. ,
Li)!
. -.

 

 

 

 

 

 

 

 

 

 

 

i p...‘ .2.
a .3:

 

 

ERSITY LIBRARIES ‘

llllllilll’i‘l\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ ~

3 1293 00914 9968

This is to certify that the

dissertation entitled

AUDITOR INDUSTRY EXPERIENCE AND
INITIAL PUBLIC OFFERINGS

presented by

Randal Jeffrey Elder

has been accepted towards fulfillment
of the requirements for

Ph.D. Accounting

degree in

 

 

 

K'Iajor professor

Date ,3 - XL—Q. ’5

 

MSU is (m Affirmative Action/Equal Opportunity Institution 0-12771

 

 

 

 

 

LIBRARY
Michigan State
'UnIversIty

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU Is An Affirmative Action/Equal Opportunity Institution
c:\clrc\datedue.nm3—p.1

 

 

AUDITOR INDUSTRY EXPERIENCE AND
INITIAL PUBLIC OFFERINGS

By

Randal Jeffrey Elder

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Accounting

1993

 

 

 

 

ABSTRACT

AUDITOR INDUSTRY EXPERIENCE AND ‘
INITIAL PUBLIC OFFERINGS

By

Randal Jeffrey Elder

This research examines the relationship between Big 6 audit firm industry
experience and auditor selection, offering underpricing, and auditor compensation
for initial public offerings (IPOs) of stock during the period 1988-1991. Auditor
experience is hypothesized to be an element of audit firm quality. Although
considerable theoretical and empirical research on audit quality exists, there is little
evidence on dimensions of audit quality other than auditor size. Auditor selection,
offering underpricing and auditor compensation are three empirical dimensions of

audit quality that have been. previously identified in the accounting literature.
Audit firm experience is examined in an IPO context because previous research has
found audit quality to be important in this setting.

This study uses multiple measures of auditor industry experience. These
measures vary along three dimensions: sales-based and client-based measures,
continuous and dichotomous measures, and three-digit and two-digit SIC code levels.
Previous research on auditor experience has generally utilized one measure of audit
firm experience.

The results of this study indicate that auditor experience is demanded by IPO
companies, although not to a greater extent than ern'sting publicly—traded companies.

The demand for auditor experience was hypothesized to be related to measures of

 

 

 

 

 

 

 

fi—fi

the information risk of the stock offering. The demand for experience was found to
be positively related to IPO company industry, but not significantly associated with

company size or age.

Most auditor experience measures were not found to be associated with lower

levels of offering underpricing. However, after controlling for individual firm

reputation effects, there was some evidence that audit firm experience is associated
with lower underpricing, especially for dichotomous measures of experience at the

three-digit SIC code level.

Audit firm industry experience was generally associated with higher levels of

audit fees. This relationship was found using both estimated audit fees, and total

 

IPO cash compensation as a proxy for audit fees.

 

 

 

Copyright by
RANDAL JEFFREY ELDER -
1993 '

 

 

 

 

ACKNOWLEDGMENTS

I wish to express my sincere appreciation to the members of my dissertation
committee - Dr. Alvin Arens, Dr. Kathy Petroni, Dr. Frank Boster, and Dr. Richard
Simonds, for their guidance, encouragement, and commitment of time and effort on
my behalf. I would especially like to express my deepest gratitude to my chairperson,
Al Arens, for the support he has given me throughout my doctoral program.

I would like to acknowledge the financial support received during my doctoral
program from the American Institute of Certified Public Accountants, and the
KPMG Peat Marwick Foundation.

There were several individuals that encouraged me to pursue a doctorate in
accounting. 1 would especially like to thank Dr. Richard Czarnecki for his
encouragement, and for being such a positive role model.

I would like to thank all my family for their support, especially my parents,
Robert and Marilyn Elder. I would also like to thank my son, Robbie, for making
it all worthwhile. Finally, my deepest gratitude and appreciation goes to my wife,

Betsy, who has continually encouraged me and made the sacrifices necessary for my

811 CCCSS.

 

 

TABLE OF CONTENTS

 

 

 

 

 

 

 

 

 

 

CHAPTER 1 - INTRODUCTION ......................................
1.1 Audit Quality ....................... ' .....................
1.1.1 Theories of Audit Quaflv .............................
1.1.2 Dimensions of Audit QualLty ..........................
1.2 Initial Public Offerings of Stock ..............................
1.3 Motivation of the Research .................................
1.4 Summary of the Research Methodology .......................
1.5 Organization of the Dissertation ............................
CHAPTER 2 - REVIEW OF RELATED RESEARCH ......................
2.1. Theories of Auditor Product Differentiation ...................
2.2 Non-IPO Empirical Evidence on Auditor Quality Differentiation
2.2.1 Auditor Selection ..................................
2.2.2 Audit Performance .................................
2.2.3 Audit Fees .......................................
2.2.4 Economic Consequences of Auditor Choice .............
2.2.5 Summary of AuditQualitLResearch ....................
2.3 Audit Quality and IPOs ...................................
2.3.1 Auditor Selection in IPOs ............................
2.3.2 IPO Audit Fees ...................................
2.3.3 Offering Underpricing ..............................
2.3.4 Summary of Audit QualiLv Research in IPOs .............
2.4 Industry Experience As a Dimension of Audit Quality ............

CHAPTER 3 - DEVELOPIVHENT OF RESEARCH HYPOTHESES,

3.1

3.2

3.3

EXPERIENCE MEASURES AND SAMPLE DESCRIPTION ......
Development of Research Hypotheses .......................
3.1.1 Demand for Auditor Experience .......................
3.1.2 Determinants of the Demand for Industry Experience ......
3.1.3 Auditor Experience and Offering Underpn'cing ...........
3.1.4 Auditor Experience and Auditor Compensation ...........
Selection and Measurement of Experience Variables .............
3.2.1 Selection of_E_xperience Measures .....................
3.2.2 Audit Firm Eigerience Measures ......................
Initial Public Offering Sample Data ..........................
3.3.1 Sample Description ................................
3.3.2 _E_xperience Measure Correlations ......................

vi

10
12

15
15
21
21
22
24
26
27
29
3O
32

41
41
41
42
44
45
47
47
51
53
53
57

CHAPTER 4- DEMAND FOR AUDITOR EXPERIENCE RESULTS ....... -. . 60

 

 

 

 

 

 

 

 

 

 

4.1 Introdtrction ....................................... ' ...... 60
4.2 Demand for Auditor Experience by IPO Companies ............. 60
4. 2.1 Random Selection Model Results ...................... 63
4.2.2 Market Share Weighted Selection Model ................ ‘ 65
4.2.3 IndusthWeighted Selection Model ............. ' ....... 67
4 2. 4 Summary of Test Results- Demand for Auditor Experience . 69
4.3 Determinants of the Demand for Auditor Experience ............ 71
4.3.1 Model Develmpment ................................ 71
4.3.2 Model Results .................................... 76

4.3.3 Summary on Determinants of the Demand for Auditor
Experience ....................................... 80
CHAPTER 5 - OFFERING UNDERPRICING RESULTS ................... 83
5.1 Introduction ............................................ 83
5.2 Underpn'cing Model ...................................... 83
5.3 Preliminary Model Tests .................................. 88
5.4 Research Results ........................................ 92
5.5 Specification Tests ....................................... 97
5.5.1 Effects of Big_6 Mergers ............................. 99
5.6 Summary and Conclusion for Underpricing Tests ................ 103
CHAPTER 6 - AUDITOR COMPENSATION RESULTS .................... 106
6.1 Introduction ............................................ 106
6.2 Auditor Compensation Model .............................. 106
6.3 Preliminary Model Tests .................................. 108
6.4 Research Results ........................................ 111
6.4.1 Compensation Regmssion with Audit Fees ............... 118
6.4.2 Effects of 136 Mergers ............................. 120
6.5 Summary and Conclusions for Auditor Compensation Tests ....... 122

CHAPTER 7 - SUMNIARY, CONCLUSIONS, AND

7.1

7.2
7.3

7.4

 

 

 

 

 

 

 

 

 

SUGGESTIONS FOR FUTURE RESEARCH ............ 123
Summary of Results ...................................... 123
7.1.1 Demand for Auditor mnence ....................... 123
7.1.2 Determinants of the Demand for Auditor Experience ...... 125
7.1.3 Offering Underpricing .............................. 127
7.1.4 Auditor Compensation .............................. 127
Comparison to Previous Audit Quality Research ................ 128
Other Results ........................................... 133
7.3.1 Measurement of Auditor Experience ................... 133
7.3.2 @pchange Effects .................................. 134
7.3.3 Effect of Mergers on Firm Reputational Capital . . . .' ...... 134
7.3.4 Replication of Previous Research ...................... 135
Limitations and Suggestions for Future Research ................ 136
7.4.1 Measurement of Audit Firm Experience ................. 136
7.4.2 Data Limitations ................................... 136

 

vii

 

 

LIST OF REFERENCES .............. , ............................... '138

APPENDICES
Description of Appendices Contents .................................. 143
APPENDIX A - BIG 8 MARKET SHARES - JANUARY, 1988 ............. 144
A] Client-Based Market Shares - Three-Digit SIC Level ............. 144
A2 Sales—Based Market Shares - Three-Digit SIC Level .............. 147
A3 Client-Based Market Shares - Two-Digit SIC Level .............. 150
A4 Sales—Based Market Shares - Two-Digit SIC Level ............... 152
APPENDIX B - BIG 6 MARKET SHARES - JANUARY, 1990 ............. 154
El Client-Based Market Shares - Three-Digit SIC Level ............. 154
B2 Sales-Based Market Shares - Three-Digit SIC Level .............. 156
B3 Client-Based Market Shares - Two-Digit SIC Level .............. 158

B4 Sales-Based Market Shares - Two-Digit SIC Level ............... 160

viii

map
3.1
3.2
3.3
4.1
4.2

4.3

4.4

4.5

4.6
4.7

5.1
5.2
5.3
5.4
5.5
5.6
6.1
6.2

LIST OF TABLES

1mg

Number of Offerings By Year and Auditor Type
Sample Descriptive Statistics

Correlation Matrix of Experience Measures
Mean Auditor Experience Levels - IPO Sample

Comparison of IPO Sample Experience Measures To Expected
Values Under Random Selection

Comparison of IPO Sample Experience Measures To Expected
Values Under Random Selection Weighted By Total Market
Share '

Comparison of IPO Sample Experience Measures to Expected
Values Under Random Selection Weighted by Industry Market

Share ‘

Logistic Regression Model - IPO Auditor Selection - Preliminary
Test

Auditor Experience Model - Continuous Measures

Auditor Experience Logistic Regression Model - Dichotomous
Measures

Descriptive Statistics by Exchange - IPO Sample 1988-1991
IPO Underpricing Model - Preliminary Tesr

