results. A summary of the study and conclusions are

presented in Chapter 6.













CHAPTER 2
REVIEW OF RELEVANT RESEARCH

Agency Theory

Agency theory investigates the contractual relationship between a principal and

an agent. The principal is typically assumed to have a utility function that is

increasing in agent performance and decreasing in payments made to the agent for

their services. The agent is assumed to have a utility function increasing in payments

received from the principal and decreasing with effort exerted. In these settings

optimal outcomes can only be achieved under certain conditions. One such setting is

when the agent's actions and private information are perfectly observable by the

principal. Unfortunately, this setting is often unrealistic for application to "real-world"

contractual negotiations. Therefore, the agency literature has expanded to include

numerous solutions to the principal/agent problem which attempt to minimize the

inefficiency created by information asymmetry, or agency costs, resulting from the

inability of the principal to observe perfectly the agent's actions and motivations.

One such solution, originally proposed by Harris and Raviv [1979] and

Holmstrom [1979], is that any costless monitor, even if imperfect, that provides some

positively correlated information regarding the actions of an agent or the overall state

of nature should be used by the principal to reduce the agency cost arising from

information asymmetry. If a monitor does not provide useful information it's use









could potentially increase the risk of the agent and lead to welfare reducing decisions.

Additionally, the authors proposed that if the monitor is costly, the benefits received

from use of the monitor should exceed the cost to ensure that net societal welfare is

improved.

The Role of Auditing

The role of the auditor has been theorized to be a potential monitor of the

principal agent relationship between the owner/investors (equity or debt) and manager

of a company. Ng [1979] posits that the value of auditing services lies in the fact that

it enhances the value of the economic information generated by the accounting

process. Auditing is viewed as assisting the user of financial statements in assessing

the quality of those statements. Ng and Stoeckenius [1979] note that the demand for

auditing arises due to the nature of the contract between managers and owners, where

owners cannot directly observe the output of the firm (i.e., financial results). In this

setting with asymmetric information, the owners/investors must rely on the earnings

reported by management to make their investment and compensation decisions. Audits

represent a costly, yet imperfect, monitor of the principal/agent relationship.

If auditing is demanded to reduce and/or monitor agency costs, do

owners/investors have a homogeneous demand for auditing services?

DeAngelo[1981b] makes the argument that auditors and their respective market find

each other based on the quality of audit product demanded, and supplied by the audit

firm. Consequently, some clients, such as a nonpublic or public company with large









management ownership, might not require the highest quality audit.' Other clients,

especially widely held public companies, potentially demand a higher quality audit

given the increased agency cost of professional management. Therefore, distinct levels

of audit quality may be demanded for different levels of agency monitoring (Wilson

[1983]).

Empirical Results

If heterogenous levels of audit quality are desired, how is audit quality to be

defined and differentiated? Given that it is costly for consumers of audit services to

evaluate audit quality, DeAngelo [1981b] argues that auditor size serves as an efficient

surrogate for audit quality. This is because larger firms, in an attempt to portray a

consistent level of quality to the market, have a reduced incentive to lower audit

quality to fit a particular engagement.2 Therefore, large firms should have higher

overall audit quality.

Research has examined evidence of heterogeneous effects created by audit firm

size differentials regarding auditor litigation (Palmrose [1988]), stock valuation

(Nichols and Smith[1983] and Eichenseher, Hagigi, and Shields[1989]), IPO valuation

(Menon and Williams [1991]), auditor switching (Chow and Rice [1982]), audit fee




'In these situations the motivation for hiring the auditor might be to simply fulfill
regulatory requirements, or to attain some other service or benefit from the auditor in
addition to the attest function.

2An issue not addressed by this research is the potentially divergent incentives
between the individual auditor and the firm. Auditors are motivated by factors such as
time pressure, client acquisition requirements, fee collections, and office profitability
which might create incentives to disregard firm quality control standards.









differentials (Simunic [1980], Palmrose [1986], Francis and Simon [1987], and Simon

and Francis [1988]), and the incidence and magnitude of "low-balling" on initial

engagements (DeAngelo [1981a], Ettredge and Greenberg [1990], Schatzberg [1990],

Simon and Francis [1988], and Turpen [1990]). The quality/size proxy most often

used in this research was the classification of firms as Big Eight vs. non-Big Eight.

While the general result from this research is that audit firm size serves as a

quality surrogate, Simunic [1980] notes that differentiated products are not observed

directly but rather are revealed by differences in prices which are associated with

differences in observed product characteristics. He notes that care must be taken when

assessing price differentials and other measures of audit quality given all the variables

that could affect auditing quality.

Supporting Simunic's propositions, Carcello et al. [1992] isolated the following

four factors, which may or may not be correlated with audit firm size, as predictors of

audit quality based on questionnaires received from financial statement preparers,

auditors, and users: 1) audit team and firm experience with a client, 2) industry

expertise, 3) responsiveness to client needs, and 4) compliance with generally accepted

auditing standards. Therefore, while firm size is the easiest proxy for auditor quality

to examine given the availability of data, this proxy is not necessarily capturing all the

quality differentiating information available to clients.

Simunic's prudence is confirmed by the presence of several inconsistencies in

the research findings regarding the ability of audit firm size to effectively proxy for

audit quality. No relationship has been found between audit firm size and an auditing







10

firm's propensity to issue qualified opinions (Chow and Rice [1982]), the magnitude of

"low-balling" on initial engagements (Simon and Francis [1988]), or auditor switching

before an initial public offering (Menon and Williams [1991]). Additionally, while the

extant literature on the competitive nature of the auditing market indicates that the

market is competitive with a fee differential to large firms who participate in the small

auditee market (Palmrose[1986] and Francis and Simon[1987]),3 no direct evidence has

been found concerning what factors allow a firm to price their services above the

competition and still obtain an engagement.

Whether the inconsistent results are driven by the poor specification of the

theoretical models utilized, or the inability of the quasi-experimental methods applied

to adequately control for potential confounding variables, is unclear. Many factors, as

mentioned earlier, could be proxied by firm size, only one of which is audit quality,

which is itself not defined in a consistent manner in the extant research.

Experimental Economics

Experimental economics provides a means to directly test an economic theory

while controlling for the effects of potentially confounding variables. Experimental

settings give the theory its "best shot" in a controlled setting such that, if the theory

fails in this setting, it's ability to handle more robust situations is questionable.

Theories which are supported in a laboratory market can later be examined under

different methodological paradigms to determine their robustness. Several excellent




3Palmrose and Francis and Simon classified companies with assets of less than
$40MM as "small".









reviews of laboratory market research have been written (e.g., in the economics

literature see Plott [1979] and Smith [1980 and 1982]).4 Smith [1982] characterizes an

economics experiment as a market environment which consists of economic agents, or

the experimental participants, the characteristics they possess, and the commodities

they are endowed with. Institutional allocation rules within the experiment control the

communications, trading environment, property rights and the rules of cost imputation.

Experimental settings usually involve manipulating the institutional setting, within the

market environment to test hypotheses regarding the equilibria sustained for

commodity prices and volumes, and the affects of such decisions on the allocative

efficiencies within the market.

When utilizing the experimental economics methodology, it is necessary to

control both the environmental and institutional parameters to maximize the proposed

theory's predictive ability. Additionally, the preferences of the experimental

participants must be controlled. Experimental economics relies on the theory of

induced demand (Smith [1976] and Plott [1979]) for this purpose. By using induced

demand, the experimenter can create supply and demand conditions for an imaginary

commodity in the laboratory by mapping commodity transactions into cash payments.

Several experimental precepts must be considered and complied with for the induced






4Vernon Smith also produces an ongoing series entitled Research in Experimental
Economics, most recently published in 1991. Smith, Schatzberg, and Waller [1987]
provide a summary of the application of experimental economics to the study of
auditing markets.









demand theory to be effective. Smith [1982] defines these precepts as privacy,

dominance, non-satiation, and saliency.5

Numerous laboratory market studies have been performed in the extant

literature. Of relevance to this paper are several studies which have examined the

demand for audit services. In particular, a study by DeJong, Forsythe, Lundholm, and

Uecker [1985], hereafter DFLU, examined the price and quality of services delivered

by sellers (auditors) to buyers in market settings where the sellers were not bound by

their quality representations made to buyers during the negotiation process. Sellers

could agree to deliver high quality services and subsequently deliver lower quality

without the buyers' knowledge. This creates a setting where the effort level of the

seller is not directly observable to the buyer, or an environment with moral hazard.

DFLU found that auditors were initially able to command higher prices by consistently

providing services above the minimum quality level in their multiperiod markets.

However, they could not isolate conditions which provided sufficient returns to prevent

auditors from opportunistically reducing their quality in later periods. To further

examine this relationship, they allowed buyers to purchase investigations regarding

auditor quality and operationalized various legal liability environments. Neither






'Briefly, nonsatiation requires that subjects always prefer more to less. Saliency
requires that subjects understand the link between their action choices and the reward
received. Dominance is achieved by paying subjects sufficient rewards such that any
potential affects of other characteristics in their utility functions are controlled.
Privacy controls for subject reactions to payments made to other participants. All
rewards must be private information.









modification significantly affected their results regarding the inability of auditors to

earn sufficient profits to maintain the provision of high quality services.

DeJong, Forsythe, and Lundholm [1985] likewise were unable to isolate

conditions to mitigate the effects of information asymmetry regarding auditor quality.

They found that sellers (auditors) utilized pricing strategies to "rip-off' buyers. Sellers

would price their services at high levels but provide low cost/quality services in

settings where buyers could not observe seller effort. The gains to sellers from

utilizing this "rip-off' strategy were greater than from the provision of consistent, high

quality service.

In a later study by Brozovsky [1990], experimental subjects were allowed to

purchase an asset that had a depreciating reputation value. Sellers (Auditors) who did

not purchase reputations earned the greatest profits in the experiment. Sellers who

purchased this reputation did command higher prices, but not at sufficient levels to

recover their reputation investment. Therefore, the investment in reputation was not

efficient from the seller's perspective.

While it is possible to question the assumptions and market variables applied in

each of these experimental markets, the consistent finding that sellers (auditors) were

not able to earn sufficient returns to encourage the maintenance of high quality

services is a significant result. While the ability to make inferences to actual market

environments from experimental settings is limited, this result follows the intuition and

concern of "real world" audit market participants that audit quality is not sufficiently

rewarded in audit prices.









Audit Quality and Peer Review

Historically, the accounting profession, through various oversight committees

and boards, has promulgated standards to regulate the quality of the auditing function.

However, this privilege is not guaranteed as noted by the Metcalf Committee Report

from the U.S. Senate (1977), which was very critical of the accounting profession's

self-regulation efforts and the quality of the attest function. The Committee was

especially interested in the profession's policies established to monitor and maintain

audit quality.

Partially in response to the Metcalf Committee Report, the AICPA issued the

Commission on Auditor Responsibilities: Report Conclusions and Recommendations

(Cohen Commission, 1978). A specific recommendation cited in the report was that

peer review be used to foster improvement in the performance of accounting firms. In

1977, the AICPA established the AICPA Division for Firms to provide peer reviews.

The AICPA Division for Firms contained two separate sections, the SEC Practice

Section (SECPS) and the Private Companies Practice Section (PCPS). Membership in

these organizations was voluntary and after joining the SECPS/PCPS firms had 12

months to undergo their first peer review. Thereafter, a peer review is performed

every third year.

Briefly, a peer review involves the examination of a firm's quality control

procedures, including compliance with generally accepted auditing standards such as

independence, proper training, and proper supervision. A sample of the reviewed

firm's audit workpapers are examined to ensure that the audits were properly planned,









performed, and reviewed and that the auditor's opinions were supported by sufficient

evidential matter. Reviewed firm's can contract with the AICPA Division for Firms or

a qualified member firm to conduct their peer review.

During the past fifteen years since the creation of the AICPA Division for

Firms, the market for accounting services has undergone radical changes. The

prohibitions on advertising and client solicitation have been lifted and competitive

pressures are continually increasing. Additionally, the recent problems within the

thrift, banking, and real estate industries have convinced regulators and legislators that

there is a need to improve the quality of the attest function (Berton [1990]).