IPO Underpricing Model - Three-digit Experience Measures
IPO Underpricing Model - Two-digit Experience Measures
Underpricing by Audit Firm

IPO Underpricing Model - Merged Firm Dummy Model
IPO Cash Compensation - Preliminary Test

IPO Cash Compensation Model » Three-digit Experience
Measures

66
68

75

77
79

87
90
94
96
100
102
110
112

"U

Table Title ' age

6.3 .IPO Cash Compensation Model — Full Sample of IPOs with . .115
Experience Variable

6.4 IPO Cash Compensation Model - Two-digit Experience . ‘ 117
Measures

6.5 IPO Audit Fee Model - Three-digit Experience Measures 119

6.6 Fee Residual by Audit Firm 121

 

 

 

Figure
1.1

2.1
2.2
2.3
3.1
3.2
4.1
4.2

7.1
7.2

7.3

LIST OF FIGURES

_T_itl_e

Summary of Research Design

Audit Quality Theories - Summary of Implications
Summary of Audit Quality Research

Comparison of IPO and non-IPO Audit Quality Research
Summary of Research Hypotheses

Summary of Experience Measures

Summary of Results - Demand for Auditor Experience

Summary of Results - Determinants of Demand for Auditor
Experience

Summary of Research Results

Comparison of Research Results to Previous Audit Quality
Research

Detailed Comparison of Research Results to Previous Audit
Quality Research

xi

if?
99
CD

124

130

131

 

 

 

 

CHAPTER 1 - INTRODUCTION

This study is an empirical investigation of the role of auditor industry
experience in initial public offerings of common stock (IPOs). The purpose of this
investigation is to assess whether auditor industry experience is an important
dimension of audit quality. Several definitions of auditor experience are tested in
an attempt to identify the types of experience valued by market participants. This
study will help further our understanding of the nature of audit quality, and the
economic role of auditing.

Several theories of the nature of audit quality exist. Based on these theories,
extensive empirical evidence has documented that the largest audit firms, the so-
called "Big 6 firms", provide quality-differentiated audit services (e. g., Palmrose [1986
and 1988], Francis and Wilson [1988]).1 Little theory or evidence exists to support
quality along a dimension other than auditor size.

The information quality theory of Titman and Trueman [1986] suggests that
audit quality is reflected by the precision of the information provided by the auditor.
Titman and Trueman further suggest that industry experience may be one dimension
of audit quality. The practice literature also supports the importance of auditor
industry experience. Professional standards require that the auditor have knowledge
of the client’s industry (AICPA [1991], AU 311.06).

This study examines the relationship between auditor experience and three
dimensions of audit quality: auditor selection, the underpricing of the IPO, and

auditor compensation.2 Most studies of audit quality, including those that examine

 

 

 

 

2

auditor industry experience, have examined a single dimension of audit quality. The
three dimensions of audit quality in this study are not independent. For example,
the cost of audit quality may affect the extent-to which audit quality is demanded.
Examination of multiple dimensions of audit quality allows consideration of their
interdependence, and increases the power of. the empirical tests to assess whether
auditor industry experience is an important dimension of audit quality.

This study is conducted in an IPO setting for offerings with a Big 6 auditor.
This setting is chosen because previous studies have demonstrated that Big 6 auditors
are demanded by IPO companies (Simunic and Stein [1987]; Menon and Williams
[1991]), and selection of a Big 6 auditor affects the level of underpricing of the
offering (Beatty [1989], Balvers, McDonald and Miller [1988]). Further, the
information quality model of Titman and Tru eman [1986] provides a theoretical basis
for a relationship between audit quality and offering underpricing. Considerable
research indicates that the Big 6 firms are quality-differentiated, and they are the
only audit firms with meaningful levels of industry experience. Therefore, this study
_ is restricted to Big 6 firms.

The remainder of this chapter is organized as follows. Section 1.1 discusses
the theories and empirical dimensions of audit quality. Section 1.2 describes the IPO
environment. Section 1.3 describes the motivation and expected contributions of the
study. Section 1.4 is a summary of the research design. Section_1.5 summarizes the

organization of the remainder of the dissertation.

 

 

 

1.1 Audit Quality
1.1.1. Theories of Audit Quality

Three main theories of audit quality exist. In the independence theory, audit
quality is determined by the auditor’s independence, which is a function of size. In
the risk-sharing model, audit quality is a function of the auditor’s ability to share
risk. The ability to share risk is also a function of audit firm size. In the
information quality theory, audit quality is a function of the quality of the
information provided by the auditor.

The independence model deveIOped by DeAngelo [1981] has been used to
explain the existence of quality differentiation by the Big 6. DeAngelo defined audit
quality as the "market-assessed joint probability that a given auditor will both (a)
discover a breach in the client’s accounting system. and (b) report the breach" (p.
186). DeAngelo focused on the reporting of the accounting breach, which is partially
a function of the independence of the auditor. She argued that large firms are likely
to be more independent.

The emphasis in DeAngelo’s theory on size has become the theoretical
underpinning for an extensive body of research that indicates that the Big 6 firms are
quality-differentiated. The empirical evidence, excluding evidence related to initial
public offerings, encompasses auditor selection (e.g., Francis and Wilson [1988]),
audit performance outcomes (e.g., Palmrose [1988]), audit fees (e.g., Simunic [1980];
Palmrose [1986]), and the economic effects of auditor choice (e.g., Nichols and Smith
[1983]; Ettredge, Shane and Smith [1988]). Cumulatively, these studies present

convincing evidence the Big 6 firms are qualityudifferentiated.

 

 

 

4

Antle [1982] argued that one role of the auditor is to share risk. Because the .
ability to share risk is a function of auditor size, the implications of the risk-(sharing
theory are similar to theindependence theory. However, the risk-sharing theory
does not imply actual differences in audit quality.

Titman and Trueman [1986] developed a model in which the information risk
of an IPO is reduced through use of a quality-differentiated auditor. Although the
Titman and Trueman model is based on an IPO setting, it can be applied more
generally to other auditor choice situations. Titman and Trueman defined audit
quality as: "the accuracy of information supplied to investors; the information
provided by a higher-quality auditor allows an investor to make a more precise
estimate of the firms’s value" (p. 160). Titman and Trueman suggested that
knowledge of industry conditions is one aspect of auditor quality.

The empirical research on audit quality in non-IPO settings is mirrored by the
research into the market for initial public offerings of stock. IPO research has
demonstrated quality differentiation by the largest firms in auditor selection (Simunic
and Stein [1986]; Menon and Williams [1991]); auditor compensation (Beatty [1989]),
and offering underpricing (Balvers, McDonald and Miller [1988]).

The empirical tests in this study are based upon the information quality
theory, as this theory allows for quality differentiation on a basis other than size.
Whether auditor experience is a form of audit quality is an empirical issue.

It is important to note that the tests in this study do not constitute tests of
one audit quality theory in favor of another. The theories of audit quality are not

mutually exclusive, and it is likely that audit quality embraces all three concepts. For

 

 

 

5

example, ‘ while DeAngelo [1981] argued that audit quality is a function of
independence, she also suggested that one benefit of auditor size is the increased
industry experience of the auditor.

1.1.2 Dimergions of Audit many

This Study identifies four empirical dimensions of audit quality. Audit quality
can affect auditor selection, audit performance, audit fees, and the economic
consequences of auditor choice. This study examines the relationship between
auditor experience and each dimension of audit quality, except audit performance.
Although audit performance may provide the clearest indication of audit quality,
direct measures of audit performance are difficult to obtain. Multiple dimensions
of audit quality are investigated, because studies along one dimension of auditor
quality have not always yielded consistent results. In addition, the results along one
dimension of audit quality can be confounded by-another dimension. For example,
auditor selection is related to auditor compensation and the economic consequences
of auditor choice.

Auditor selection involves both the measurement of the extent to which audit
quality is demanded, and the identification of specific company factors that will lead
to the choice of a higher-quality auditor. According to the Titman and Trueman
[1986] model, one benefit of the use of a high-quality auditor is the reduction of
5 information risk and its effect on the underpricing of the offering. In the models
developed in this study, the demand for high-quality auditors is a function of the

information risk of the offering.

 

 

 

 

6

Economic theory has been used to predict that audit quality will be associated
with higher audit fees. Several empirical studies, (e.g., Palmrose [1986]; Francis and
Simon [1987]) demonstrate that Big 6 audit firms charge higher fees. However, in
attempting to explain observed levels of auditor concentration within industries, it
has alsobeen argued that industry experience may lead to economies of scale that
result in lower audit fees (Danos and Eichenseher [1986]). Measurement of the cost
of audit quality is important, because cost is one determinant of demand.

The economic consequences of auditor choice are the economic benefits and
costs associated with auditor choice. The economic consequences can include any
benefit or cost associated with the choice of auditor. Although the costs include the
audit fee, this generally has been regarded as a separate area of research. In this
study, the economic consequence of auditor choice is the effect of auditor experience
, on the underpricing of the initial public offering of stock. Previous empirical studies
by Balvers et al. [1988] and Beatty [1989] indicate that underpricing decreases with
the use of a high-quality auditor. Offering companies wish to minimize underpricing

since it reduces the amount of offering proceeds.

1.2 Initial Public Offerings of Stock

Initial public offerings have several features that make them desirable for
study. The information environment is such that auditor reputation might be
expected to play an important role in the process. As discussed in the previous

section, the information quality theory of Titman and Trueman [1986] is based on

 

 

 

7

an IPO setting. Institutional features of the market also suggest a dynamic
environment for audit choice.

The IPO environment is one marked by information asymmetry between the
offering entrepreneurs and investors (Leland and Pyle [1977]). Ex ante uncertainty
exists concerning the value of the investment because less information exists about
the IPO company. While publicly-traded companies are subject to analyst and media
scrutiny, privately-held companies are less scrutinized. Many IPO companies are
young entities, so less of an operating history exists for estimating firm value.

The level of underpricing of the offering is related to the ex ante uncertainty
concerning firm value (Rock [1982]; Beatty and Ritter [1986]). Previous studies by
Balvers et al. [1988] and Beatty [1989] indicate that underpricing is less with a
higher-quality auditor. Studies in other settings have generally been unable to find
a significant economic consequence associated with auditor choice (Nichols and
Smith [1983]; Ettredge, Shane and Smith [1988]).

Initial public offerings are also an ideal environment for examining auditor
selection. IPOs have been associated with the decision to change auditors.
(Carpenter and Strawser [1971]; Menon and Williams [1991]). For many young IPO
companies, the auditor at the time of the IPO will likely be the company’s first
auditor. For the sample of IPO companies in this study, the median company age
was approximately four years. Accordingly, the relationship between auditor
experience and company features may be less noisy for IPO-companies than for

existing publicly-traded companies.