Following the recommendations of the Metcalf Committee, the SEC recommended in

1987 that all CPA Firms involved in the audit of publicly traded companies undergo

peer review. In 1989 the AICPA membership adopted such a requirement for all firms

that perform audits of publicly held companies. For those firms to retain their

membership in the AICPA, they must be a member of the SECPS of the AICPA

Division for Firms. An effect of the AICPA and SEC actions was to increase the

regulation of the auditing market by placing an additional cost on firms wanting to

maintain their AICPA membership and/or their publicly traded clients.6'7



6The SEC does not require membership in the AICPA for auditing firms engaged
by public companies, so firms can continue to perform public company audits without
joining the SECPS if they are not members of the AICPA.

7Previously, in 1987, the AICPA's members approved the requirement that all
member firms of the AICPA participate in a mandatory quality review program.
Quality review differs from peer review in several ways. The most notable is that
quality reviews are not "public" information. This research will focus on the peer
review process where the resulting reports are public information.









This requirement was not well received by AICPA members from small

accounting firms. In a survey of small accounting firms, many of whom had

voluntarily joined, and subsequently withdrawn from the SECPS or PCPS, Wallace

[1988] found that firms' concerns centered mainly on the added cost of undergoing

review, the lack of perceived benefit from the entire peer review process, and the

inability of membership in the AICPA Division for Firms to increase client contacts

and generate new business.8 These concerns appear to be justified as the AICPA has

acknowledged that the cost of on-site reviews could place a disproportionate expense

on firms with only a small number of audit clients (Huff and Kelly [1989]).

Peer review is a costly process which provides a potential monitor of audit

quality to prospective purchasers of audit services and users of financial statements.

Therefore, it is necessary to examine the benefits provided by the peer review process

to ensure that the benefits of peer review exceeds it's costs. The extant literature on

the benefits derived from peer review is limited. One study by Francis, Andrews, and

Simon[1990] looked for evidence that peer-reviewed audit firms in the AICPA

Division for CPA Firms are perceived as quality-differentiated auditors as evidenced

by the ability to command a fee premium. If a sufficiently large fee premium exists,

then audit firms should undergo review voluntarily. If not, then regulation might be

required to coerce firms into the peer review process. Using data from 1984 and



8While the direct cost of peer review is moderate for small firms, the indirect cost
of review could be substantial. The overall cost of complying with a new set of
quality control procedures, increased continuing education and liability insurance
requirements, and required concurring partner review on all public company audits,
could substantially increase the smaller firm's costs.







17

1985,9 for client companies with sales of less than $125MM and audits performed by

non-Big Eight auditors, the authors found no systematic fee differential between peer

reviewed and non-peer reviewed firms. The authors present two interpretations of this

result. The first is that peer review costs are trivial so a fee differential is not required.

However, they note that the lower percentage of membership among small firms is

prima facie evidence that there is an economic cost to joining the AICPA Division for

Firms. Second, the auditing market is not willing to pay for a quality differentiated

audit. This gives audit firms no incentive to join even if the cost is trivial. This

interpretation supports the aforementioned concerns of firms regarding the cost benefit

relationship of peer review.'0

Deis and Giroux [1992] found peer review to be a significant explanatory

variable in determining audit quality, as defined for reviews of auditor workpapers for

audits of Texas Independent School Districts by the Audit Division of the Texas

Education Society. Their findings are the first to show that peer reviewed firms

maintain higher quality levels. However, their study did not examine the ability of

peer review to increase audit fees.






9Note that this time period is before the AICPA mandated peer review requirement
became effective.

1An alternative interpretation is that the econometric techniques utilized were not
able to efficiently isolate the effects of peer review from the numerous confounding
factors that are present in the auditor selection process. This interpretation motivated
the selection of the experimental markets methodology for this research so that
potentially confounding factors could be controlled.









How societal welfare is affected by the peer review of auditors is not directly

measurable. The consumer potentially receives a higher quality product. However, an

argument can also be made that many consumers of audit services do not require, or

desire, a higher quality and higher cost audit. Another concern is the effect of the

requirement on the participants in the audit market. The effect of peer review is

unlikely to be consistent across firms. Small firms, who wish to continue their

membership in the AICPA, may perceive peer review as an increase to their cost of

operations which reduces their ability to compete in the auditing market. A

contrasting view is that if clients find it difficult to determine an audit's quality, lower

quality firms might be benefitting from the overall quality perception of the

profession. Lower quality firms could also have a competitive advantage over other

firms due to their lower costs of operation.

Another concern relates to the quality of attest services provided. Regardless

of the characteristics of audit demand that exists in the auditing market, anecdotal

evidence suggests that the current monitoring devices such as Ethics Committees and

State Oversight Boards, which require suspected low quality auditors to be "turned-in"

by their peers, are ineffective at maintaining quality at the prescribed standard for the

profession (GAO [1986]). Therefore, while rational consumers are aware that

differential quality levels are provided by auditors, the lack of differentiable quality

information induces consumers of audit services to value the audit product uniformly









across suppliers." In this setting, the profession is ripe for the entry and success of

"free riders" who carry the proper certifications but do not uphold the quality

standards of the profession.2 Currently, these individuals and firms are difficult to

identify due to the lack of a sufficient monitoring system. This could contravene the

effectiveness of the attest function for outside parties who rely on the audited financial

statements to make informed contractual decisions.

Peer review represents the accounting profession's response to regulators

concerns about the quality of the attest function. The impetus behind peer review was

to provide all audit clients and their related investors with an acceptable level of audit

quality (Huff and Kelly [1989]). However, an important incremental benefit from peer

review is that auditing firms might be able to differentiate their services based on their

inclusion in the peer review process. Peer review could help eliminate "free riders" by

forcing them to adhere to a set of minimum quality standards. This would reduce the

provision of low quality service in the profession and therefore increase the overall

quality assessment of all firms, even if firm specific reputations cannot be developed.

Product differentiation is a perplexing area for CPA firms because it is difficult

for client firms to differentiate between their services and those of a competing firm.

This tends to make the demand curve more elastic and increases firm's incentives to




"For example companies have been shown to perceive audit quality as
homogenous across firm classification as Big Eight or non-Big Eight (e.g. Simunic
[1980]).

'2This depends on the other barriers to entry that may be present. However, such
barriers are likely to be lower, or less costly to achieve, for lower quality firms.







20

engage in price competition. This is evidenced by accounting professionals responding

to a recent Public Accounting Report [PAR] FAX poll [1992] which indicated that

fee-based competition is increasing. Eighty-six percent of respondents characterized

fee-based competition as high or extreme. An analysis of auditor changes during 1987

in PAR [1988], revealed that fee considerations were the most often cited reason, by a

2-to-1 margin, for changing auditors. Thirty-seven (Sixteen) percent of auditor

switches from Big Eight (other) firms were motivated by the ability to reduce fees.

The Present Research

The purpose of the present study is to address the implications of the peer

review process on audit demand, audit quality provision, and audit pricing within the

auditing market. More specifically, I will examine the influence of peer review on

auditor's quality decisions and the impact of the peer review process on purchasers of

audit services in assessing the quality and related value of audit services. The primary

question addressed in this study is whether peer review provides information to

purchasers of audit services which allows sufficient audit fees to encourage and

support the provision of high quality auditing services.

Given that audit quality has been notoriously difficult to define and measure, I

will define audit quality as audit firm compliance with GAAS.'3 This is the standard

monitored by the peer review process and has been identified by Carcello et al. [1992]



13The aforementioned litigation explosion and the GAO report [1986] on audit
quality which found that 34% of governmental audits examined had serious departures
from GAAS support the proposition that a significant number of audits do not comply
with GAAS.









as one of the major factors potential clients consider when assessing auditor quality.

This research will rely on the peer review process as a monitor of auditor quality

(compliance with GAAS) which avoids many of the difficulties associated with the

audit size proxy.14 Peer review is public information which indicates whether a firm

is, or is not, complying with GAAS. In this setting firm size is not important. All

firms who comply with GAAS can receive an unqualified peer review report.
































'4This definition of audit quality does not allow examination of the possibility that
firms can provide audit services of a higher quality than that required to comply with
GAAS. This is an area for future research.













CHAPTER 3
DEVELOPMENT OF MODELS AND HYPOTHESES

Introduction

The objective of this section is to develop a buyer-auditor model that is both

compatible with the theoretical features of research interest and receptive to laboratory

market operationalization. I will define the market environment, determine the buyers'

and auditors' optimal behavior, and discuss the related incentives of the players within

each market setting.

Model Settings and Definitions

Consider a market in which every individual (player) can be either a buyer' or

an auditor. A buyer can employ the services of an auditor to attest to the current state

of nature (S). The state of nature represents the true financial position, stated in

accordance with GAAP, of the audited company.2 The state of nature will be



'The buyer will represent individuals who utilize audited financial statements in
making contracting and investing decisions. This could be an investor who seeks to
monitor the owner/manager, or an owner who seeks to monitor the manager. A buyer
could also represent equity investors or shareholders, debt investors or creditors, and
include other groups that are affected by management behavior such as employee
groups.

2A possible effect that is not investigated by this research is the influence of audit
quality on management effort. If management is confident that it's diligent efforts will
be accurately reported by the audit this might increase managers' incentives to exert
more effort. This research exogenously provides probabilistic information about the
state of nature to buyers and does not consider the manager's incentives regarding the
accurate reporting of the true state of nature.

22







23

dichotomous taking on a low value (SL) or a high value (SH), respectively (SL,SH, E S).3

Information asymmetry is present in the market because buyers are not able to observe

the state of nature directly. Ex ante, the buyers believe that the state of nature is SH

with probability t, or P(S=SH)=), and conversely P(S=SL)=1-).4 Ex post, the buyers'

information about the state of nature will be based on the audited financial statements.

The payoff to the buyer will be determined by the report of the audited financial

statements. Therefore, the buyer receives compensation based on the audited financial

statements report and not the true state of nature. Note that the buyer must retain an

auditor in order to receive compensation.5

An auditor's ability to attest to the actual state of nature is dependent upon their

investment in audit quality. Each audit's quality (q) can be either high (q) or

low (qL), respectively (qH, qce q). The cost of providing qH (qL) quality is represented

as CH (CL). Incurring CL implies that in performing the audit the auditor might not

perform sufficient procedures and/or does not come to the correct assessment from

procedures performed.6 Low quality audits will have more difficulty detecting the true


3Dichotomous variables will be utilized in the formulation of the model to
facilitate analytic tractability and reduce notational complexity.

4This proxies for buyers private information about the true state of nature.

5Buyers receive compensation in the experimental setting only by purchasing
auditor services. This relates to the "real" world in that companies cannot be traded
on a public stock exchange without audited financial statements. Additionally, audits
are usually required by financial institutions who provide debt capital. Therefore,
auditing is often a prerequisite to attaining debt or equity capital.

6This could be due to either an insufficient level of training or knowledge
regarding GAAP and GAAS applications or the opportunistic reduction of audit
quality provided.







24

state of nature and will be less likely to discover a material departure from GAAP. To

provide high quality services, the auditor will be required to make additional

investments in audit quality (CH > CL).7 Incurring CH increases the likelihood that the

true state of nature will be detected.8

For each contract completed between an auditor and buyer, the auditor will

privately observe a probabilistic signal (a,) on the state of nature.9 The following

conditional probabilities capture the interaction among the auditor's report (R), the

actual state of nature (S), and the audit's quality level (q,) (a, for i = qH, qL):

P(R= SI q= q )= a

P( R # SI q = qi ) = 1-ai

The auditor's ability to observe the true state of nature will be an increasing function

of the audit quality (qi) level chosen, or aH > aL.10





7These investments could consist of quality increasing measures such as continuing
education training, concurring partner review, and firm adherence to a quality control
plan that address numerous areas of control procedures within the firm itself.

8Given that audit quality is dichotomous, there will be no lag between the quality
investment decision and the ability to provide the corresponding quality of service.
Auditors can choose to make quality improving or reducing decisions at the beginning
of each period by selecting the quality they will provide before contracting with
buyers.

9Each buyer can purchase an audit during each period. Auditors can contract with
more than one buyer.

'OThe audit's ability to incur each type of reporting error is symmetrical for states
of nature that are high or low. While this does not capture the potential reality that
this relationship is not symmetrical, the qualitative results are not affected by this
assumption.









Each auditor will then issue a private report (R), that represents the audited

financial statements, to their buyer indicating a high (H) state or a low (L) state,

respectively (H, L e R). Based on this report the buyer will receive a high (low)

payoff denoted as XH (XL). The auditor will be required to truthfully report the state

observed."

The decisions that will emerge from the aforementioned market setting will be

affected by the competitive pressures in the marketplace, the players preferences

including their demand for audit quality, and the beliefs of the buyers regarding the

quality of the auditor's services. To consider each player's preferences it is important

to remember that each player's objective is to maximize his or her own expected

utility. Buyers derive utility from contracting with an auditor for reporting services.

Auditors derive utility from being engaged by buyers at compensatory rates for

reporting services.12






"Note that, if the quality decision is made before contracting, adverse selection is
present, if after, moral hazard is present. This model addresses the adverse selection
issue by not allowing the auditor to deviate from their quality decision and requires
truthful reporting of the outcome observed.

'2Given that the state of nature and auditor quality levels are uncertain, expected
utilities cannot be evaluated without a specific assumption as to the risk preferences of
the market participants. Consistent with extant research, throughout this paper I will
assume that the buyers and auditors are risk-neutral. If the assumption of risk
neutrality does not hold, players behaviors could potentially deviate from the solutions
predicted by the theoretical model due to characteristics of their personal utility
functions that were not adequately controlled in the experimental setting. See DeJong,
Forsythe, Lundholm, and Uecker [1985] for a detailed discussion of the risk neutrality
assumption.









The Buyer's Valuation of Audit Services

The buyer's valuation of audit services requires explicit consideration of two

factors. First, the impact of auditor reporting errors must be considered. To clarify,

auditor reporting errors can occur under two scenarios. First, when (R=LIS=SH) the

state was high but the auditor did not detect the true state which results in a payoff to

the buyer of XL instead of the appropriate payoff of XH. Second, when (R=HI S=SL),

the true state of nature was low and the payoff received should have been XL.

However, the audit attested to financial statements reporting the state as high so the

buyer received XH. When this occurs, a penalty (X) that represents costs generated by

inaccurate audited financial statements will be incurred by the buyer.'3 This penalty is









"The ability of various liability regimes to affect auditor effort has been
previously examined (e.g., DFLU [1985]) and will be explicitly excluded from this
experiment. Therefore, the auditor's only loss from providing low quality service is
the potential effects if, and when, this information becomes known. No direct
penalties will be assessed to the auditor for misreporting. Note that in a utopian world
the use of achieved audit risk as a measure of audit quality would be optimal. While
such measures may be useful in cases in which audit risk is readily evident ex post, it
may be difficult or impossible to ascertain audit risk in cases where the effect is long
delayed and partial. Audit failure might not show up for several years, and there
might then be doubt as to whether it was caused by the audit, by the client, or by
some other circumstance. An alternative and perhaps less expensive means for
averting quality deterioration may be some form of screening device by application of
licensing or some form of minimum quality standard (Leland [1979]). This conforms
to the information provided by the peer review process and addresses the actuality that
investors incur substantial losses from audit failure. Investors potentially lose a
substantial portion of their investment currently, while the auditors loss is proportional
and incurred at a much later date when the litigation is finally settled.







27

imposed in a subsequent period when the true state of nature is revealed.4 Therefore,

the market will generate an incremental cost for buyers for all periods where the

auditor's report does not equal the true state of nature or (R S).

Next, the buyer's preference for audit quality must be considered. A

differential demand for audit quality may arise when an investor has an opportunity set

of potential monitors of manager behavior. These include, but are not limited to,

knowledge of general economic conditions, related industry performance, internal

auditing and accounting reports, personal conversations with management, and the

external audit. Additionally, as the investor's participation in the management of the

company increases, the ability to use internal information to monitor the actions of

management also increases, implying less need for, and value of, an external audit.

Since the external audit represents just one of the numerous monitors that

investors can use to monitor management, the demand for accuracy is affected by the

reliance placed on the audited financial statements." Wilson [1983] shows that in

situations where investors have limited access to reliable monitors of management

behavior, the demand for a more accurate audited financial statement, or more

diagnostic information signal, increases. Conversely, investors who have access to

other monitors that accurately indicate the true state of nature are more amenable to



'4This revelation is a proxy for information that can be revealed through third
party actions such as litigation, press releases of analysts, court filings, etc. Consistent
with extant empirical markets research, the potential future penalties from auditor
misreporting are not discounted.

"Reliance indicates to what degree investors utilize the audited financial
statements in making their contracting and investing decisions.









receiving less diagnostic information from the audit. Note that in this situation the

investor potentially has better information about the inherent risk present and may be

willing to accept higher levels of audit risk, which is reflected in an unwillingness to

pay higher audit fees. If the audit risk being provided by high quality audits is lower

than desired, buyers may not be willing to pay for it. Therefore, the demand for audit

accuracy can be affected by the buyer's reliance on the attest function.

This discussion suggest that a heterogenous demand for auditing services may

exist. The preference for audit accuracy is affected by the degree of reliance on the

audit as an indicator of the true state of nature. To incorporate the heterogenous

demand for audit accuracy a weighting factor (8) will be utilized which can be either

high (SH) or low (6L), respectively [8 e 8L, 8H]. At 8L (6H) this factor indicates a low

(high) level of reliance on audit accuracy and will therefore reduce (increase) the

influence of costs incurred by the investor from inaccurate auditor reporting.

Based on the aforementioned assumptions and variable descriptions, the

remainder of this section will derive the buyer's expected valuation of audit services.

It is assumed initially that audit quality is known to the buyer. Then uncertainty about

audit quality will be incorporated into the analysis.

The buyer's expected value (V) from contracting for an audit of type q, is

affected by the true state of nature (S), the auditor's reporting accuracy (a), the payoff

received based on the audit report (X), the level of reliance on audit accuracy (8), and

the audit fee (Fi). Using E as the expectations operator this is expressed formally as:







29

E(V qi) = P(S=SH)P(R=S q=qi)Xn + P(S=SH)P(RSS Iq=q,)X, + (1)
P(S=SL)P(R=S I q=q)XL + P(S=SL)P(RS q=qi)(X, -8L) F116

Applying the notation previously described produces:

E(V|qi) = xaiXH + (1 ai)XL + (1- )iXL + (1 X)(1 a )(XH- 8X) Fi (2)

Equation 2 indicates that when audits perfectly reveal the true state of nature

(so a, =1) then auditor misreporting cost are eliminated. The buyer will receive the

exact compensation they were entitled to, or X, when the true state of nature was high

and XL when the true state of nature was low.17 However, if audits are unable to

perfectly report the true state of nature, or a, < 1, then auditor reporting errors occur.

Auditors could misreport the state of nature to be low (high) when the true state of

nature is high (low) which reduces buyers' expected value.

To analyze the buyers' valuation of differential audit quality, the incremental

effect of audit quality (qi) on the expected valuation from auditor services follows

from equation 2 by applying substitution and simple algebra:

E(VI qH) E(VI qL) = (a aL)[(20 1)(XH XL) + 8(1 0)]-(FH FL) (3)

Equation (3) can be simplified by determining the relationship that holds when

the buyer is indifferent regarding the provision of audit quality. When

E(V I qH) E(V I q ) = 0, equation 3 becomes:



'6To facilitate the operationalization of this model into the experimental setting,
the reliance factor will only be applied to the costs generated from misreporting a low
state as high. Incorporation of the reliance factor to costs generated from misreporting
a high state as low does not qualitatively affect the predictions of the model.

"When ai = 1, the second and fourth terms of equation 2 are eliminated.
Therefore, when 0 = 1, or the state of nature is high, then E(Vlq,) = XH. Likewise,
when 0 = 0, or the state of nature is low, then E(V|qi) = X,.


j







30

FH FL = (aH aL)[(2d 1)(XH XL) + 68(1 )] (4)

Analysis of equation 4 indicates that the auditor's incremental ability to

accurately report on the true state of nature, as measured by (ac aL), affects buyer

willingness to pay higher fees. If buyers can assess audit quality, analysis of equation

4 indicates that the incremental willingness to pay for high quality audit services is

increasing in aH, X, and 8, ceteris paribus.18

The interactive between the presence of auditor misreporting cost (X) and the

buyers reliance on audits (8) indicates that both are necessary, but not sufficient,

conditions for a demand for high quality auditing services to exist. This indicates that

the demand for high quality auditing services is influenced by a combination of the

presence of auditor misreporting cost and the reliance on the audit by the buyer in

making contracting decisions. This interaction will be subsequently referred to as the

buyer's reliance cost.

This analysis shows that the buyer's valuation of audit services is affected by

several independent and interacting factors. However, as shown by equation 4, before


"This requires the assumption that aH > aL which follows from the additional
investment made in audit quality as detailed earlier. Given that the second term of
equation 4 can produce a positive or negative value depending on the magnitude of 4,
note that buyers willingness to pay the incremental fees required for the provision of
high quality audits is increasing in settings where

(-2( + 1)(X, X2)
[1 -4]


This indicates that a buyer's desire for audit accuracy is affected by their level of
reliance on auditor reporting accuracy, the relative payoffs for a high and low state,
and their prior information about the true state of nature.









any of these factors have an affect on the valuation, the buyer must first be able to

assess the differential audit quality present in the marketplace.

The Auditor's Valuation of Audit Services

In the market, audit quality is not perfectly determinable by buyers. With

incomplete information regarding the true state of nature and information asymmetry

about the audit quality level it is difficult for the buyer to assess auditor quality based

on prices as auditors who provide low quality will be able to perfectly emulate the

behavior of auditors who provide high quality. In this setting Shapiro [1983] shows

that low quality services will be priced at cost, or FL = CL, and high quality services

will be priced at

FH > CH + v(CH FL)/4 (5)

Where v represents the time period for the seller to develop a reputation for providing

high quality services and p represents the time period that high quality services will be

provided in subsequent periods.

As an adaption of Shapiro's model, a review process, operationalizing the peer

review process, is incorporated into the model. Auditors will receive a report based

on the quality of audits provided during the periods) under review. This report (r)

will indicate if the audits were of high (h) or low (1) quality (h, 1 e r). The first

review period will include the initial period after the auditor indicates they desire to

enroll in the review process. Subsequent reviews will be performed on a regular

interval if requested by the auditor. To receive a high peer review report all audits

conducted during the periods) under review must be of the high quality type (qH).









Auditors can provide the peer review report information to buyers in subsequent

negotiations. Only truthful revelation of the peer review report received by auditors is

possible.

Since the initial peer review report is received at the end of the first period

under peer review, the auditor has a potential loss from the provision of high quality

services of rl(FL CH ),19 where rI represents the number of audits provided in the

initial period under review. Therefore, the fee required to recover this loss and

motivate the continued provision of high quality services by auditors is:

Xc(FH CH) + qr( FL CH) 0 (6)

or

FH 2 CH + rI(CH FL)/Xa (7)

where x represents the number of future periods high quality services will be provided

at compensatory rates and a represents the number of audits sold during this period.

Note that this model assumes that the time period to detect "shirking" on the

peer review report is the same as the time period required for the provision of high

quality audits to obtain a high peer review report. Given that the time required to

establish a high quality peer review report is potentially shorter than the time period

for the buyer to detect shirking, it is important to note that this pricing does not

necessarily assure that the auditor's optimal strategy is to maintain the provision of

high quality services. If auditors choose to lower their quality in future periods,



"This potential loss is created by the buyers incentives to accept the lowest quality
offer when quality differentiating information is not available before the purchase
decision.









profits could be earned until this behavior is detected by buyers. Noting this, the

timing of subsequent peer reviews could affect the "honesty" of auditors since it is

possible for them to provide low quality services for the periods between review

without the buyer's knowledge.

If auditors adopt the strategy of initially providing high quality services to

obtain a positive peer review report (r=h) and then providing less costly low quality

services until detected by the peer review process, the profits earned are:

7tp(FH CL) + r(FL CH) (8)

where 7 represents the number of periods between reviews and p represents the

number of audits sold each period the auditor is reducing quality. Therefore, for

consistent high quality services to be provided the profits from the provision of

consistent quality must at least equal the profits from opportunistically reducing

quality, or comparing equations 6 and 8:

Xo(FH CH) + 1q(CH FL) > rp(FH CL) + n(CH FL) (9)

It is assumed that a=p, or the number of clients retained by the auditor should

not be affected by the auditors strategy in the periods between review reports as the

buyer's information has not been updated. If the auditor provides consistent high

quality services the subsequent review report will confirm this allowing the auditor to

continue to present the high (h) peer review report. If the auditor shirks and provides

low quality services, buyers will become aware of this. Therefore, the number of

periods a auditor can benefit from the continual production of high quality services is









greater than from opportunistically reducing quality in the periods between peer

reviews, or X > n.