 

 

 

 

 

 

8

It is known that the distribution of clients across industries is not random
(Eichenseher and Danos [1981]). However, little is known about the process by
which audit firms gain market share in particular industries. IPO companies are a

potentially important part of the changes in industry client distributions.

1.3 Motivation of the Research

The primary purpose of this study is to determine whether auditor industry
experience is an important element of audit quality. The empirical research on audit
quality presents convincing evidence that the Big 6 are the quality providers of audit
services in the market for publicly-traded clients. However, existing research
provides no insight into specific factors, other than audit firm size, that accounts for
the differential audit quality of these firms. Because the Big 6 dominate the market
for publicly-traded companies, definitions of audit quality that partition the airdit
market between Big 6 and non-Big 6 firms are not very informative as to the nature
of audit quality.

Differences in audit quality based on audit firm industry experience would
significantly alter the way in which the audit market is viewed. Most empirical
research views the Big 6 as having a relatively uniform level of audit quality.
Experience-based differences in audit quality would not only allow for differences in
audit quality, among the Big 6 firms, but also within individual audit firms, dependent
upon the client’s industry. Audit quality would also not be restricted to the Big 6.
Other national and regional firms could also be quality providers of audits in those

industries where they have significant levels of experience.

 

 

 

 

9

Auditor quality differentiation is an important issue to auditors, clients and
capital market participants. Knowledge of the economic benefit and cost of auditor
choice should be of direct interest to clients and capital market participants. For
clients, greater knowledge about the dimensions of auditor quality may help them
make more informed auditor selection decisions. Capital market participants can
better understand how auditor type affects information and the value of their
investments. For CPA firms, knowledge of the role of auditor experience in auditor
selection can benefit'CPA firms in their marketing strategies. Although this study
examines the IPO market for Big 6 firms, the results may potentially be of interest
to non-Big 6 firms. By developing industry-specific expertise, these firms may be
able to demonstrate a sufficient quality level to retain clients that go public. This
issue is an important one for non-Big 6 firms since clients often change to a Big 6
auditor prior to going public (Carpenter and-Strawser [1971]). i

From a theoretical perspective, examining the role of experience provides a
direct test of the information quality hypothesis in Titman and Trueman [1986]. The
existence of quality differentiation on a basisother than size is not predicted by other
theories of audit quality.

Auditor industry experience is the most promising area for examining audit
quality on a basis other than size. Initial empirical studies into the role of auditor
experience on audit pricing did not suggest that auditor experience was an important
dimension of the audit market (Palmrose [1986]). However, more recent research
suggests that auditor experience is related to the auditor change decision (Williams

[1988]), and the change in fees due to a change in auditor (Ettredge and Greenberg

 

 

 

10

[1990]), Studies in the governmental sector have suggested the exiStence of an
experience-related fee premium (Ward, Elder and Katteltrs [1993]), and that auditors
with industry experience perform audits of higher quality (Deis and Giroux [1992a]).
Previous research has not demonstrated economic consequences associated with the
choice of an experienced auditor, nor has it simultaneously considered multiple
dimensions of the auditor choice decision. This researcher is unaware of any other
audit firm attributes that have been empirically examined in a market setting, or that

are directly referenced in the professional literature.

1.4 Summary of the Research Methodology

. This study examines the relationship between industry experience and auditor
selection, underpricing of the offering, and auditor compensation. Industry
experience is hypothesized to be a dimension of audit quality and is the research
variable of interest. Auditor selection, offering underpricing and auditor
compensation represent three dimensions of the effects of audit quality in the audit
market.

Regression models are developed for each element of the choice decision to
control for other factors that affect auditor selection, auditor compensation, and
offering underpricing. Initial public offerings of stock were chosen as the research
environment because auditor choice has previously been identified as being

important in this context. Figure 1.1 summarizes the research design.

 

 

11

 

 

Auditor
Selection

 

 

 

 

 

 

Auditor Industry
Compensation Experience

 

 

 

 

 

 

 

 

 

Economic
Consequences -
IPO Underpricing

 

 

 

DIMENSIONS DIMENSION OF
OF AUDITOR AUDIT
CHOICE QUALITY

Research Environment -
Initial Public Offerings

Figure 1.1
Summary of Research Design

 

 

 

12

The first phase of the auditor selection decision is measuring the extent to
which auditor experience is demanded by IPO. companies. Models of auditor choice
will then be developed in which auditor type is the dependent variable, and the
independent variables are company characteristics expected to result in demand for
auditor experience. Auditor experience is included as an independent variable in
models of offering underpricing and auditor compensation. Auditor experience is
predicted to be associated with lower offering underpricing, consistent with a
reduction in the information risk associated with the offering. Auditor experience
is also hypothesized to be associated with an audit fee premium.

IPOs issued during 1988-1991 were selected for examination. This is the most
recent period for which data is available. This period also spans the period in which
mergers reduced the Big 8 to the Big 6. Examination of the pre- and post-merger
periods may provide insights into the impacts of these mergers on the reputational

capital of the largest CPA firms.

1.5 Organization of the Dissertation

The remainder of the dissertation is organized into six chapters. Chapter Two
contains a description and discussion of previous literature concerning auditor
product differentiation. Included in Chapter Two is a review of theories of audit
quality, a summary of empirical evidence on audit quality, and a review of the
literature on auditor industry experience. Research hypotheses are developed in
Chapter Three. A description of the experience measures and sample is also

contained in Chapter Three. The remaining four chapters contain the results and

 

 

 

 

13

conclusions of the study. Chapter Four contains the results of the tests of auditor
selection. Included in this chapter are measures of the . demand for auditor
experience, as well as models of the determinants of the demand for audit
experience. The results of the tests of the relationship between auditor industry
experience and offering underpricing are contained in Chapter Five. Tests of the
relationship between auditor experience and auditor compensation are included in
Chapter Six. Chapters Five and Six also present results of tests in which membership
in the Big 6 is the measure of audit quality. These are presented as a replication of
previous research, and as a basis of comparison for the results of the tests of auditor
industry experience. Chapter Seven includes a summary of the findings of the
research, a discussion of the limitations of the study, and suggestions for future

research.

 

 

 

14
ENDNOTES

The Big 6 firms are Arthur Andersen, Coopers & Lybrand, Deloitte & Touche, Ernst
and Young, KPMG Peat Marwick, and Price Waterhouse. In late 1989, the firms of
Ernst & Whinney and Arthur Young merged to create Ernst & Young. In early
1990, the firms of Deloitte, Haskins and Sells and Touche Ross merged to create
Deloitte and Touche. Although the period examined in this study spans these two
mergers, the current Big 6 terminology is used.

Underpricing is a measure of the initial return on newly traded stock. The term
underpricing refers to the fact that the average initial return is positive.

 

 

 

CHAPTER 2 - REVIEW OF RELATED RESEARCH

This chapter consists of a review of the theoretical and empirical literature on
audit quality. The existing research is used a basis for developing empirical
predictions concerning the relationship between auditor industry experience and
auditor selection, offering underpricing, and audit fees for IPO companies.

Section 2.1 describes the theories of audit quality and summarizes their
empirical predictions. Section 2.2 reviews empirical research on audit quality in non-
IPO settings. Section 2.3 describes the environment for initial public offerings of
stock and reviews previous research on auditor selection, audit fees, and offering
underpricing in an IPO setting. Section 2.4 reviews previous research on auditor

experience as a dimension of audit quality.

2.1. Theories of Auditor Product Differentiation

There are three main theories of auditor product differentiation. Each theory
addresses a different dimension of auditor quality. These theories address the
auditor’s independence, risk-sharing ability, and the information quality provided by
the auditor. .

The theories differ in some of their implications along the four dimensions
of audit quality: auditor selection, audit performance, audit fees, and the economic
consequences of auditor choice. Except for audit performance, the empirical
predictions derived from these theories are often dependent upon other theories or

assumptions. Auditor selection models depend upon theories of specific factors

)_.a
'JI

 

 

16

which give rise to the demand for audit quality. The economic consequences of
auditor choice depend upon characteristics 'of the securities markets, such as the
information environment. Finally, predictions about audit fees depend upon
assumptions about the structure of the audit market.

The independence theory in DeAngelo [1981] states that audit quality is
dependent upon the "market assessed joint probability that a given auditor will both
a) discover a breach in the client’s accounting system, and b) report the breach (p.
186). DeAngelo focused on the probability that the auditor will report the breach,
which she posited is a function of the auditor’s independence. She further argued
that large firms are likely to be more independent since any one client is insignificant
to the firm’s practice. The implication of this model is that the largest audit firms
are quality differentiated.

DeAngelo’s theory is appealing because of its clear link between auditor size
and auditor quality based on independence. Auditor independence is the foundation
of the independent auditing function. It is probable that an audit firm will be less
likely to withstand client pressure when that client constitutes a large portion of an
audit firm’s practice. V

The independence theory is less appealing because it focuses on the reporting,
rather than the detection element of the audit function. Auditing is primarily an
evidence-gathering process, and it is likely that if firms differ in quality, it is on the
basis of their audit procedures. Cushing and Loebbecke [1986] found that the largest
audit firms differ in their audit approaches. Further, under the independence theory,

quality-differentiation depends upon the apparent intentional failure to report an

 

 

 

17

error due to lack of independence. This researcher is unaware of any empirical
evidence to support the contention that certain audit firms are more likely to report
errors.

The independence theory is clear in its prediction that large audit firms
perform audits of higher quality, because they are more likely to report errors.
Accordingly, the demand for the largest audit firms should be related to those
company characteristics that create a demand for audit quality. The market should
react differently to information contained in the audit reports of these firms, because
it is of higher quality. The largest firms will also receive a fee premium if quality is
priced in the marketplace and there are no scale economies associated with auditor
size.

Antle [1982].has suggested that one role of the auditor is to share risk. As
with the independence theory, the implication of this model is that the largest firms
are quality-differentiated. Larger firms with a greater capital base are better able
to absorb risk, so their services are more valuable. That is, Big 6 firms are quality-
differentiated on their ability to share risk. Although risk-sharing appears to be one
plausible explanation for the choice of a Big 6 firm at the time of an IPO, this line
of reasoning is seldom advanced in empirical tests.

Although both the independence and risk-sharing theories indicate that the
largest firms are quality-differentiated, they differ in their empirical predictions. The
risk-sharing theory does not by nature imply differences in actual audit quality. In
the risk-sharing model, the demand for audit quality is also clearly linked to company

risk characteristics. The consequences of use of a large audit firm are related to the

 

 

18

benefits of sharing risk. Large audit firms earn a fee premium due to the assumption
of additional risk.