Based on these assumptions the required fee for the auditors continued

provision of high quality services is determined by solving equation 9 for FH, or

FH (XCH- tCL)/(X-7) (10)

Analysis of equation (10) provides a significant insight. As n, or the time period

between reviews increases, the auditor's fee required for the provision of high quality

increases.20

Analysis of Player Strategies

Single period market

As a benchmark for decision making it is useful to consider the incentives that

the players would face in a single period market. Note that this precludes utilization

of the peer review process as the peer review report is not available until the

beginning of the second market period. With incomplete information regarding the

true state of nature and information asymmetry about the audit quality level it is

difficult for the buyer to assess auditor quality based on prices. Auditors who provide

low quality will be able to perfectly emulate the behavior of auditors who provide

high quality. As a result of the asymmetric information regarding auditor quality, the




20The incremental impact of n on FH is shown by application of the quotient rule
to be
aFH X(CH-CL)
9 (X-)>0
aTc (z 7C)2







35

expected value the buyer receives from audit services is only affected by fees paid, as

shown in equation 4. Therefore, the buyer's optimal response is to accept the lowest

fee offered. If auditors reason that buyers will only accept the lowest fee, the auditor's

best response is to provide less costly low quality services. Therefore, only low

quality audits will be provided in the single period setting. The auditors expected

profits will be

F-CL (11)

To determine the prices which will derive in the single period market requires

a further model specification of the nature of market supply and demand for auditor

services. An advantage of utilizing the laboratory markets methodology is that the

structure of market supply and demand can be specified to best address the problem of

research interest. This is accomplished by allocating market surplus from trade to

buyers and auditors in the manner best suited to experimental hypothesis testing. The

control over quantities which can be purchased (sold) can be used to generate excess

demand (supply), which, at the extreme, allows the buyer or auditors to capture the

entire surplus available in a laboratory market. Auditors will capture the market

surplus in markets with excess demand, and buyers will capture the market surplus in

markets with excess supply (Smith [1982]).

The structure of supply and demand used in this study is one of excess supply.

As noted earlier, buyers can only purchase one audit but auditors can sell to as many

buyers as they can contract with. The produces as excess supply of audit services.

This demand and supply structure is illustrated in Figure 3-1. In an environment of







36

excess supply, auditors will be driven to cost based pricing. Setting equation (11) to

zero determines the price for low quality auditing services of

FL = CL (12)

Multiple period markets

The previous pricing was developed with an assumption of a single-period

world. However, when there are multiple periods, as required to operationalize the

peer review process, there can be numerous alternative solutions. Note that the buyer

could be better off (and the seller no worse off) with an outcome involving high

quality and a commensurate market price.2' Yet, the single period strategy for the

auditor is to provide low quality auditing services. In this case, the availability of

additional solutions requires a refinement of the predicted outcome. One useful and

standard refinement is the notion of Pareto dominance. Here, players will tend to

cooperate such that everyone receives the highest payoffs, if such cooperation is

possible. A possible aid to this coordination problem is the use of the peer review

signal. The accounting profession's adoption of the peer review process has the

potential benefit of allowing buyers to differentiate between the quality of respective

auditors. This could increase buyer willingness to pay for high quality services, in



21The parameters used in the experiment, as detailed in Table 1, produce the
following expected values. In settings without reliance costs, buyers expected value
from the purchase of high (low) quality auditing services is 426 cents (450 cents)
inducing a preference for low quality. In settings with reliance costs the buyers'
expected value from the purchase of high (low) quality auditing services is 372 cents
(180 cents). The cost of providing high (low) quality services is 200 cents (100
cents). Therefore, in settings with reliance costs, where high quality is preferred, the
buyer can capture an additional 92 cents by purchasing high quality services (372-200
+ 180-100), assuming audits are priced at cost, without reducing the auditors profit.









settings where high quality is demanded, while motivating the auditor to consistently

provide high quality services. If effective, peer review potentially allows a multi-

period solution that is Pareto superior to the single period solution.22

The previous discussion indicates that players optimal contracting strategies

regarding the quality of auditing services are affected by the buyer's reliance costs, the

availability of peer review, and the timing of peer review. The following discussion

develops predictions for strategies of buyers and auditors in each market setting and

provides comparative hypotheses for testing.

Markets without peer review

The analysis for markets without peer review is straightforward. Buyer

willingness to pay for increased reporting accuracy is shown in equation 4 which is

reproduced here:

FH FL = (aH aL)[(24 1)(XH XL) + 86(1 0)] (4)

As shown earlier, if buyers cannot assess the differential reporting accuracy provided

by high quality services they will treat the purchase of audits as a lottery, since



22An alternative solution is that produced by backward induction which allows the
extension of the single-period solution to multiperiod environments with a known
endpoint (e.g., Selten [1978]). Milgrom and Roberts [1982] show two factors that can
discourage backward induction. First, information asymmetry must be present within
the market setting, or players cannot have complete information about others players
current decisions and motivations. Second, the setting must include repeated
interactions with some possibility of players observing either the other player's past
behaviors or the outcomes from other players past behaviors. Information asymmetry
gives the players reason to forecast future actions based on past behaviors which can
motivate players attempts to develop reputations for honoring contractual relationships
(Kreps et al. [1982]). Therefore, while a single-period solution can be obtained in a
multi-period setting, alternative solutions can exist.









(H = XL (or aH aL = 0) by implication, irrespective of their reliance on auditor

reporting accuracy (86). In this setting, the buyer's incremental willingness to pay for

higher quality auditing services reduces to

FH F = 0 (13)

If auditors reason that buyers will only accept the lowest fee, the auditor's best

response is to provide less costly, low quality services. Therefore, only low quality

audits will be provided in markets without peer review irrespective of the buyer's

reliance on auditor reporting accuracy. As previously shown, competition will drive

prices for low quality audit services to CL.

Markets with peer review, no reliance

In markets with peer review, buyers willingness to pay for audit quality must

be addressed. In markets were buyer's have no reliance costs, or 8X = 0, equation 4

reduces to

FH FL = (aH aL)[(2< 1)(XH XL)] (14)

In this setting buyer's do not rely on auditor reporting accuracy. Analysis of

equation 14 indicates that increased auditor reporting accuracy, or a( > aL, has a

differential impact on the buyer's willingness to pay increased fees based on the

buyer's ex ante information about the state of nature(4). To establish experimental

settings where buyers would not prefer higher levels of audit accuracy, in the absence

of reliance costs, 0 was set equal to .4. In this setting buyers' will not alter their

decisions based on the peer review information. Buyers' incremental willingness to

pay for auditor reporting accuracy becomes F, = FL.







39

Auditors willingness to provide more accurate reporting services is reflected in

equation 10 which is repeated here:

FH > (XCH- 7CL)/(x-7i) (10)

Based on the previous assumption that x > 7 and additional assumption that x, n > 0,

the minimum valuation of the right hand side of equation 10 is CH.23 Given that CH >

CL by definition, FH must be greater than FL for the auditor to provide high quality

services. However, the buyer's valuation of audit quality is FH = FL. This lack of

congruence in the valuation of audit services, in markets with peer review where

buyer's have no reliance costs, will lead to auditor provision of less costly, low quality

services with prices driven to FL = CL by competition.

Markets with peer review and buyer reliance

In settings with peer review, where buyers rely on auditor reporting accuracy,

the relationship for the provision of high quality services can be shown by equating

the price at which buyers and auditors are willing to purchase/provide high quality

services. From equations 10 and 4 this relationship can be written as

(XC,- CL)/(X-7) = (aH aL)[(2 1)(XH XL) + 6X(l )] + FL (15)

In this setting the availability of peer review may provide sufficient information to

buyers to allow their valuation of high quality audit services, as represented by the

right hand side of equation (15), to equal or exceed the auditors required fee for high

quality services which may motivate them to provide high quality audit services.


2Equation 10 reduces to FH = CH when x = 1 and 7r = 0.









This analysis indicates that the availability of peer review will only influence

the quality/price combination of audit services in settings where buyers rely on auditor

reporting accuracy. Additionally, the timing of the peer review process influences the

relationship shown in equation 15. Earlier, it was shown that as the periods between

peer reviews increase, or 7 gets larger, the required price for the auditor to provide

high quality services increases. Therefore, the buyer's valuation of reporting accuracy

must also increase to maintain the relationship. Therefore, the interactive effect of the

availability of peer review and the buyers' demand for reporting accuracy could be

affected by the timing of the peer review process.

Hypotheses

The previous discussion indicates that players optimal strategies regarding the

quality of auditing services to contract for are affected by the buyer's reliance

costs (6?), the availability of peer review, and the timing of peer review (n). Specific

hypotheses will be presented to examine the quality, and respective pricing, of audit

services provided in each market. Additional hypotheses will deal with the market's

economic efficiency. Economic efficiency is a measure often considered in

experimental markets and details the amount of resources generated by trade. This

will be examined by measuring the total surplus generated in each setting, referred to

as market efficiency. Market efficiency is optimized in each setting as decisions near

the model's price predictions. Additionally, the allocation of the generated surplus

between the buyers and auditors, referred to as allocative efficiency, will be examined.

As noted earlier the structure of supply and demand within the markets indicates that







41

the buyers will capture the surplus generated by the market as the auditors are forced

to cost-based pricing by excess competition. As the price paid for audit services nears

the model's prediction, allocative efficiency will increase.

From the previous analysis the following predictions are made.

H, The proportion of high quality audit services provided will be increasing
in the buyers level of reliance (6) only in markets where peer review is
present and this interaction will be more likely as peer review timing
(n) decreases.

H2 The prices of the audit services provided will be increasing in the
buyers level of reliance (8) only in markets where peer review is
present and this interaction will be more likely as peer review timing
(it) decreases.

H3 Allocative efficiency will be increasing in the buyers level of reliance
(6) only in markets where peer review is present and this relationship
will be more likely as peer review timing (n) decreases.

H4 Market efficiency will be increasing in the buyers level of reliance (5)
only in markets where peer review is present and this relationship will
be more likely as peer review timing (7t) decreases.

A final hypothesis will examine the influence of peer review timing and

reliance costs on auditor decisions regarding the consistent provision of high quality

auditing services in markets with peer review. Remember that an auditor, after

receiving a high peer review report based on past audit quality provision, can

subsequently elect to provide low quality audit services without detection until the

next peer review. Given that auditors incur no penalty for this strategy, other than a

potential loss of future business, they will not be concerned with the buyer's level of

reliance. Additionally, the model indicates that the time period between peer reviews

and the required price for the continued provision of high quality services are









positively correlated. Therefore, peer review timing may affect an auditor's

willingness to provide consistent high quality. This leads to the following prediction:

H5 The proportion of auditors "shirking" by providing low quality audits in
the market periods between peer reviews, will decrease as peer review
timing becomes more frequent.

Dependent variables

Information was captured on the following dependent variables to test the

proposed hypotheses:

H1 Audit quality: the number of high quality audits purchased as a ratio of the
total audits purchased for each market period.

H2 Market price: the mean of the prices paid for audits for each market period.

H3 Allocative efficiency: the difference between the market surplus captured by
the buyers and sellers in each market. In markets with optimal decision
making the buyer should capture the surplus so allocative efficiency is
increasing in buyers surplus.

H4 Market efficiency: the total surplus generated in each market setting.

H5 Auditor shirking: the percentage of low quality audits provided when the
auditor was indicating their quality to be high with a high peer review report.















Price






Equilibrium = >


Demand



Supply


Supply







Demand


Quantity


Figure 3-1: Forced Excess Supply













CHAPTER 4
RESEARCH DESIGN AND EXPERIMENTAL METHODS

Introduction

The present chapter will convert the conceptual model detailed previously into

a series of laboratory market experiments (see Table 4-1). As these markets are

created it will be important to ensure that the internal validity of the experiment is not

compromised. Internal validity first requires the development of a successful

laboratory market. This requires adherence to the theory of induced value (Smith

[1976] and Plott [1979]), while controlling for the precepts documented by Smith

[1982]. To comply with these precepts, actual monetary payoffs will be used to

ensure dominance and nonsatiation. The experimental parameters were set such that

experiment participants could expect to earn $25 for their participation. Privacy was

maintained by paying each subject in cash privately after the experimental session had

ended. Subjects were instructed on several occasions verbally, and in writing, to not

discuss the experimental session with anyone until the date given that all experimental

sessions would be complete. Responses to the post-experimental questionnaire

indicate that internal validity was sustained (see appendix B).

Next, it is important to control for demand effects and subject interaction in

the experimental setting. The goal is to ensure that other aspects of the subjects'

utility functions, such as fatigue, level of interest, and stress, do not materially change









during the experiment. This was accomplished by limiting the experiment to a

reasonable period of time, approximately 2 1/2 hours, and controlling for experimental

demand created by instructions, definitions, conversations, and feedback. Additionally,

subjects were randomly assigned to the player types to control for possible subject

variation.

The experimental instructions used generic terms such as buyers and sellers as

opposed to auditors and clients to control for priors. Additionally, no mention was

made that this experiment represented the auditing market. The output of the audit

process was converted to experimental commodities with nondescriptive definitions.

The goal was to control for potentially confounding effects from real world

institutions. It was stressed that this experiment was not an examination of classroom

knowledge, and that only the information provided in the experiment should be used to

make decisions.

Laboratory Markets

In the experimental market detailed below, the presence or absence of peer

review was contrasted with the presence or absence of reliance costs for buyers of

audit services. Twelve markets were conducted with a total of four in the no reliance

cost setting, and eight in the reliance cost setting to allow for an examination of two

separate peer review institutions.' This includes a three year review that






'The use of 4 markets in the no reliance cost setting (or 8X=0) was driven by the
models prediction that peer review would not impact market results in this setting.









operationalizes the existing structure in the profession and a one year review that

provides information regarding the effects of a more timely review process

(Figure 4-1).

This research adopted an information structure similar to Kachelmeier [1991]

and represents a compromise of an assumption of game theory, namely that all players

have perfect information about other players' valuation, cost, and utility functions, and

the precepts of induced value that require all such information to remain private in the

experimental setting. For this research, the composition of payoffs was publicly

communicated to all players in the laboratory market instructions. However, these

instructions stopped short of communicating actual valuation and cost amounts.

Buyers were informed that auditors (described as sellers in the experimental setting)

incurred one of two possible investigation costs where the higher (lower) cost

investigation had a 90 percent (50 percent) chance of determining the true state of

nature. Similarly, auditors were informed that buyers received greater compensation

for high reports and in reliance cost settings, that buyers could incur additional costs

from auditor misreporting. By publicly revealing relative payoff and cost information,

while maintaining the privacy of specific values, an environment advantageous to

game theoretic strategies is maintained while avoiding the potential confounding

effects of violating the experimental precepts.2





2Kachelmeier and Shehata [1992] found that when cost and redemption values
were public information that these experimental markets were significantly less
competitive than private information markets.









The reliance cost settings were operationalized with an ex post charge to

buyers at the end of the game for all periods when the auditors report revealed a high

state of nature when the actual state of nature was low. To improve the information

environment of Kachelmeier [1991], where buyers had no information regarding the

total amount of additional charges incurred until the end of the game, this experiment

provided the buyers with the expected value they were earning based on the quality of

audits they might have received (high or low). Therefore, the buyers could see the

expected impact of a auditors provision of high and low quality audits as the game

progressed while not allowing buyers to directly infer an auditor's quality during the

execution of the game.

Subjects

A total of 144 subjects for the laboratory markets were obtained from student

groups, mainly graduate students from the College of Business Administration. Each

subject participated in one the 12 markets (eight subjects to each market).3 Subjects

were informed they were volunteering to participate in an experimental session lasting

approximately three and one-half hours. While it would be preferable to use "real"

owners, business managers and auditors, the constraints imposed due to the monetary

and time considerations render this infeasible. Additionally, Plott [1982] notes that as






3Twelve subjects were enlisted for each experimental setting to ensure that the
required eight subjects were present. If more than eight subjects arrived, eight
subjects were randomly selected from the subjects present and the remaining subjects
were paid a small fee and dismissed.









long as real people pursue real incentives in a laboratory market, the choice of

subjects should not be critical.

Procedure

After arriving at the experimental location each player received the same set of

written instructions. An instructional period was then completed where the game

instructions were read aloud to players and a pre-experimental questionnaire was

completed and discussed to ensure the market structure was understood. After random

assignment to computer workstations,4 which determined player classification as a

buyer or auditor, a training phase (no cash earned) was then conducted that consisted

of three experimental market periods that used the same computerized trading

mechanism as the experimental phase. The purpose of this training phase was to

allow the subjects to learn about the computer mechanism, not the behavior of others.

For this reason, the experimenter gave the subjects the values to enter. Subjects were

informed that these values were to illustrate the options available to them and were not

to be considered as recommended behavior.

After the training phase was completed, the experimental market periods were

conducted. Each market consisted of two sub-markets where players first participated

in a non-peer reviewed market for 15 periods and then a peer reviewed market5 for the




4Physical barriers were present in the computer laboratory to prevent participants
from making visual contact with other players. Participants were instructed that verbal
contact was prohibited and this was monitored by the administrator.

sIn the experimental setting the peer review report was referred to as an "analysis"
report.









remaining periods of the game (the average length of this second market was 15

periods). No information about the second market was given before the conclusion of

the first. The parameter values used are given in Table 4-2. The values were set to

induce an indifference towards audit quality in the no reliance cost settings and to give

the buyer a preference for high quality audits in the reliance cost settings. The

sequence of decisions is detailed in Figure 4-2 and Figure 4-3 for the non-peer

reviewed and peer reviewed markets, respectively.

Before each peer reviewed market began, an identical set of additional

instructions was read aloud by the experiment administrator while all participants

reviewed their own copy of the instructions which detailed the peer review process.

Each participant was polled privately by the administrator to ensure their

understanding of the additional information before the market was restarted. At the

conclusion of this market, participants were asked to complete a post-experimental

questionnaire. They were paid their cash earnings (privately) for all trading periods,

which ranged from $7 to $42. Analysis of the post-experimental questionnaire

responses indicated that participants understood the instructions and how their

decisions would affect their earnings as indicated in appendix B.

In order to mitigate end game effects, players were informed at the end of

period 27 that at the end of each subsequent period, the computer would roll a "die"

with values 1 through 6. If the roll was a six, the game would end, otherwise the







50

game would continue on to the next period. This information was displayed on each

player's computer at the end of each affected period.6

The markets utilized a posted offer institution. Auditors submitted offers that

were visible to all buyers.7 Buyers chose a specific auditor. Note that this

institutional setting encourages the prices to converge from "above", or prices will

decline towards the cost of production in the competitive markets, since initially the

auditors have the informational advantage (Smith [1982]). Given that auditor profits

will be driven to zero by competition, auditors were compensated through endowments

for each market period. Buyers received an initial endowment of sufficient funds to

ensure they did not incur a net loss in the initial periods) when they were likely to

pay too much for audit services. Endowments also allowed the payment structure to

be manipulated so that all experimental participants had the potential to receive the

same expected compensation.










6Thus there was a on-sixth chance of the market ending on any day after day 27.
This essentially produced a maximum discount factor of 20 percent (p/(l-p)) for the
subjects. This creates an expected number of experimental days of 33 per market.
Therefore, without the players knowledge, the computer always rolled a six at the end
of the 33rd "day" if the game reached that point.

7As is common practice in experimental markets settings, to speed the convergence
to cost based prices, successful offers were displayed to all players at the end of each
period. Auditors did not directly observe other auditor's offers, but they were aware of
the level at which contracts were being exchanged.









Table 4-1
Operationalization of Conceptual Model to Laboratory


Investor/Manager model Laboratory Markets


State of nature can be high or low State of nature has a high or low value

Auditors sell their audit services to Sellers sell investigations about state to
buyers. Buyers.

Buyer compensation (X,, X2) is Buyers compensation from the
contingent on the auditor's report, experimenter is based on whether the
(R = H, L) with X, > X2. state's value is reported as high or low.
Payment will be greater for the high
valued state.

Auditors pay the cost of providing a Sellers pay either a high or low cost to
high or low quality audit. The quality investigate and report on the state. The
provided is positively correlated with ability to detect the value of the state is
the ability to detect the true state of positively correlated with this cost.
nature.

A probabilistic signal regarding the true At the end of applicable markets buyers
state of nature is publicly reported for must pay the experimenter a penalty for
specific buyers, who have engaged all periods in which the investigation
auditors. A penalty is assessed for all report was high, but the true state was
periods where (R=HI S=SL). low.

Auditors can undergo peer review which Seller's can undergo an investigation of
will evaluate their audit quality level, their production process. A report will
Peer review reports are issued at the end be issued after the initial year, and
of the review period, every year (third year) thereafter,
indicating the investigations rendered
were high or low

A high/low peer review report will be A high report can only be given if the
given if the cost of the audit produced is costs of the investigations produced are
high/low for the period under review, high for the entire period under review.
Otherwise a low report will be issued.
Buyer players will be instructed that a
high/low report indicates that the
investigations produced were high/low.









Table 4-2
Glossary of Variables


Variable Description Value'

CH, CL E c Auditor's cost level, high/low 200 or 100

SF F Fee paid for high and low quality
Fn, F1 E F
services

qn, qL E q Audit's quality level, high or low
h, 1 e r Peer review report on audit quality
H, L e R Auditor's report

H, S, ES Probabilistic state of nature with .40 or .60
values of high ( ) or low (1-+)

Compensation paid by
XH, XL e X experimenter to buyer based on 300 or 600
auditor's report (H, L e R).
The probability the auditor is able
a,, aL E ai to detect the true state of nature .90 or .50
given his/her quality level
Misreporting costs incurred by 900
buyer when (R=HI S=SL)
The probability that a buyer relies
8L, 8H E 6 on the audit as a monitor of the 10
0%. or 100%
investor/manager relationship
8, Reliance costs 0 or 900


'Values utilized in this experiment were framed in "cents". Variables for which no
value or probability is listed were determined in the negotiations between market
participants.









Table 4-2 (continued)
Glossary of Variables


Variable Description Value

V Buyers expected value from
contracting for auditor services

rT number of buyers engaged in
initial period auditor is in the peer
review process

X number of periods high quality
services will be provided after
initial engagement
C number of buyers engaged during
X period.
p number of buyers engaged in
periods the auditor is
opportunistically reducing quality.
7 number of periods between peer 1 or 3
reviews


F-












Not Not
Available Availae A e Available
Available" Available Available


No Reliance costs**
Reliance costs


A,


A2


C, C2 D, D2


*The three year and one year peer reviews denotes the review timing manipulation.

"Unavailable and available denotes the presence of peer review or the peer review
manipulation. Cells A, D, represent first 15 market periods and cells A2 D2
represents market periods 16 +.

*"No reliance costs and reliance costs denotes the reliance manipulation. There were
2 replications of cells A and B, and 4 replications of cells C and D for a total of 12
markets.



Figure 4-1
Experimental Settings


B


B2 I


3- Year Peer Review*


I Year Peer Review


I










Auditors privately select quality level

4-


Auditors submit offer to buyers

4-


Buyers accept or reject auditors' offers




Auditor observes a state of nature and issues
report to buyer


End of Game
(Reliance settings only)


True state of nature revealed to buyers,
penalties assessed


Figure 4-2
Non-peer Reviewed Settings










Auditors elect to initially join or remain in peer review*


Auditors privately select quality level



Auditors submit an offer, with/without peer
review report, to buyers



Buyers accept or reject auditors' offers



Auditor observes a state of nature and issues
report to buyer


Payoffs made to buyers based on auditor's report


Peer review report received


End of Game
(Reliance settings only)


True state of nature revealed to buyers,
penalties assessed

'After initially joining peer review, the decision to remain under review occurs in
subsequent periods based on the term of the review process examined, one or three
periods.