Titman and Trueman [1986] developed a theoretical model based on
information quality of the effect of audit quality on the pricing of an IPO. In their
model, an owner can signal private information about the value of the firm through
the choice of the quality level of the auditor. Titman and Trueman do not define
specific characteristics of high-quality auditors in their model. Although the model
is applied to an IPO context, it is more general and can be applied to other auditor
and non-auditor choice situations.

In their model, IPO quality is revealed by auditor choice. Because a high
quality auditor reveals more accurate information to the market, only firms with
favorable information to reveal will choose to disclose more accurate information
through selection of a high quality auditor. Low-quality companies cannot mimic
high-quality companies by hiring a better auditor because the high-quality auditor is
more likely to reveal unfavorable information about the company. i

F eltham, Hughes and Simunic [1991] have criticized the Titman and Trueman
model for its assumption that the auditor produces new information. They argue
that this is inconsistent with the audit process. They appeal to a model by Datar,
Feltham and Hughes [1991] in which the auditor strictly provides an examination of
the report proposed by the entrepreneur.1

Their criticism of the Titman and Trueman model takes a static view of the
audit process. Auditing is a dynamic process in which both the entrepreneur’s

financial report and the auditor’s opinion can change based on the actions of the

 

 

 

19
other party. The use of the term "proposed report" (p. 376) by Feltham, Hughes and

Simunic is itself an indication that the report can change as a result of the audit
process. That is, information is- generated. Indeed, a focus of the audit is to detect
intentional and unintentional errors. By their nature, unintentional errors would not
be known to the entrepreneur or to investors.

The Titman and Trueman model is an attractive characterization of the audit
market, because actual audit performance leads to product differentiation in their
model. Further, this differentiation is not dependent on the size of the CPA firm.
This is not to suggest that the largest CPA firms are not quality-differentiated. They
have the resources to hire the best personnel, and can devote substantial resources
to training and developing technical practice aids. However, the Titman and
Trueman model does not preclude quality differences among the Big 6, nor does it
preclude the performance of quality audits by non-Big 6 firms.

Figure 2.1 summarizes the potential implications of the three audit quality
theories along the dimensions of auditor selection, audit quality, audit fees, and
economic consequences of auditor choice. It .is important to note that these theories
are not mutually exclusive, and they have several overlapping predictions.

Only the information quality theory allows for quality differentiation on a
basis other than size. The information quality theory provides a theoretical basis
of investigating other potential dimensions of audit quality, including auditor industry
experience. As the last column of Figure 2.1 indicates, if auditor industry experience

is a relevant dimension of audit quality, it may be related to all four empirical

dimensions of audit quality.

 

 

20

 

 

Auditor
Selection

Audit
Performance

Audit
Fees

Economic
Consequences

Independence

Not specified.
Depends on other
factors (i.e.,
agency factors)
which would lead
to a demand for
auditor quality.

Function of
independence
which is based on
audit firm size.

Large firms earn a
premium due to
higher
reputational

quality.

Higher quality
audit perceived as
such by market,
which presumably
acts differently to
higher quality
information.

Risk-Sharing

Risky firms choose
higher quality
auditors since they
most benefit by the
sharing of risk.

No quality effect
based on risk-
sharing.

Large firms earn a
premium due to
their ability to
share risk.

Sharing of risk
affects overall risk
of company.
Market reacts to
effect of lower risk.

Information

Quality

High quality
companies choose
high quality
auditors.

Higher quality
firms provide
higher quality
audits defined by
information

quality.

High quality firms
earn a fee
premium.

Market reacts
favorably to higher
information

quality.

 

 

Figure 2.1
Audit Quality Theories
Summary of Implications

 

 

 

 

21

2.2 Non-IPO Empirical Evidence on Auditor Quality Differentiation
I In this section, existing empirical studies on audit quality in non-IPO settings

are discussed. These studies almost exclusively compare Big 6 and non-Big 6 firms.
The review of these studies is organized along the dimensions of auditor selection,
audit performance, audit fees, and economic effects of auditor choice. Most of the
empirical research relates to audit fees.
2.2.1 Auditor Selection

The auditor selection process can be characterized as a process of feature-
matching between the auditor and client. Both client and auditor attributes can be
the subject of study. In empirical audit quality studies, audit quality has generally
been the dependent variable, and client attributes that give rise to the demand for
a given level of audit quality are the variables of interest.2 Most of the empirical
research on auditor selection has focused on auditor changes. This reflects an
inherent interest in the decision to change auditors.

Francis and Wilson [1988] have modeled the demand for auditor quality as
a function of the agency costs of a company.3 Francis and Wilson examined the
association of auditor changes with both changes and levels of agency costs. The
agency costs variables include individual variables that had been tested individually
in previous studies of auditor choice.4 The variables tested include managerial
ownership, the existence of bonus plans, diffusion of ownership, leverage, and
subsequent issues of stock or debt.

Francis and Wilson examined two models of audit quality, which they viewed

as potentially competing theories. The continuous model of audit quality is based

 

 

 

22

on DeAngelo’s theory of auditor independence. Audit quality in this model is
measured as the log of sales dollars audited. The competing model’is the brand
name model, based on an economic theory of product differentiation by Klein,
Crawford and Alchian [1978]. In the test of this model, audit quality is measured by
the usual Big 6/non-Big 6 dichotomous classifications

Francis and Wilson found that changes in audit quality in the brand name
model are weakly associated with both levels and changes in agency costs. A change
in quality in the brand name model is a change to (from) a Big 6 auditor from (to)
a non-Big 6 auditor. They did not find support for the continuous model of audit
quality. The overall explanatory power of each model was low. Their generally weak
results are consistent with earlier studies which examined the relationship between
specific agency variables and auditor choice.

The support of the brand name model over the continuous model can be
interpreted as consistent with the information quality theory, since quality in the
brand name model is on a dimension other than size. This interpretation should be
made with caution. Their test is not a powerful test for discriminating the brand
name and continuous models of quality since the definitions of audit quality in the
two models are highly correlated.

2.2.2 Audit Performance

Very little direct research evidence on audit quality exists. Palmrose [1988]
studied auditor litigation for the period 1960-1985 and found that non-Big 6 firms
as a group have higher litigation activity than Big 6 firms. She interpreted this result

as consistent with quality-differentiation by the Big 6. However, the lack of control

23

forclient factors associated with litigation activities suggests caution in interpreting
this result as a direct test of any theory of audit quality.

Palmrose also found significant differences in litigation activity within the
Big 6, and that litigation activity varies within industry classifications. Her results
support the existence of quality differentiation within the Big 6. However, her
sample sizes were small, and the results were sensitive to the form of analysis.

Imhoff [1988] performed a survey of financial analysts to determine if
perceived accounting quality differed among the. clients of different CPA firms.
Because the quality of reported accounting numbers is affected by the audit process,
his study is an implied test of audit quality. Imhoff did not find significant
differences in perceived accounting quality between clients of Big 6 firms, or between
clients of Big 6 and non-Big 6 firms. His research suggests that there are not
significant differences in audit quality between CPA firms. His research can also be
interpreted that the demand for audit quality does not differ significantly among
companies. This result is consistent with the Francis and Wilson [1988] study that
found only a weak association between auditor changes and agency costs.

A recent study by Deis and Giroux [1992a] examined differences in audit
quality among small CPA firms in the audits of Texas school districts. Audit quality
was measured by the subjectively weighted score of the results of a Quality Control
Review by the Audit Division of the Texas Education Agency.

Deis and Giroux found that audit quality increases with the number of school
district clients. They interpreted this 'result as consistent with DeAngelo’s

independence theory, although they acknowledge that this result is potentially

 

 

24

confounded with industry experience effects. However, theindependence theory
suggests that independence is a function of total audit firm size, not size in a
particular industry. Deis and Giroux could have controlled for total audit firm size.
If they had controlled for total audit firm size, a significant relationship between
industry experience and audit quality would have provided evidence that industry
experience is a dimension of audit quality.
2.2.3 Audit Fees

The link between audit quality and audit fees is complex. Palmrose [1986]
discusses three hypotheses that provide for a price differential between Big 6 and
non-Big 6 firms. The first hypothesis is that the Big 6 charge higher prices due to
monopoly power. The second hypothesis is that the Big 6 receive a fee premium
because they provide a higher quality service. The third hypothesis is that the Big
6 charge lower prices due to scale economies. The three hypotheses are not mutually
exclusive. In fact, all three relationships can be present in the market, hindering the
ability to make any inferences about quality from pricing data. However, the
maintained hypothesis in most audit fee studies is that the Big 6 receive a fee
premium because they provide audits of higher quality.

In an early study of audit fees, Simunic [1980] found that on average the Big
6 charged lower fees than non-Big 6 firms, which he interpreted as consistent with
the existence of scale economies.6 The lower fees charged by the Big 6 was
interpreted as supporting the existence of competition, and a lack of monopoly rents.

However, most subsequent research has found a fee premium for the Big 6.

 

 

25

While Simunic found that the Big 6 on average charged lower fees, he. found
that Price Waterhouse did charge significantly higher fees. This is the first empirical
evidence of potential reputational differences among the largest CPA firms.

Simon [1985] performed a replication of Simunic’s study and was unable to
demonstrate a fee premium for Price Waterhouse. Simon attributed the difference
in results to increasing competition that diminished Price Waterhouse’s ability to
charge a premium. If this interpretation is correct, it would require a fairly sudden
change in competitive forces. Simunic’s sample consisted of audit fees for the year
1977; Simon studied audit fees for the period 1978-1983. A more plausible
explanation is that the studies used different samples and different definitions of
audit fees. Simunic gathered his data by survey, while Simon’s sample was based on
voluntary disclosures of audit fees. Simunic used the sum of audit fees and the cost
of the internal audit function; Simon used the audit fee only.

Palmrose [1986] found that Price Waterhouse was able to charge a premium
in the office equipment industry. Price Waterhouse had the largest market share in
this industry, suggesting the existence of a fee premium for industry experience.

More recent fee studies have generally found a fee premium for the Big 6, at
least for smaller publicly-traded companies (e. g., Palmrose [1986]; Francis and Simon
[1987]). Audit fee studies in the Australian market by Francis [1984] and Francis
and Stokes [1986] are also consistent with the existence of a fee premium for the Big
6. In a study of auditor changes, Ettredge and Greenberg [1990] found that first-year

reduction in fee was smaller with the change from a non-Big 6 to Big 6 auditor. This

result is also consistent with a fee premium for the Big 6.

 

 

 

26

. There is also evidence that the Big 6 command a fee premium in the
governmental sector. In a study’of audit fees for large US. cities, Rubin [1988]
found a fee premium for the Big 6. However, the fee premium was only significant
(.10 level) for the largest cities. Copley [1990] also found a fee premium for the Big
6 in a study that largely replicated Rubin.

Fee research in the governmental sector suggests that size-related fee
premiums extend beyond the Big 6. Deis and Giroux [1992b] found that audit fees
for Texas school districts are positively related to the size of a local firm measured
by the number of offices.

2.2.4 Economic Consequences of Auditor Choice

Two studies have examined the relationship between choice of auditor and
capital market activity. The results of these studies were directionally consistent with
quality-differentiation by the Big 6, but were not statistically significant.

In an early study, Nichols and Smith [1983] investigated the market reaction
when a company changed from a non-Big 6 to a Big 6 auditor. Although Nichols
and Smith found a positive reaction to the change to a Big 6 auditor, the reaction
was not statistically significant.

In a similar vain, Ettredge, Shane and Smith [1988] compared the association
between earnings forecast errors and security returns for companies audited by Big
6 firms versus companies audited by non-Big 6 firms. They expected a larger
association for clients of Big 6 firms on the premise that the information provided
by Big 6 firms is of higher quality. Their results were consistent with this hypothesis,

but were not statistically significant.

 

 

27
2.2.5 Summaxy of Audit Qualiy Research

Figure 2.2 is a summary of the results of empirical tests in which auditquality
is defined as membership in the Big 6. An extensive body of literature on audit
quality exists. Although the results of individual studies are not always significant,
the cumulative evidence in this area is overwhelmingly supportive that the Big 6
firms are quality-differentiated providers of audit services, at least in the market
segments studied. There does not appear to be a strong need for further empirical
work in this area unless it provides a test that discriminates among competing
theories.

The research in this area has dealt almost exclusively with publicly-traded
companies, a market dominated by the Big 6. Research in the governmental sector
has provided preliminary evidence that audit quality is a function of size for non-Big

6 firms.

 

 

 

 

28

 

Dimension of
Audit Quality

Results

Comment

 

Demand for Audit
Quality

Strong support

Publicly-traded companies
overwhelmingly prefer Big 6
auditors.

 

Determinants of
the Demand for
Audit Quality

Weak support

Explanatory power of models
with size related variable is high.
Relation to specific theories,
such as agency theory, is weak.

 

Audit Performance

Support

Limited evidence based on
litigation.

 

Many studies indicate the

 

 

Audit Fees Strong support presence of a fee premium for
the Big 6.

Economic Not supported Results directionally consistent,

Consequences ‘ but not statistically significant.

 

 

 

. Figure 2.2
Summary of Audit Quality Research

 

 

29

2.3 Audit Quality and IPOs

The research on auditor quality differentiation in IPOs mirrors the research
in non-IPO settings, except there have been no direct attempts to measure audit
quality. These empirical studies appeal either to the independence or information
quality theories in supporting an expectation that Big 6 auditors will be quality-
differentiated. Auditor quality is then inferred from the study of auditor selection,
auditor fees, and the economic impact of auditor selection on offering underpricing.

IPOs have received considerable research attention because of the relationship
between the information environment and the level of underpricing of the offering.
Underpricing refers to the fact that the first return for newly traded stocks is on
average, positive. The initial return is defined as (CP-OP)/OP, where CP is the
closing (bid) price on the first day of trading, and OP is the offering price. Ritter
[1991] reports that the initial returns on IPOs for the period 1960-87 was 16.4
percent.7

Underpricing models developed by Rock [1986] and Beatty and Ritter [1986]
explain underpricing as a function of information asymmetry between informed and
uninformed investors. Underpricing is necessary to maintain the interest of the
uninformed investors. The presence of rationing of shares in the IPO market
prevents uninformed investors from realizing the average return on IPOs. Rationing
of shares is greatest for the shares which increase the most in price, so that
uninformed investors are more likely to receive an allocation of overpriced shares.
Without the presence of underpricing, uninformed investors would suffer losses on

their IPO investments, and drOp out of the market. The level of underpricing is

 

30

related to the level of ex ante uncertainty about the valuation of the issue, which
reflects the degree of information asymmetry. Beatty and Ritter [1986] demonstrate
that this pricing equilibrium is enforced by the reputation of the offering investment
banker.

In a setting where valuation is related to the ex ante uncertainty regarding the
value of the firm, there is a natural role for the auditor in reducing information risk.
A high quality auditor can reduce the ex ante uncertainty of the offering by providing
more precise information concerning firm value.

Initial public offerings also provide an environment where signaling is
important. Leland and Pyle [1977] deveIOped a signaling model in which the
entrepreneur’s private information about future cash flows of the firm is revealed by
the percentage of ownership offered in the IPO. In the Titman and Trueman [1986]
model, the choice of auditor is also a signal as to the owner’s private information.
However, the difference in information quality between high and low quality auditors
prevents low-quality companies from mimicking the signal by hiring a high quality
auditor. . That is, a high quality auditor will be more likely to reveal the
entrepreneur’s unfavorable information.

2.3.1 Auditor Selection in IPOs

Empirical studies which address the issue of auditor quality differentiation and
auditor selection in IPOs include Simunic and Stein [1987] and Menon and Williams
[1991]. The empirical tests in Menon and Williams are theoretically founded upon
the quality differentiation theory of Titman and Trueman [1986], and DeAngelo

[1981]. Simunic and Stein develop their own model of differentiation, but the

 

 

 

31

implications of their model are not substantively different from the DeAngelo [1981]
and Titman and Truman [1986] models. I

Simunic and Stein [1987] differentiate audit quality along three dimensions:
control, credibility and product line. Control relates to how the audit firm
contributes to organization control. Credibility relates to how the audit serviCe is
perceived by outside users.8 Product line refers to the breadth of services offered
by the CPA firm.

Simunic and Stein test for product differentiation using initial public stock
offerings that occurred in 1981. In their sample they distinguish between Big 6,
second tier and local firms, but delete the second tier from their tests. Because
client size may also affect auditor choice, they include control variables for total
assets, geographic dispersion, and whether the financial statements are consolidated.
All three control variables are significant in a logistic regression of the choice
between a Big 6 and local auditor.

Simunic and Stein model the demand for auditor credibility as an increasing
function of leverage, the reputation of the investment banker, and the use of a firm
commitment offering.9 Consistent with their hypotheses, choice of a Big 6 auditor
is positively associated with firm commitment offerings and the use of a ranked
investment banker. Use of a Big 6 auditor is negatively correlated with leverage,
contrary to their predictions. They interpret their results as consistent with higher
auditor credibility for the Big 6. I

Menon and Williams [1991] followed an approach similar to that found in

Simunic and Stein, except they examine auditor changes in the two year period prior

 

 

 

32

. to an IPO. They performed a logistic regression in which the dependent variable is
'3 whether a company retained its local firm or switched to a Big 6 firm. Consistent
with the results in Simunic and Stein, the decision to replace a local auditor was
associated with the type of offering, reputation of the investment banker, and size
of the IPO company.

2.3.2 IPO Audit Fees

Beatty [1989] used the residuals from a regression of total cash compensation
paid in the IPO as a proxy for auditor reputation. The total cash compensation in
the IPO served as a proxy for audit fees. Total cash compensation includes auditor
fees, legal fees, and miscellaneous expenses. The use of audit fees as a proxy for
auditor reputation is based on the hypothesis that higher quality auditors receive a
fee premium.

Estimated audit fees were available for part of Beatty’s sample; A the
correlation between audit fees and total cash compensation was .77. Beatty
interpreted the high degree of correlation as indication that total cash compensation
is a good proxy for audit fees.

Beatty found that the average fee residuals for the Big 6 are positive, and the
average fee residuals for the second tier and local firms are negative. His most
notable finding is the existence of significant differences in the fee residuals among
the Big 6 firms. This finding is important because it challenges the assumption of
uniform quality among the ’Big 6. If this finding is correct, it suggests that other
meaningful differences in audit quality may exist among the Big 6. However, caution

is warranted in interpreting this result. The lower fee residuals of certain members

 

 

33

of the Big .6 are based upon tests which , group these firms with the largest firms in
the second tier. Empirical research on audit fees by Francis and Simon [1989]
suggests that the second tier are more like local firms than the Big 6.

2.3.3 Offering Undexpricing

Use of a high quality auditor is expected to be associated with lower
underpricing of the offering through the reduction of information risk, or by signaling
the entrepreneur’s private information. Balvers, McDonald and Miller [1988]
developed a model in which both investment banker and auditor reputation serve to
reduce underpricing. An important feature of their model is the presence of an
interaction term between investment banker and auditor reputation. The interaction
term indicates that the effect of investment banker and auditor reputation is reduced
when the value of both variables is high. Omission of the interaction term would
tendto obscure the relationship between investment banker and. auditor reputation
and, underpricing.

Balvers et a1. obtained results consistent with their model. The coefficient for
auditor reputation and investment banker reputation was significant and associated
with lower underpricing. The coefficient for the auditor-investment banker
interaction term was also significant and associated with higher underpricing.

Beatty [1989] used the residuals from the model of IPO cash compensation
described in the previous section as a surrogate for auditor reputation. The fee

residual proxy for auditor reputation was significantly associated with lower

underpricing.

34

Beatty’s approach is unique because of the way the audit fee residual is used
in the underpricing model. The fee residual. is allowed to vary across each
observation, including IPO observations involving the same audit firm. The usual Big
6/non-Big 6 classification treats quality as a global characteristic of the firm.
Allowing the reputation variable to vary across observations within the same firm is
consistent with quality varying within the firm.

It should be noted that Beatty did not control for auditor size using the usual
Big 6/non-Big 6 classification. Since fee residuals were positive for the Big 6 and
negative for other firms, the significant relationship between underpricing and the
fee residuals may be capturing differences between Big 6 and non-Big 6 firms.10
2.3.4 Summa of Audit uali Research'in IPOs

Figure 2.3 is a comparison of the results of studies of audit quality in an IPO
setting with the results of studies in non-IPO settings. The research on audit quality
in IPOs is generally consistent with the findings in non-IPO settings that the Big 6
provide quality-differentiated services. The IPO research is limited to auditor
selection, economic consequences (offering underpricing), and auditor compensation.
This researcher is unaware of any direct research into audit quality in IPO settings.
In addition, the results in Beatty [1989] provide some indication of differences in

reputational quality among the Big 6.

 

35

 

 

 

 

 

 

 

 

 

 

 

Dimension of Results Results Comment on IPO
Audit Quality Non-IPO Studies IPO Studies Studies
Demand for Strong support Strong support IPO companies
Audit Quality prefer Big 6 auditors
Big 6 auditors
Determinants Weak support Moderate preferred in firm
of the Demand support commitment
for Audit offerings, and with
Quality use of ranked
investment banker
Audit Support No direct No direct evidence
Performance evidence
Audit Fees Strong support Supported Limited evidence
Economic Not supported Supported Underpricing lower
Consequences with high quality
auditor.
Figure 2.3

Comparison of IPO and non-IPO

Audit Quality Research

 

 

36
2.4 Industry Experience As a Dimension of Audit Quality .