Figure 4-3
Peer Reviewed Settings













CHAPTER 5
RESEARCH RESULTS

Introduction

The no reliance/non-peer reviewed settings (A, and B,) were used to provide a

benchmark for determination of the interactive effects for the factors of reliance, peer

review, and peer review timing. The purpose of this study is to compare the effects of

these factors on player's behaviors, not necessarily to test whether a particular solution

holds in a selected experimental cell. Thus, the hypotheses will be analyzed as

comparisons across markets rather than as test of point predictions.'

The markets were designed to examine market rather than individual participants

behavior. Therefore, the variables tested represent the average responses for each

market.

Hypotheses 1-4 were analyzed by examining each of the respective dependent

variables using a repeated-measures analysis on peer review (not available, available),

reliance costs (no reliance, reliance), and peer review timing (3 year, 1 year), where

peer review is a within-markets factor and reliance costs and peer review timing are

between markets factors. Hypothesis 5 was analyzed using a between markets design


'Potential learning effects are mitigated by only using the last 8 periods for each
market in the data analysis. Other cutoff periods produced results that are
qualitatively the same. Previous experimental research has found that experience
effects are mainly present during the initial stages of a market as indicated by a more
rapid convergence to predicted vales (i.e., Schatzberg and Sevcik [1993]).

57







58

with the factors of reliance and peer review timing, and the respective interaction term,

for markets where peer review was available (markets A2, B2, C2, D,). The results are

organized by dependent variable and will incorporate a discussion of the specific

hypotheses affected.2

Results

Audit Quality

The results for audit quality are reported in Figure 5-1 and Panel A of Table

5-1. In support of H,, the audit quality provided was significantly affected by the

interaction of the availability of peer review (PR), the timing of peer review (PRT),

and the presence of reliance costs (R) (F(,~)=5.62, p=.0226).3 To isolate the significant

effects producing this interaction the analysis was extended by examining the simple

interaction of reliance and peer review for the 3-year and 1-year peer review markets,

respectively. The 3-year peer review markets, as shown in Panel B of Table 5-1, did

not reveal a significant simple interaction (F(,4) = .01, p=.471).4 The simple main

effects of peer review (F(1,4)=.29, p=.310) and reliance (F(,4)=1.82, p=.125) were also




2The following analysis adopts the "analysis of simple effects" approach to
analyzing three-way interactions as documented by Keppel [1991, p.445-452]. Briefly
this analysis begins by looking at the three-way interaction and proceeds to the
appropriate analysis of simple interaction effects and/or main interaction effects.

3The p-values indicated in the results section are for a directional (one-tailed) test
of significance.

4As recommended by Keppel [1991] p. 448, when heterogeneity of within-groups
variance was present in the simple interaction analyses, separate error terms based on
the within-group mean squares were calculated for each of the interaction effects at
different levels of peer review timing.









insignificant. This indicates that audit quality was not significantly affected by the

availability of a 3-year peer review in markets with or without reliance cost.

In the 1-year peer reviewed markets, the simple reliance (R) x peer review

(PR) interaction (Table 5-1, Panel C) was significant (F(,,4)=6.59, p=.031).

Examination of the simple-simple effect of peer review in 1-year markets with reliance

costs produced a significant increase in audit quality. The percentage of high quality

audits provided was 7.81% (62.50%) in markets without (with) peer review

(F(,3)=13.66, p=.017). The availability of peer review did not produce a significant

increase in audit quality in markets without reliance costs. The percentage of high

quality audits provided was 15.63% (12.50%) in the markets without (with) peer

review (F(,7,)=.25, p=.352).

These results provide strong support for H,. The proportion of high quality

audit services provided was increasing in the buyers level of reliance in markets where

peer review was present and this interaction was only significant in markets with the

1-year peer review. In markets where buyers demanded high quality services, the 1-

year peer review process allowed buyers to isolate high quality auditors which

increased the provision of audit quality within those markets.

Market prices

The results for market prices are reported in Figure 5-2 and Panel A of Table

5-2. H2 is not supported in that market prices were not significantly affected by the

interaction of the availability of peer review (PR), the timing of peer review (PRT),

and the presence of reliance costs (R) (F(,8)=.30, p=.299). Additional analysis of









market price behavior isolated the following significant effects. First, the the main

effect of reliance was significant (F(8,)=4.40, p=.035). Market prices were uniformly

higher in markets with reliance costs over markets with no reliance costs. This

indicates that the presence of reliance lead to higher prices irrespective of the presence

or timing of peer review. Review of the market cell means (Table 5-2) indicates that

buyers were potentially mislead by auditor pricing strategies in the reliance settings.

Prices increased from an average of 128.67 [(128.95 + 128.39)/2] in markets with no

reliance costs to 167.61 [(177.30+157.92)/2] in markets with reliance costs in settings

without peer review. This provides empirical evidence to support Chan and Leland

[1982]. They developed a theory which shows that in settings where prices are

costless to observe but quality is costly to observe that sellers are able to extract

surplus from uninformed buyers surplus by reducing quality rather than prices. In this

study, market prices were not significantly affected by the interactive effects of

reliance and peer review due to the buyers' willingness to pay high prices in settings

where audit quality was not determinable.5

Second, the interaction between peer review and peer review timing produced

significant results (F(,8) =3.36, p=.050). Analysis of the simple main effect of peer

review timing produced significant results for the markets with the 3-year peer review

(F(o5) = 5.11, p=.0367) indicating a reduction in average prices in markets with peer

review. When combined with the earlier result that quality did not increase in markets

with the 3-year peer review, this indicates that the 3-year peer review did provide


'See Appendix C where players strategies are presented.









sufficient information for buyers to determine that low quality audits were being

provided which led to reduced prices. Average prices did increase, although not

significantly (F(, 5)=1.39, p=.145) in markets where the 1-year peer review was present

for the markets with reliance costs. These results indicate that peer review provides

quality discriminating information to buyers that allows them to make more informed

purchase decisions. Irrespective of the quality provided, prices moved toward the cost

of providing the quality purchased, high or low respectively.

Allocative efficiency

The results for allocative efficiency are reported in Figure 5-3 and Panel A of

Table 5-3. In support of H3, the allocative efficiency attained was significantly

affected by the interaction of the availability of peer review (PR), the timing of peer

review (PRT), and the presence of reliance costs (R) (F(18)=5.34, p=.025). To isolate

the significant effects producing this interaction the analysis was extended by

examining the simple interaction of reliance and peer review for the 3-year and 1-year

peer review markets, respectively. The 3-year peer review markets, as shown in Panel

B of Table 5-3, did not reveal a significant simple interaction (F(,4) = .01, p=.472).

This indicates that the buyers' ability to capture market surplus was not significantly

affected by the availability of a 3-year peer review in markets with or without reliance

cost.

In the 1-year peer reviewed markets, the simple reliance (R) x peer review

(PR) interaction (Table 5-3, Panel C) was significant (F(,4)=7.70, p=.025).

Examination of the simple-simple effect of peer review in 1-year markets with reliance









costs produced a significant increase in the buyers' ability to capture market surplus.

The buyers' profits less auditors' profits were -12.98 (126.84) in markets without (with)

peer review (F(,3,=9.29, p=.028). The availability of peer review did not produce a

significant increase in the buyers' ability to capture market surplus in markets without

reliance costs. The buyers' profits less auditors' profits were 351.03 (294.22) in the

markets without (with) peer review (F(,, )=5.22, p=.132).

These results provide strong support for H3. Allocative efficiency was

increasing in the buyers level of reliance in markets where peer review was present

and this interaction was only significant in markets with the 1-year peer review. In

markets where buyers demanded high quality services, the 1-year peer review process

allowed buyers to isolate high quality auditors which increased their capture of the

available market surplus within those markets.

Market efficiency

The results for market efficiency are reported in Figure 5-4 and Panel A of

Table 5-4. In support of H4, the market efficiency attained was significantly affected

by the interaction of the availability of peer review (PR), the timing of peer review

(PRT), and the presence of reliance costs (R) (F(1,8)=2.27, p=.085). To isolate the

significant effects producing this interaction the analysis was extended to examination

of the simple interaction of reliance and peer review for the 3-year and 1-year peer

review markets. The 3-year peer review markets, as shown in Panel B of Table 5-4,

did not reveal a significant simple interaction (F(,4) = .00, p=1.0). This indicates that

the markets' ability to generate surplus for allocation to buyers and auditors was not









significantly affected by the availability of a 3-year peer review in markets with or

without reliance cost.

In the 1-year peer reviewed markets, the simple reliance (R) x peer review

(PR) interaction (Table 5-4, Panel C) was marginally significant (F(,,4)=4.06, p=.057).

Examination of the simple-simple effect of peer review in 1-year markets with reliance

costs produced a marginally significant increase in the markets' ability to generate

surplus. The total market surplus was 86.20 (158.59) in markets without (with) peer

review (F(o3)=3.09, p=.088). The availability of peer review produced a marginally

significant decrease in total market surplus in markets without reliance costs. The

total market surplus was 376.56 (323.44) in the markets without (with) peer review

(F(1o,=18.07, p=.074).

These results provide support for H4. Market efficiency was increasing in the

buyers level of reliance in markets where peer review was present and this interaction

was only significant in markets with the 1-year peer review. In markets where buyers

demanded high quality services, the 1-year peer review process allowed markets to

increase their generation of available market surplus. Combining this results with the

results provided for allocative efficiency, the 1-year peer review allowed a pareto

dominant outcome to be achieved.

Auditor behavior

To examine auditor behavior, the percentage of low quality services provided

by auditors when they were presenting high (h) peer review reports, or the percentage

of auditors shirking, was used as the dependent variable (Table 5-5). Hypothesis 5







64

predicts a main effect of peer review timing. To improve the robustness of this test an

ANOVA model using reliance costs, peer review timing, and the interaction term was

utilized. The interactive effect of peer review timing and reliance (F(,,)=.03, p=.437)

and the main effect of reliance (F(,8)= 1.85, p=.106) were insignificant. The main

effect of peer review timing indicates a significant increase in shirking from 28.36%6

for the 1-year peer review to 64.56% for the 3-year peer review (F(,,)=4.29, p=.036)

supporting H,.

To examine this issue more closely, the required fees for the provision of high

quality services were calculated from equation (10) using the 3-year and 1-year peer

review paradigms.7 The required fee for the provision of high quality services was

260 and 214 cents for the 3-year and 1-year markets, respectively. The average prices

charged for high quality services was 232.38 and 217.71 for the 3-year and 1-year

markets, respectively.8 Analysis indicates that the 1-year market prices were not

significantly different from the fee required to maintain high quality, or 217.71 vs. 214



6The main effects for the 1-year (3-year) peer review were calculated as 28.36% =
(41.67 % + 15.05 % )/ 2 (64.56 % = (75.00 % + 54.11 %) / 2) from the cell means
presented in Table 5-5.

7These calculations were made using 8 for the value of 7, or the number of
periods the auditors might expect the game to continue. Auditors using a greater
(lessor) number of periods would have experienced lower (higher) valuations. The
average prices paid of 217.71 (232.38) for the 1-year (3-year) markets were sufficient
to support the consistent provision of high quality audit services if the auditors
expected at least five or more (11 or more) future periods in the market.

8This was calculated by taking the accepted prices for each market for audits of
high quality. Prices below 200, or the cost of providing high quality audits that
represented low-balling, were not included as this test examines the ability of prices to
sustain high quality services after the initial engagement.







65

(t(29)=1.331, p=.1938). However, the 3-year market prices were significantly below the

required high quality fee, or 232.28 vs. 260 (t(14)= -8.719, p < 0.0001). This inability

of markets that utilized the 3-year peer review to sustain high quality prices could

have affected the provision of high quality services and potentially motivated the

"shirking" behavior isolated. This suggests that auditor behavior regarding honest

representation of quality is affected by the timing of the peer review process and the

buyer's willingness to pay the requisite prices to reward "honest" performance.

Additional analysis

The extant research utilizing experimental markets in settings examining quality

provision (i.e., Dejong, Forsythe, and Lundholm [1985], and Brozovsky [1990]) has

shown that sellers (or auditors) who choose to provide low quality services earn a

greater profit than those choosing the provision of high quality desired by buyers. For

this study, this was examined by determining the profits for auditors who provided

high and low quality services, respectively. This analysis is restricted to markets

where peer review was available, or where the ability to distinguish quality was

present.