Knowledge of the client’s industry is an important element of the planning
and performance of the audit (AICPA [1991], AU311.06). Specific references to
industry experience are included in the quality models of DeAngelo [1981] and
Titman and Trueman [1986]. DeAngelo suggests that one reason companies may
change to a Big 6 auditor at the time of an IPO is their greater industry-specific
knowledge in comparison to smaller CPA firms (p. 185). Titman and Trueman note
that many IPO companies are start-up operations, and suggest "The auditor provides
such firms with expertise in information processing as well as knowledge of industry
conditions" (p. 162).

There are considerable differences in the client distributions between the Big
6 firms across industries (Rhode, Whitsell and Kelsey [1974], Eichenseher and Danos
[1981]). Industry concentration is related to regulation of the industry (Danos and
Eichenseher [1986]). Danos and Eichenseher suggest that this relationship is due to
economies of scale. However, it is also plausible that regulation leads to differential
demand for audit quality in the form of auditor industry experience.

Evidence linking auditor experience and audit quality is scant, although more
recent research has addressed the role of auditor experience. Perhaps the most
important is a study by Deis and Giroux [1992a]. Deis and Giroux found that audit
quality for CPA firms which audited Texas school districts was increasing with the
number of school districts audited. Although Deis and Giroux, interpret this as
consistent with the DeAngelo independence theory, it indicates that industry

experience is an element of audit quality as evidenced by actual audit performance.

 

 

37

In a follow-up study, Deis and Giroux [1992b] examined whether audit quality
and industry experience were determinants of audit fees and audit hours for Texas
school districts. Of particular interest is the variable for industry specialization
measured by the number of school district clients. This variable had a significant
negative relationship with the number of audit hours. Industry experience was
negatively associated with audit fees as well, although the t-statistic of 1.51 was not
significant based on a two-tail test. Deis and Giroux interpreted their results as
indicating the presence of efficiency due to industry specialization.

Although the industry experience variable in Deis and Giroux [1992b] was
negatively associated with audit fees, it is important to note that part of the influence
of industry experience on audit fees [is indirect through its effect on audit quality.
Deis and Giroux also find that audit fees and audit hours are positively associated
with audit quality (more precisely, they find that audit hours and audit fees are
negatively associated with lack of quality) and the size of the firm. Firm size and
industry experience were associated with higher audit quality in their previous study.

Ward, Elder and Kattelus [1993] find the Opposite relationship between audit
fees and industry experience. They found that one firm with the largest share of the
municipal audit market in Michigan received a significant fee premium. One
possible explanation for the difference in results in the two studies is that the
continuous measure of experience in Deis and Giroux captures economies of scale,

while the dichotomous measure of auditor experience in Ward et al. captures brand

name effects.

 

 

38
The studies by Deis and Giroux [1992b] and Ward, Elder and Kattelus [1993]

relate to the governmental sector. Ettredge and Greenberg [1990] found a similar
relationship in a study of fee .cuts at the time of an auditor change for commercial
entities. The fee cut was larger with the change to a more experienced auditor using
a continuous sales-based measure of experience. This is consistent with greater audit
efficiency associated with increasing auditor experience. However, the fee cut was
smaller if the change was to an auditor cited by the company as having greater
industry expertise. The result for this dichotomous variable is consistent with a fee
premium for industry experience.

Palmrose [1986] included an indicator variable for auditor experience in her
study of audit fees in the private sector. The variable was positive, but not
significant in her model. As discussed in section 2.2.3, Palmrose found some
evidence of fee premiums in specific industries.

The research on the relation of auditor experience to audit quality has been
limited, but is rapidly growing. Except for Deis and Giroux [1992a], this evidence
has related to audit fees, with conflicting results. Palmrose’s inability to find a fee
premium for industry experience may reflect the effects of using a broad—based
sample. Both the Deis and Giroux, and Ward et al. studies are examples of studies
which use a specific industry. Another alternative to use of a specific industry setting
is to examine audit fees in a specific context, such as IPOs.

As in previous'auditor experience research, this study tests the hypothesis that
auditor industry experience is a dimension of audit quality. The current study differs

from previous research in several ways. First, an attempt is made to link the

 

 

 

 

39

prediction that auditor experience is a dimension of audit quality to the information
quality theory. SecOnd, this study considers multiple experience measures, where
previous research has considered singular measures of experience. Third, the current
study also addresses the demand for auditor experience, and the economic
consequences associated with selection of an experienced auditor. These dimensions
of audit quality have not been addressed in previous research on auditor industry
experience. Finally, this study investigates the role of auditor experience in IPOs,
a unique private sector setting. Previous research on the relationship of audit fees
and auditor industry experience in the private sector have used more general settings,
with conflicting results.

In Chapter Three, specific hypotheses are developed as to the relationship
between auditor industry experience and auditor selection, audit fees, and IPO
underpricing. The experience measures used in the study are also described in

Chapter Three.

 

 

 

 

10.

4o
ENDNOTES

’ The Datar, Feltham and Hughes (1991) model can be considered to be somewhat

of a hybrid of the independence and informational quality theories since high quality
auditors are assumed to be both more discriminating (information quality) and more
resistant to client pressure (independence).

Several researchers have also attempted to identify audit firm characteristics
associated with the decision to change auditors (e.g., Eichenseher and Shields
[1983]), and audit quality (e.g., Schroeder, Solomon and Vickrey [1986]). These firm
characteristics include firm-wide and audit team specific characteristics.

See Watts and Zimmerman [1990] for a review of the agency literature in accounting.
See Francis and Wilson [1988] for a review of these studies.

The term "brand name" is somewhat misleading in the sense that brand names are
generally thought of as attaching to individual firms. In Francis and Wilson, brand
name is indicated by belonging to a class of firms (the Big 6).

It is important to point out that Simunic reported results using the audit fee scaled
by total assets, and the audit fee plus salaries paid to internal auditors scaled by total
assets. The coefficient for the Big 6 (excluding Price Waterhouse) was negative in
each model, but was only statistically significant in the model using audit fees alone.

Most of Simunic’s analysis is based on audit fees plus internal audit salaries.

Ritter [1991] indicates that underpricing is a short-term phenomenon. Over a longer
period spanning three years following the offering, IPOs underperform the market.

Both Simunic and Stein (1987) and Menon and Williams (1991) use the term auditor
credibility to refer to how audit quality is perceived by financial statement users.
This is essentially the same as audit quality as used in this study.

In a firm commitment offering, the underwriter guarantees the sale of the specified
number of shares at the specified price. In a best efforts offering, the underwriter
attempts to sell as many shares as possible, subject to minimum and maximum
numbers of shares.

Copley [1991] uses fee residuals as a proxy for audit quality in a model of financial
statement disclosure quality by municipalities. The audit fee residual had greater
explanatory power than a Big 6/nOn-Big 6 dummyvariable in one model, but not in
another model. '

 

 

CHAPTER 3 - DEVELOPMENT OF RESEARCH HYPOTHESES,
EXPERIENCE MEASURES AND SAMPLE DESCRIPTION

In the previous chapter, existing research was described that linked audit
quality with auditor selection, audit performance, audit fees, and the economic
consequences of auditor choice. Preliminary evidence in the governmental sector has
provided evidence that auditor industry experience is a dimension of audit quality.
This study examines whether auditor industry experience is a dimension of auditor
quality for IPO companies. Specifically, the relationship between auditor experience
and auditor selection, audit fees, and the underpricing of the offering for IPO
companies that choose a Big 6 auditor is examined. Although direct measurement
of audit performance in an IPO setting is a potentially important research issue, it
is beyond the scope of this study.

The specific research hypotheses about the relationship between auditor
experience and auditor selection, offering underpricing, and auditor compensation
are developed in the following section. The measures of auditor experience are

described in Section 3.2.

3.1 Development of Research Hypotheses
3.1.1 Demand for Auditor Experience

Menon and Williams [1991] found an increased demand for Big 6 audit firms
over local audit firms by investment bankers and their clients at the time of an IPO,

which they interpreted as a demand for increased auditor credibility. If auditor

 

 

42

industry experience is a relevant dimension of auditor quality, auditor experience will

also be demanded by IPO companies.

H1: IPO companies will have a positive demand for auditor industry
experience.
Hypothesis 1 predicts that IPO companies will display a preference for
experienced auditors. An explanatory model of the demand for auditor experience

by IPO companies is develOped in the following section.

3.1.2 Determinants of the Demand for Industfl Experience

Models by Rock [1986] and Beatty and Ritter [1986] describe IPO
underpricing as increasing with the information risk of the offering. Consistent with
these theories, the demand for auditor experience is modeled as a function of the
information risk of the IPO firm. Firms for which less information exists are
perceived as being more risky due to greater uncertainty about future returns
(Clarkson and Thompson [1990]). Datar, Feltham and Hughes [1991] predict a
positive relationship between IPO company risk and auditor quality after controlling
for the percentage of ownership offered.

The empirical proxies used in this study for information risk are IPO company
size, IPO company age, and the industry of the IPO firm. The demand for auditor
experience is predicted to be negatively associated with IPO company size and age.

Auditor experience is predicted to be positively associated with IPO company

membership in a specialized industry.

 

 

 

43

Greater information exists. for large publicly-traded firms than for small
publicly-trade firms (Freeman [1987]). Consistent with the information environment
for publicly-traded firms, less information is expected to exist for smaller IPO
companies, leading to an increase in information risk. Accordingly, the demand for
experience is predicted to be negatively associated with the size of the company.1

HZA: The industry experience of the auditor selected by the IPO

company will be negatively associated with the size of the IPO
company.

Hypothesis 2A predicts that smaller companies will demand greater levels of
auditor industry experience due to the information environment. Similarly, less
information is expected to exist for companies with shorter operating histories.
Beatty [1989] argued that the amount of information for IPO companies is partially
a function of company age, and he found that underpricing was higher for firms with
shorter operating histories. Accordingly, the demand for auditor experience is
expected to be negatively associated with the age of the IPO company.

H23: The industry experience of the auditor selected by the IPO company

will be negatively associated with the age of the IPO company.

Hypotheses 2A and 2B predict that the demand for auditor experience will
be greater when less information is available about the IPO company. The demand
for auditor experience is also expected to be greater when the industry information

environment is complex. An indicator variable for the existence of industry-specific

 

 

 

 

 

 

44

accounting standards or an industry audit guide is used as a proxy measure of
industry accounting complexity.

HZC: The industry experience of the auditor selected by the IPO
company will be positively associated with IPO company
membership in a specialized industry.