Utilizing the difference in profits between high and low quality providers as the

dependent variable, an ANOVA analysis was conducted (see Table 5-6) using reliance

costs, peer review timing, and the interaction term. The interaction (F(,7)=.68, p=.217)

and main effects of peer review timing (F(,9)=1.12, p=.162) and reliance (F(,9)=.05,

p=.413) were clearly not significant. Examination of the profit differences by cell

indicates that, in reliance markets that utilized the 1-year peer review, auditors









providing high quality services earned greater profits, although not significantly so.

These results indicate that auditors who chose to provide high quality services were

able to earn similar profits to auditors providing low quality services. This created an

environment where auditors were willing to provide the quality level desired by the

buyer, contrary to the results derived in prior studies.

Summary

These results presented here indicate that audit quality provision (H,),

allocative efficiency (H3), and market efficiency (H4) were increasing in the buyers

level of reliance in markets where peer review was present but this interaction was

only significant in markets with the 1-year peer review. In markets where buyers

demanded high quality services, the 1-year peer review process achieved a pareto

superior solution by allowing markets to increase their generation of available market

surplus while increasing the allocation of that surplus to buyers through the provision

of increased audit quality in support of H,, H3, and H4.

The examination of market prices (H2) did not reveal a significant interactive

effect of reliance, peer review, and peer review timing. Market prices were affected

by the presence of reliance costs, irrespective of the presence or timing of peer review.

Buyers were mislead by auditors pricing strategies and paid excessive prices for low

quality audit services in settings where peer review was not available.

Finally, an examination of auditor willingness to honestly represent their

intended quality level (H,) found that the more timely 1-year peer review produced

significantly lower levels of auditor shirking in support of the hypothesis.







67

Chapter 6 will review the results presented here in relation to the model offered

previously. Interpretations of the results, limitations, and potential for future research

will also be discussed.









Table 5-1
Audit Quality


Cell Means
3 Year Peer Review
Not
Available Available


1 -Year Peer Review
Not
Available Available


No Reliance 0.00% 3.13% 15.63% 12.50%
Reliance 19.53% 21.88% 7.81% 62.50%

Peer Review Timing = Timing of peer review, 1 or 3 periods
Reliance = Preference for high quality
Peer Review = Peer review availability, yes or no.

aThe p-values indicated are for a directional (one-tailed) test of significance.

bThe cell means represent the percentage of audits provided that were high quality
(low (high) quality = 0 (1)) in each market setting.


Panel A: ANOVA Results
All Markets

Sources DJE SS F Pr > F

Peer review timing (PRT) 1 .10188 2.37 .0813
Reliance (R) 1 .21584 5.01 .0278
PRT*R 1 .00050 .01 .4581

Between error term 8 .34460

Peer review (PR) 1 .10842 5.32 .0250
PR*PRT 1 .07082 3.48 .0496
PR*R 1 .10842 5.32 .0250
PR*PRT*R 1 .11444 5.62 .0226a

Within error term 8 .16296









Table 5-1 (continued)
Audit Quality


Panel B: ANOVA Results
3 Year Markets

Sources DF SS F Pr > F

Reliance (R) 1 .09770 1.82 .125

Between error term 4 .21521

Peer Review (PR) 1 .00199 0.29 .310
Rx PR 1 .00004 0.01 .471a

Within error term 4 .02771



Panel C: ANOVA Results
1 Year Markets

Sources DL SS F Pr> F

Reliance (R) 1 .11865 3.67 .064

Between error term 4 .12939

Peer Review (PR) 1 .17725 5.24 .042
Rx PR 1 .22282 6.59 .031a

Within error term 4 .13525


Reliance = Preference for high quality
Peer Review = Peer review availability, yes or no.

aThe p-values indicated are for a directional (one-tailed) test of significance.





































Audit Quality: 1-year Review
Peer Review x Reliance Interaction


70.00%


50.00%


30.00%
15.63%

10.00%
12.50%


-10.00% ---
No Reliance


Peer Review Availability
SNo Peer Review Peer Review


Reliance


Figure 5-1
Audit Quality


70.00%


50.00%


30.00%


10.00% 3.13%

0.00%
-10.00% I
No Reliance
Reliance Level

I No Peer Review Peer Review


Reliance









Table 5-2
Market Prices


Cell Means
3 Year Peer Review
Not
Available Available


1 -Year Peer Review
Not
Available Available


No Reliance 128.95 105.36 128.39 127.11
Reliance 177.30 157.52 157.92 178.38

Peer Review Timing = Timing of peer review, 1 or 3 periods
Reliance = Preference for high quality
Peer Review = Peer review availability, yes or no.

aThe p-values indicated are for a directional (one-tailed) test of significance.

'The cell means represent average market prices paid for audit services for each
market setting.


Panel A: ANOVA Results
All Markets

Sources DF SS F Pr > F

Peer review timing (PRT) 1 158.35 .06 .404
Reliance (R) 1 10956.50 4.40 .035
PRT*R 1 129.44 .05 .463

Between error term 8 19940.86

Peer review (PR) 1 195.31 .55 .240
PR*PRT 1 1304.15 3.66 .050
PR*R 1 217.50 .61 .229
PR*PRT*R 1 107.10 .30 .299a

Within error term 8 .16296












Market Price: 3-year Review
Peer Review x Reliance Interaction

200.00

180.00 177.30


. 160.00 5752

| 140.00

120.00
105.36
100.00
No Reliance Reliance
Reliance Level

N No Peer Review 4. Peer Review




Market Price: 1-year Review
Peer Review x Reliance Interaction

200.00

180.00
8.38
(D ^^ 157.92
.A 160.00

140.00
128.39
120.00 127.11

100.00 I
No Reliance Reliance
Reliance Level

u No Peer Review +. Feer Review


Figure 5-2
Market Prices









Table 5-3
Allocative Efficiency


Panel A: ANOVA Results
All Markets


F PrL>F


Peer review timing (PRT)
Reliance (R)
PRT*R

Between error term


Peer review (PR)
PR*PRT
PR*R
PR*PRT*R

Within error term


4269.35
472442.48
5439.58


8 55328.26


12135.81
204.79
12601.31
13178.92


8 19742.83


Cell Meansb


3 Year Peer Review


Not
Available


1 -Year Peer Review


Not
Available Available


No Reliance

Reliance


Peer Review Timing=
Reliance =
Peer Review =


301.47 356.47 351.03 294.22
-26.99 25.80 -12.98 126.84


Timing of peer review, 1 or 3 periods
Preference for high quality
Peer review availability, yes or no.


'The p-values indicated are for a directional (one-tailed) test of significance.

bCell means represent the average difference in surplus captured by buyers and
auditors for each market setting. The expected optimal surplus available for no
reliance (reliance) markets was 350 (172). If buyer's make the decisions predicted by
the theoretical model, the difference (buyers surplus auditor surplus) will approach
these values.


.62
68.31
.79


.227
.001
.200


4.92
.08
5.11
5.34


.029
.390
.027
.025"


Available









Table 5-3 (continued)
Allocative Efficiency


Panel B: ANOVA Results
3 Year Markets


SS

289635.09

21637.28

7746.78
3.24

6353.43


53.54



4.88
0.01


DL

1

4

1
1

4


Reliance (R)

Between error term

Peer Review (PR)
Rx PR

Within error term


Panel C: ANOVA Results
1 Year Markets


Pr>F

.005



.153
.025a


Reliance (R)

Between error term

Peer Review (PR)
RxPR

Within error term


DF

1

4

1
1

4


SS

188246.97

33690.98

4593.82
25776.99

13389.40


22.35



1.37
7.70


Reliance = Preference for high quality
Peer Review = Peer review availability, yes or no.

aThe p-values indicated are for a directional (one-tailed) test of significance.


Pr>F

.001



.046
.472a























































Figure 5-3
Allocative Efficiency









Table 5-4
Market Efficiency


Panel A: ANOVA Results
All Markets


Peer review timing (PRT)
Reliance (R)
PRT*R


1056.19
333472.24
2687.51


Between error term

Peer review (PR)
PR*PRT
PR*R
PR*PRT*R

Within error term


8 19068.78


134.73
1397.28
5061.81
5445.35


8 19183.47


Cell Meansb


3 Year Peer Review


Not
Available


1 -Year Peer Review


Not
Available Available


No Reliance
Reliance


359.38 360.94 376.56 323.44
88.02 87.29 86.20 158.59


Peer Review Timing=Timing of peer review, 1 or 3 periods
Reliance=Preference for high quality
Peer Review=Peer review availability, yes or no.


aThe p-values indicated are for a directional (one-tailed) test of significance.

bCell means represent the average total surplus (buyer surplus + auditor surplus)
generated by each market setting. The expected optimal surplus for no reliance
(reliance) markets was 350 (172). Values may exceed the expected optimum due to
randomness in the markets.


.44
139.90
1.13


.262
.001
.160a


.06
.58
2.21
2.27


.409
.234
.092
.085


Available










Table 5-4 (continued)
Market Efficiency


Panel C: ANOVA Results
1 Year Markets

Sources

Reliance (R)

Between error term

Peer Review (PR)
Rx PR

Within error term


SS

138143.08

1440.97

247.58
10503.65

10339.08


Reliance = Preference for high quality
Peer Review = Peer review availability, yes or no.

'The p-values indicated are for a directional (one-tailed) test of significance.


Panel B: ANOVA Results
3 Year Markets

Sources DF SS .E_ Pr > F

Reliance (R) 1 198016.67 44.93 .001a

Between error term 4 17627.81

Peer Review (PR) 1 .46 .00
Rx PR 1 3.50 .00

Within error term 4 8844.39


DFE

1

4

1
1

4


383.47



.10
4.06


Pr.>F

.001



.386
.057"





0.00 In
No Reliance


Mark


500.00

400.00
376.
. 300.00 3 3.

, 200.00

100.00

0.00 1
No Reliance


78



(et Efficiency: 3-year review
Peer Review x Reliance Interaction










88.02
87.29
I
Reliance
Reliance Level
M No Peer Review +. Peer Review



cet Efficiency: 1-year review
FPeer Review x Reliance Interaction










158. O

86.20

Reliance
Reliance Level
No Peer Review + Peer Review



Figure 5-4
Market Efficiency









Table 5-5
Auditor Behavior


Cell means
3 Year Peer Review


No Reliance
Reliance

Peer Review Tim
Reliance


1 -Year Peer Review


75.00% 41.67%
54.11% 15.05%


ling =


Timing of peer review, 1 or 3 periods
Preference for high quality


aThe p-values indicated are for a directional (one-tailed) test of significance.

bCell means represent percentage of auditors shirking, or providing low quality
investigations when providing a "high" peer review report.


Panel A: ANOVA Results
Markets with peer review only

Sources DE SS F Pr > F

Peer review timing (PRT) 1 .34931 4.29 .036
Reliance (R) 1 .15045 1.85 .106
PRT*R 1 .00218 .03 .437a

Error term 8 .65125









Table 5-6
Seller Surplus Analysis


Cell meansb


3 Year Peer Review


No Reliance
Reliance


Peer Review Timing =
Reliance


1 -Year Peer Review


18.22 13.62
29.76 -6.85


Timing of peer review, 1 or 3 periods
Preference for high quality


aThe p-values indicated are for a directional (one-tailed) test of significance.

bCell means represent average profits of auditors providing low quality investigations
less average profits of auditors providing high quality investigations.


Panel A: ANOVA Results
Markets with peer review only

Sources DF SS F Pr > F

Peer review timing (PRT) 1 849.24 1.12 .162a
Reliance (R) 1 39.95 .05 .413
PRT*R 1 512.19 .68 .217

Error term 7 5287.22













CHAPTER 6
SUMMARY AND CONCLUSIONS

Summary of the Study

The hypotheses developed in Chapter 3 predicted that peer review would have

an effect on the quality and price of audit services, the allocation of market surplus

between buyers and auditors and the total market surplus generated, only in settings

where high quality auditing services were demanded. It was also predicted that the

timing of the peer review process would impact the affect of peer review. In addition,

a hypothesis was developed to examine the auditor's willingness to provide consistent

high quality auditing services when utilizing peer review.

To test these hypotheses, the experiment described in Chapter 4 was conducted.

Student subjects participated in a controlled experimental market setting where the true

purpose of the study was disguised. The study was conducted over a short period of

time, two weeks, to minimize information transfer between participants. The markets

were structured to comply with the precepts of experimental economics where the

subjects earn cash compensation based on their decisions made within the

experimental setting.