Hypothesis 2C predicts that the demand for auditor experience will be greater
when the IPO company belongs to a specialized industry. In addition to company
size, age, and membership in a specialized industry, the auditor selection models in
Simunic and Stein [1987] and Menon and Williams [1991] suggest several additional
variables associated with the auditor selection decision. The full specification of the
model, including additional control variables, is included in Chapter 4.

3.1.3 Auditor Experience and Offering Underpricing

As described in section 2.1, Titman and Trueman [1986] have developed a
model in which the use of a high quality auditor can decrease offering underpricing
through the reduction of information risk. Previous research by Balvers, McDonald
and Miller [1988] and Beatty [1989] have demonstrated thatunderpricing is lower for
offerings with a Big 6 auditor. If auditor experience is a dimension of audit quality,
greater levels of auditor industry experience will be associated with lower
underpricing of the offering.

H3: The underpricing of the IPO will be negatively associated with
the industry experience of the auditor.

 

 

 

 

 

45

In addition toauditor experience, the underpricing models in Beatty [1989]
and Balvers, McDonald and Miller [1988] have identified several additional factors
associated with the underpricing of IPOs. The complete underpricing model,
including additional control variables, is described in Chapter 5.

3.1.4 Auditor Experience and Auditor Compensation

In section 2.2.3, several existing studies were cited in which Big 6 audit firms
received a fee premium. The maintained hypothesis in these studies is that a fee is
consistent with higher audit quality for Big 6 firms. If auditor industry experience
is a relevant dimension of audit quality, it will also be associated with a premium in
the audit fee.

H4: Auditor industry experience will be associated with higher audit

fees.

Hypothesis four predicts that experienced auditors will receive a fee premium.
The audit fee is especially important because it affects the cost of choosing a quality
auditor. The model for auditor compensation is based on Beatty [1989] and is
described in Chapter 6.

A summary of the research hypotheses is contained in Figure 3.1.

 

 

 

 

 

46

 

Dimension of audit
quality

Research hypothesis

 

 

Audit demand

H1:

IPO companies will have a positive demand
for auditor industry experience.

 

Determinants of
demand for auditor
experience

The industry experience of the auditor
selected by the IPO company will be
negatively associated with the size of the
IPO company.

The industry experience of the auditor
selected by the IPO company will be
negatively associated with the age of the
IPO company.

The industry experience of the auditor
selected by the IPO company will be
positively associated with IPO company
membership in a specialized industry.

 

IPO underpricing

The underpricing of the IPO will be
negatively associated with the industry
experience of the auditor.

 

 

Auditor compensation

 

Auditor industry experience will be
associated with higher audit fees for the
audit of IPO companies.

 

 

Figure 3.1

Summary of Research Hypotheses

 

 

 

3.2 Selection and Measurement of Experience Variables
3.2.1 Selection of Experience Measures

In the development of the research hypotheses, auditor industry experience
was referred to as if it were a unidimensional construct. However, there are many
potential measures of auditor industry experience. Relevant dimensions of auditor
experience include (1) the use of sales-based or client-based measures of experience,
(2) the number of digits in the SIC codes used to measure experience, and (3) the
use of continuous or dichotomous measures of experience.

This researcher is unaware of any theory to suggest whether experience based
on sales dollars audited or number of clients audited is more relevant. Previous
studies of auditor choice have referred to industry experience or expertise in a
general sense (e.g., Eichenseher and Shields [1983], Shields [1984]). Studies that
directly incorporate experience measures have-used sales-based measures (Palmrose
[1986]; Ettredge and Greenberg [1990]) and client—based measures (Deis and Giroux
[1992a]). Studies of industry concentration (e.g., Eichenseher and Danos [1981])
have used client, sales and fee based measures.

_ This researcher believes that industry experience is best captured by client-
based measures. This belief is based on the notion that experience is best gained by
experience in many different settings. However, because no theory as to the relevant
dimension of auditor experience exists, the empirical tests in this study incorporate
both client-based and sales-based measures.

There are specific trade-offs associated with use of two-digit and three-digit

SIC codes to measure experience. Three-digit SIC data captures more specific

 

48

industry experience. Industries at the three-digit level probably are more
representative of what is thought of by an "industry". Danos and Eichenseher [1982]
have posited that three-digit codes are most consistent with thein-house training of
the largest CPA firms. Danos and Eichenseher [1986] have further argued that
three-digit classifications conform with the classification of audit markets in most
research. Most empirical studies which incorporate auditor experience measures
have performed analysis at the three-digit level (e. g., Danos and Eichenseher [1982];
Palmrose [1986]), and. even at the four digit-level (Ettredge and Greenberg [1990]).
Accordingly, the primary research focus in this study is on three-digit experience
measures.

Industriesfor empirical tests were limited to those with more than 25
companies. This requirement was imposed to ensure that the measures of
experience were meaningful. The minimum industry size requirement resulted in the
loss of a substantial number of observations for the empirical tests. To minimize
data loss, empirical tests were also conducted using broader two-digit experience
measures.

Continuous measures of experience allow for greater variation in the
experience measures. However, dichotomous measures may provide greater power
by capturing the perception of expertise. That is, a firm is either perceived as being
experienced in an industry, or it is not. Ettredge and Greenberg [1990] and Deis and
Giroux [1992a] used only continuous measures; Palmrose [1986] used both

dichotomous and continuous measures. Dichotomous measures have a ready

 

 

49

economic interpretation, and also can be readily compared to results of tests in which
membership in the Big 6 is the measure of audit quality.

Based on the previous discussion, this researcher prefers dichotomous
measures based on number of clients at the three-digit SIC level. However, no
theory as to how to measure firm experience exists, and alternative measures of
auditor experience can be readily constructed. Accordingly, this study tests multiple
measures of industry experience that include sales-based measures, continuous
measures, and measures based on two-digit SIC codes.

Figure 3.2 summarizes the experience measures used in this study. The use
of multiple experience measures increases the risk of finding a spurious relationship
between auditor experience and auditor selection, IPO underpricing and audit fees.
However, the use of multiple measures will also provide insights into the relationship

between these various measures, and the trade-offs associated with their use.

 

 

50

Experience Measure Definition

Continuous measures:

CLSHARE Individual Big 6 firm share of the market for that industry
segment based on number of clients audited defined as:

CLS- = Market share for firm i in industry j.
C Number of clients for firm i in industry j.

SASHARE Individual Big 6 firm share of the market for that industry
segment based on sales of clients audited defined as:
SASij = 2k Rijk
22D; Rrjk
SASfi = Market share for firm i in industry j based on sales
audued.
Rijk = Sales revenue of client k of firm i in industry j.

Dichotomous measures:

CLEXP 1 if the Big 6 firm is the industry client market leader
0 otherwise
SAEXP 1 if the Big 6 firm is the industry sales market leader
0 otherwise
LGEXP 1 if Big 6 firm audits the largest company in the industry

based on total sales
0 otherwise

The number (2) or (3) after each of the above variables indicates measurement at the two-
digit and three-digit SIC level, respectively.

Figure 3.2
Summary of Experience Measures

51

3.2.2 Audit Firm Experience Measures

Measures of individUal audit firm industry experience were determined using
the Disclosure, Inc. Compact Disclosure SEC (CD) database. This source was
chosen for the experience measures because it includes the full population of
publicly-traded companies.

The primary SIC code, revenues, and auditor name were obtained for each
company in the CD database. Foreign firm and companies with zero sales were
excluded. The data was sorted by SIC code and sorted within SIC code by auditor
to calculate individual audit firm market shares for each industry. Industries were
discarded if they contained less than 25 companies. This was done to limit the
analysis to those industries of sufficient size such that industry experience was a
meaningful measure. Industry measures were deveIOped at both the two-digit and
three-digit SIC code level.

The measures described in Figure 3.1 were initially calculated using the
January 1988 version of the CD database. This version of the database was selected
because it corresponds with the beginning of the period for which IPO data was
collected. A summary of the continuous measures is contained in Appendix A.
Appendix A.1 contains the measure CLSHARE3, and Appendix A.2 contains the
measure SASHARE3. Appendix A3 and A4 contain the measures CLSHARE2 and
SASHAREZ, respectively.

The tables in the appendix supp6rt the existence of substantial variation in
market share within industries. For example, for the measure CLSHARE3 in

Appendix A.l, a chi-squared test rejects at the 5 percent significance level the

 

 

 

_fi
52

hypothesis that the number of clients is uniformly distributed among the eight audit
firms that previously made up the Big 6 for 27 of the 87 industries.2

Many of the industries with substantial concentration levels are specialized
industries. Industries were classified as being specialized if separate accounting rules
or an industry audit guide exists for that industry. Using this procedure, 21
industries were classified as specialized. The chi-square test of industry
concentration based on number of clients was significant for 13 of these 21
industries. As expected, a binomial test rejects at the .01 significance level the null
hypothesis that the distribution of audit clients is unrelated to classification of the
industry as specialized. This result is consistent with the finding in Eichenseher and
Danos [1981] that industry concentration is related to industry regulation, since many
of the specialized industries would also be considered to be regulated industries.

Several aspects of the experience data introduce noise that may mitigate
against finding a significant relationship between experience and auditor selection.
The use of primary SIC codes ignores the other industries in which a company may
operate. However, it can be argued that except for a few conglomerates, most
companies are identified with the industry of their primary SIC code. Similarly, the
sales figures include sales from other industry segments.

During the period under study, two mergers occurred among the largest CPA

 

firms, reducing the Big 8 to the Big 6. Appendix B contains a summary of the
experience measures subsequent to these mergers. This data was calculated using

the January 1990 version of the CDdatabase.3

 

53

In the following section, the sample for performing empirical tests is

described. Included in this section is a description of the correlations between the

various experience measures used in this study.

3.3 Initial Public Offering Sample Data

3.3.1 Sample Descrigfion

 

Firm commitment offerings that took place during the period 1988 through
1991 were selected for study. The most recent period for which data was available
was chosen for study to facilitate data collection. Four years of data was expected
to generate a sufficient sample size. Because of the stock market crash of October,
1987, it was considered desirable not to include this year in the study. The test
period spanned the period of two large mergers in the public accounting profession.
Calculation—of experience measures and performance of empirical tests in the post-
merger period may provide additional insights into the effect of these mergers on the
audit market and firm reputations.

Two primary data sources were used in this study. Offering data for the
period 1988-1989 was collected from Going Public: The IPO Reporter (Going
__P_ul_)l_i<_:). Data for offerings which occurred in 1990 and 1991 was collected primarily
from the Disclosure, Inc. Compact Disclosure D’33 (D33) database. Some additional
observations were added from Going Public for 1990 and 1991. The D33 database
begins in 1990 and is not available for the earlier period. The D33 database includes

a more extensive set of variables than Going Public, including measures of auditor

 

compensation. Primary information collected from GoingPublic and D33 includes

 

 

 

54

the offering date, offering proceeds, offer price, auditor, auditor compensation, and
investment banker. The pricing data necessary to compute underpricing was
collected from the DailyStoglg Price Record. The age of the company was collect
from Standard and Poor’s Comorate Directory.