Summary and Discussion of Results

The results of this study indicate that the information provided by the peer

review process is useful to buyers in assessing the quality of services they are








82
receiving. The results suggest that the heterogeneous demand for audit quality creates

a disparate valuation of the peer review process. In markets with no reliance costs,

where buyers were not concerned with auditor reporting accuracy, there was no desire

to identify high quality providers. In these markets the peer review process did not

significantly affect the quality or price of audits provided, or the generation or

allocation of market surplus within the markets.

In markets with reliance costs, the information provided by the 1-year peer

review process led to a reduction in the information asymmetry regarding audit quality

that increased the quality of audits provided in the market. Additionally, allocative

efficiency and market efficiency were improved. The 3-year review process currently

utilized within the profession did not significantly alter the market results from the

non-peer reviewed markets.

The timing of the peer review significantly affected auditor "honesty". Markets

with the 3-year peer review incurred significantly higher levels of shirking over the 1-

year peer reviewed markets.

These results imply that the peer review process has the potential to address

regulator's concerns regarding the quality of the attest function but that the timeliness

with which the information is received is critical. Support is also provided for the

imposition of mandated peer review to improve the audit quality provided within the

accounting profession. In markets where buyers had reduced valuations of audit

quality, or no reliance costs, they were unwilling to pay the fees required to motivate

and sustain high quality audit services. Additionally, in markets with reliance costs








83
where high quality was demanded, while buyers were willing to pay higher prices, the

absence of quality differentiating information in markets with no peer review allowed

auditors to provide low quality services at high prices. The results of this study

support the ability of the peer review process to provide quality differentiating

information to buyers of audit services that improves their ability to purchase high

quality auditing services, if desired.

Peer review could be viewed as the profession's response to the potential

agency relationship between the purchaser of audit services and the ultimate end user

of audit services. Managers may have different levels of reliance costs than investors,

which creates an inconsistency between their respective valuations of audit services.

Mandated peer review could be used to control the managers' decisions by removing

the low quality auditor from the market. In essence, regulators are trying to protect

the end user of the audited financial statement even if it is more costly to

management. This is consistent with results from extant economics literature which

show that the benefits from increased minimum quality standards are not distributed

equally to market participants. High quality providers and demanders benefit from

increased quality standards at the expense of market participants who provide or

demand lower quality services (e.g., Leland [1979] and Shapiro [1983]).

Limitations

Several limitations are inherent in the research design used in this study. First,

an experimental laboratory market necessarily abstracts from the external market. In

laboratory settings only the critical decision variables are theoretically developed and









utilized. To the extent that significant variables were excluded, the ability to make

inferences to settings outside the laboratory setting is compromised. Every variable

which affects decision making regarding the auditor selection process obviously can

not be incorporated in an experimental setting. This would make the markets too

complex. The strength of laboratory markets is the ability to control for confounding

variables that are not of interest to a particular study. However, these same variables,

by their exclusion, reduce the generalizability of the experimental results.

Second, the use of college students, who might make different decisions than

experienced auditors, managers, and business owners/investors, could lead to different

results than would be derived in the auditing market. Given that the experimental

design does not require any knowledge regarding the institutional variables within the

auditing market, this effect should be minimal. Not withstanding these limitations, the

experimental laboratory markets framework provides an effective means to study

issues which are hopelessly obscured by confounding variables in the empirical

markets.

Future Research

Future research should utilize multiple methods from experimental and

behavioral perspectives in an attempt to isolate which form peer review should take to

achieve the optimal cost/benefit relationship for auditors and buyers. Specific areas

that could be examined include the incorporation of a third party to the audit purchase

decision to play the role of the manager who purchases the audit that is separate from

the investor who utilizes the audit report. This could isolate the effects of having the








85
audit purchased by the manager when the audit report is actually used by the investor.

In this light, peer review could be viewed as the profession's response to the potential

agency relationship between the purchaser of audit services and the ultimate end user

of audit services. Managers may have different levels of reliance costs than investors

which creates an inconsistency between their respective valuations of audit services.

Future studies should also examine conditions that might produce low-balling

behavior by auditors as this finding is prevalent in the extant empirical research (e.g.,

Francis and Simon [1987] and Ettredge and Greenberg [1990]). In the settings used in

this study, the audit was the sole monitor of the true state of nature which increased

buyer willingness to pay higher fees, even in settings with no information about audit

quality. This reduced the auditors need to initially price their services below cost.

Future research could examine similar settings to this study with additional monitors

of the true state of nature available to buyers to see if this alters auditors behavior

regarding the pricing of high quality audit services.

A possible effect that is not investigated by this research is the influence of

audit quality on management effort. If management is confident that it's diligent

efforts will be accurately reported by the audit this might increase managers' incentives

to exert more effort. This research exogenously provided probabilistic information

about the state of nature to buyers to assist in their valuation of audit services and

does not consider their incentives regarding the accurate reporting of the true state of

nature. Future research could examine manager's incentives to purchase accurate







86

reporting services based on their private information about the true state of nature and

their effort level.













APPENDIX A
EXPERIMENTAL MATERIALS

The materials reproduced in this appendix were given to subjects assigned to

the experimental settings at indicated in Figure 3. The materials are presented in the

following order: (1) consent form; (2) general instructions for all participants; (3)

player specific information; (4) experimenter's analysis report information which was

distributed and read aloud at the end of period 15 in each experiment; (5) startup

instructions for practice periods (6) startup instructions for experiment; (7) pre-

experimental questionnaire; (8) post-experimental questionnaire; and (9) supplemental

questionnaire used for assessment of subjects risk preferences.









CONSENT FORM



This experiment is being conducted by Jeff L. Payne, Ph. D. Candidate, at the
University of Florida. My campus address and phone are Business 267, Fisher School
of Accounting, 392-0155. This is an experiment in the economics of market decision
making. You are free to withdraw from the experiment at any time and will be paid
any compensation earned up to that point. You do not have to answer any question
that you do not wish to answer. Although you are guaranteed to earn $ 7 for your
participation, most subjects earn between $15 and $35 for 2-3 hours of participation.

This experiment will be conducted on computer terminals. No special skills
are required to be able to participate in this experiment. Before the experiment begins
you will be provided with the information and instructions you will need to understand
the mechanics of the experiment. It is important that you follow the instructions
carefully. To ensure that 8 participants were present for this experiment, 12 "slots"
were available on each sign-up sheet. If more than 8 participants are present the
participants who will be allowed to continue will be determined by random draw. The
remaining participants will be paid $7.00 and excused from the session.

At the end of the experiment, when you receive your cash payment, you will
need to sign a payment report showing that you received the payment indicated. For
purposes of the experiment, your identity will remain private information and you will
only be identified by your participant number should you agree to participate in this
experiment. All cash payments will be made in a sealed envelope so that other
participants will not know how much money you made.

If you agree to participate in this experiment please read and sign the
following:

I have read and I understand to procedure described above. I agree to participate in
the procedure and I have received a copy of this description.




Signature Date


Approved for use through February 28, 1996









GENERAL INSTRUCTIONS FOR ALL PARTICIPANTS

Note: This appendix copies verbatim the laboratory market instructions read to
experimental participants. The actual instructions were printed using a large type font,
with boldface where appropriate for emphasis. Each participant received that same set
of instructions.


Thank you for your participation in this experiment!

There are 3 phases to today's session:

Phase I

1. You will receive a set of general instructions which will describe the
experiment you will participate in today (all participants will receive the same
set of general instructions).

2. A 10 15 question questionnaire will be administered to ensure that all
participants understand the instructions for the game and how their decisions
can affect their earnings. After the questionnaire has been completed and all
questions have been answered, participants will be randomly assigned to be
either buyers or sellers and will be escorted to the appropriate computer station
to begin the experiment.

3. A computer instruction and practice period will be completed to familiarize you
with the experimental setting and provide a forum for questions about
participation in the experiment. Input sheets will be provided to assist you in
understanding the rules of the game. Note that the values used during this
practice session in no way relate to the values to be used in the actual
experimental setting.







90

Phase II

Experimental Session



In this phase the actual experiment will be conducted. The experiment will
continue until you are instructed by the experimenter to stop. Each experimental
period will consist of the several steps explained in the general and participant specific
instructions.

All participants will be paid the compensation earned during the experiment at
the end of the experimental session. Earnings will be determined as explained in the
instructions provided.









Phase III

Conclusion of Experiment


At the end of the experiment each participant will be asked to complete a
questionnaire about the experimental session. This will contain a few short questions
about your background and evaluation of the experimental settings.

Upon completion of the questionnaire, participants will be paid in cash for their
accumulated earnings from the experiment and dismissed. The cash payment will be
in a sealed envelop to maintain privacy. Please return all experimental materials to
the experimenter at this time.


FINAL NOTE

Please do not discuss this experiment with the other participants, or any other person
at any time during the experimental session or thereafter. Do not compare payments
amounts or strategies. It is very important for future experiments of this type at the
University of Florida that each participant make their own decisions and not be
influenced by conversations with those who have previously participated in this or
other similar experiments. Additionally, each experiment setting is different and the
strategies that produce the highest earnings in one setting are not necessarily the
optimal strategies in the other settings. Therefore, discussing your strategies with
other players can reduce their ability to make optimal decisions.


Thank you again for your participation!










General Instructions

Everything contained in these instructions and everything you hear in this
session is an accurate representation of the experiment you will be involved in. This
is not a test of classroom knowledge. This is an experiment in economic decision
making. Be sure to ask any questions that you may have during this instruction period
and ask for assistance at any time during the experimental session. Given the close
proximity of the computer stations, please try to minimize your verbal and physical
reactions to information received during the experimental session.

The instructions that follow describe a market experiment in which participants
earn money by making certain decisions. By participating in this experiment, the
amount of money you earn will depend upon the decisions you make. The nature of
these decisions will depend on whether you play the role of a buyer or seller.

The following information and the attached diagrams will explain the sequence
which will be followed during today's experimental session. Please carefully review
these instructions and the related diagrams as the experimenter reads them aloud. The
experimenter will provide you with the opportunity to ask any questions that you
might have before the experiment begins. Feel free to make notes on these
instructions which might help you during the experimental session. You can refer to
these instructions at any time during the execution of the experiment.

There are three player "types" (or roles) in this experiment, buyers, sellers, and
the experimenter. You can only be assigned to the role of a buyer or a seller.










Sellers

Sellers increase their cash balances by selling investigation services to buyers.
Sellers can provide "high" investigation services or "low" investigation services. High
investigations are more costly to the seller than low investigations. All sellers who
provide the same quality of service will incur the same investigation cost.

If a seller performs a High investigation s/he will have a 90% chance of
accurately reporting to the experimenter the true value of a buyer's experimental unit
(This will be explained in more detail later). If a seller performs a Low investigation
s/he will have a 50% chance of accurately reporting the true value of the unit.
Therefore, a seller who provides a high investigation will have a greater opportunity to
accurately report the actual value of the unit. Sellers will select the investigation level
they will provide at the beginning of each period but can change investigation levels
between experimental periods if they choose to do so.

Example: Suppose the actual unit value is high. A seller who performs a high
investigation can expect to report a high value for the unit 9 times out
of 10. Similarly, if the actual unit value is low, a seller who performs a
high investigation can expect to report a low value for the unit 9 times
out of 10. Suppose the actual unit value is high, a seller who performs
a low investigation can expect to report a high value for the unit 5
times out of 10. Similarly, if the actual unit value is low, a seller who
performs a low investigation can expect to report a low value for the
unit 5 times out of 10.

Sellers will make offers to sell their investigations to buyers. Sellers will make
the same offer to all buyers for each experimental period. Each seller can sell their
investigation services to as many buyers as they can successfully contract with.
Therefore, it is possible for some sellers to be hired by more than one buyer, or for a
seller to not be hired by any buyer, during an experimental period.

If a seller is hired by more than one buyer, the seller will generate a separate
report for each buyer. The seller must report the unit's level determined from their
investigation. They will not be allowed to report a different value than their
investigation revealed.

Buyers will be able to observe which seller is making the offer and the amount
of the offer before making a purchase decision. Buyers will not be able to directly
observe the level of investigation that the seller has elected to provide.