All offerings for the period 1988-1991 in Goingjublic and the D33 database
were initially selected for testing, excluding IPOs for mutual funds and investment
trusts. The sample includes unit offerings, in which shares and warrants to purchase
additional shares are priced as a unit. Pricing tests for these offerings were done on
a unit basis.

Going Public contains only firm commitment offerings. Best efforts offerings
were excluded from the sample because of the difficulty associated with obtaining
this data. Best efforts offerings are generally smaller and more risky, and less likely
to use a Big 6 auditor. Accordingly, inclusion of best efforts offerings is less
important for this study which examines auditor experience within the Big 6.

Table 3.1 is a summary of the number of offerings by year and auditor type
for offerings for which the auditor could be identified. Consistent with previous
research, most offerings involve Big 6 auditors. The primary focus of this study is
the role of auditor experience within the Big 6. However, the preliminary data
analyses include offerings with second tier and local audit firms as a manipulation

check, and as a partial replication and comparison to previous research.

 

 

55

Table 3.1 .
Number of Offerings By
Year and Auditor Type

 

 

 

 

1988 1989 1990 1991 Total
Big 6 140 128 127 319 714
Second Tier 16 10 16 16 58
Local 10 12 14 33 69
Total 166 150 157 368 841

 

Table 3.2 contains descriptive statistics for the sample, including offering
proceeds, offer price and underpricing. Consistent with previous research (e.g.,
Balvers, McDonald and Miller [1988]), offerings with Big 6 auditors are much larger,

are less underpriced, and are more likely to involve a ranked investment banker.

 

56

7 Table 3.2
Sample Descriptive Statistics

 

 

Big 6

Second

tier

Local

All

Offerings

Mean (standard deviation)

 

 

Percent
Number (Millions) with
of Offering Offer Under- Ranked
Offerings Proceeds Price Pricing Investment
Banker
714 42.56 10.87 .119 .458
(82.40) (5.42) (.208)
58 15.94 7.31 .198 .190
(23.38) (3.83) (.319)
69 9.54 6.38 .188 .058
(21.78) (4.56) (.277)
841 38.01 10.26 .130 .407
(77.17) (5.46) (.225)

 

 

 

 

 

57

3.3.2 Experience Measure Correlations

In the preceding section, the sample used for conducting empirical tests was
described. Of the 714 offerings with a Big 6 auditor in Table 3.1, a three-digit (two-
digit) experience measure could be computed for 489 (662) of the offerings. Table
3.3 contains a correlation matrix of these variables for the IPO firms included in the
study.4 The SIC codes for the IPO companies were identified from the CD
database.

Several relationships are apparent from Table 3.3. As expected, the highest
correlations are between continuous and dichotomous variables within measures
(sales or clients) at the same SIC code level (two or three digit), followed closely by
the same measures across SIC code level. Thelowest correlations are between the
sales and client measures. The lower correlations between client and sales measures
suggest that they potentially contain differential information as to auditor experience.
These results are also consist with Eichenseher and Danos [1981] finding that
industry concentrations vary substantially, dependent upon the concentration measure
utilized. The correlations also suggest that two-digit measures are relatively good
surrogates for three-digit measures.

The auditor experience measures will be incorporated into the models of
auditor selection, offering underpricing and IPO cash compensation as tests of the
research hypotheses. Empirical tests of auditor selection for the sample are reported
in Chapter 4. The tests of IPO underpricing are included in Chapter 5, and tests of

auditor compensation are contained in Chapter 6.

 

 

 

58

 

29/2 EnormoEoQ we Eeomewm “O: E £033 £885 can macaw:
5953 noun—280 of 8% 585 £58 38-033 65. S. 05 :w Eeanwa o8 macaw—oboe =<

 

.62.: opoU UHm théé of 8 83238 3:0 wEZoZ: $565000 E8 New M Z
._o>o_ £50 0.6 :w.€-v85 05 am one 8382: 2: we 05 Ho 25 p.823 $565000 How owe N Z
2: am. 8. .8. MN. Nm. mm. 8. R. E. N963
of mm. P. mm. 2.. 8. 8.. 3.. cm. Neaam
2: Q. S. :. S. S. mm. em. mass
2: 9.. om. 8. 8. as. 3.. Nana
2: .2. R. mm. mm. me. $20
03 S. S. me. 2. 2.63
2: am. am. :.. mesa
2: am. 2. E56
2: 5.. 2.8
2: 3.5

 

N53 Nessa. mess gem 35 R53 macaw mesa team 35

 

 

 

moi—mam: wouotomxm .3 £32 :owflohoo
an. «Ewe

 

 

 

 

 

59
ENDNOTES

Smaller companies are less likely to have in-house expertise available for the
production of information necessary for the offering. The operations of smaller
firms may also receive greater benefit from the industry knowledge of the auditor.

The chi-square test is known to be unreliable when cell frequencies are less than 5
(Hoel, Port and Stone 1971; p. 5). When the analysis is restricted to industries
where the Big 6 audit more than 40 companies, the chi-square test is significant for
18 of 40 industries.

Although a uniform distribution assumption is consistent with homogeneity among
the Big 6, the largest CPA firms do differ in size. When the expected industry
distribution is based on overall market shares of each firm, the chi-square test is
significant for 24 of the 87 industries.

This data could be created by adding the market shares of the firms involved.
Because the mergers occurred approximately at the midpoint of the period under
study, the experience measures were recomputed. The mergers may precipitate some
auditor changes due to conflicts with client competitors. It is unlikely, however, that
many of these changes were reflected in the January 1990 CD database.

It is possible to develop a correlation matrix for the experience measures for the full
population of publicly-traded companies. However, the development of such
measures is not the primary focus of this study.

 

 

 

7” a

CHAPTER 4 - DEMAND FOR AUDITOR EXPERIENCE RESULTS

4.1 Introduction

 

This research addresses the role of Big 6 audit firm industry experience in the
initial public offering market. In Chapter 3, specific hypotheses were developed as
to (1) the demand for auditor industry experience by IPO companies, (2) the

determinants of the demand for experience by IPO companies, (3) the relationship

 

of auditor experience to offering underpricing, and (4) the relationship of auditor
experience to auditor compensation.

In this chapter, the hypotheses related to the demand for auditor experience
are empirically tested for a sample of initial public offerings that occurred during the
period 1988 to 1991. The sample, variables and data collection process were
described in Chapter 3. The demand for auditor experience by IPO companies is
examined in Section 4.2. The determinants of the demand for auditor experience are

tested in section 4.3.

4.2 Demand for Auditor Experience by IPO Companies

 

The first research hypothesis predicted that IPO companies would have a

positive demand for auditor industry experience.

 

H1: IPO companieswill have a positive demand for auditor industry
experience.

60

 

 

61

Table 4.1 presents summary measures of auditor experience levels for the
IPOs involving a Big 6 auditor for which experience measures could be computed. ‘

These variables are defined in Figure 3.2 in Section 3.2.

 

 

 

 

 

 

Table 4.1
Mean Auditor Experience Levels
IPO Sample
(n=489) (n=662)
Three-digit Two-digit
SIC Level SIC Level
Std. Std.
Mean Dev. Mean Dev.
CLSHARE 14.50 7.27 13.40 5.43
SASHARE 17.83 16.04 16.31 11.83
CLEXP .28 .45 .26 .44
SAEXP .23 .42 . .20 .40
LGEXP .19 .39 .17 .37

 

 

 

The auditor experience levels can be interpreted as follows, using client share

 

at the three-digit SIC level as an example. For the average offering in the sample,

the auditor’s industry market share based on number of clients was 14.50 percent.

 

The variable CLEXP indicates that the auditor was the industry leader based on
client market share at the three-digit level for 28 percent of the offerings. The
experience measures at the two-digit and three-digit levels are quite similar, although

the values for the three-digit measures are higher.

 

 

 

   

 

62

. The raw numbers do not interpret whether IPO companies demand auditor.
industry experience. In conducting empirical tests of this hypothesis, baseline
prediction models must be, developed to assess whether IPO companies display a
preference for auditor experience. Three successively more stringent baseline models
are used in the empirical tests. The three models are the random selection model,
the market share weighted selection model, and the industry share weighted selection
model. The sample mean for each experience variable will be compared to the
expected mean implied under each of these models to assess whether there is a
positive demand for auditor industry experience by IPO companies.

These comparisons will be made for all of the computed experience measures.
However, because the selection decision for an IPO company involves the choice of
an auditor by an individual IPO company, client-based experience measures appear
to be the most appropriate basis of comparison. Client-based measures weight each
audit client equally, and are not sensitive to the presence of large audit clients.

The random selection model assumes that selection of any of the Big 6 firms
is equally likely. Geographic and other factors make this supposition untrue for an
individual company’s choice of auditor firm. However, such factors are assumed to
average out over many observations. The random selection model is consistent with
the theoretical and empirical literature that views the Big 6 as a homogenous group.

The market share weighted selection model recognizes that the Big 6 firms
are not equal in size. Using this model, the probability that a given Big 6 firm will
be chosen is equal to that firm’s total market share based on the comparison metric

used (sales or clients).

 

 

 

 

 

 

 

 

63

The third baseline model is the industry share weighted selection model.
Under this model, the probability of an audit firm being chosen is the firm’s market
share within the industry of the IPO company. This model recognizes that the
distributions of clients among the Big 6 is not uniform across industries (Eichenseher
and Danos [1981]). This model provides a stringent test of whether IPO companies
demand auditors with greater industry experience than non-IPO companies.

4.2.1 Random Selection Model Results

The first benchmark of comparison is a random selection model, in which the
choice of auditor is made randomly among the Big 6 audit firms. The following
example will serve to illustrate the computation of the benchmark. In industry SIC
code 363, Big 6 firms audit 88.9 percent of the market based on number of clients.
The expected auditor market share for an individual firm in this industry based on
six firms is 14.82 (88.9 + 6). Each sample observation has an expected value based
on the industry of the IPO company.1 The overall sample expected value is equal
the mean for all observations. For all dichotomous measures, the appropriate
expected value would be .167 (1 -:— 6).2

Calculation of expected auditor experience levels is complicated by changes
in the number of Big 6 audit firms due to mergers during the period. The offering
sample was divided into periods based on the number of audit firms in the Big 6.
The expected value for the entire period is equal to the weighted average of the

expected value in the subperiods.

 

 

 

 

64

Table 4.2
Comparison of IPO Sample Experience Measures
To Expected Values Under Random Selection

 

 

 

 

 

 

(n=489) ‘ (n=662)
Three-digit Two-digit
SIC Level SIC Level
Sample Exp. Sample Exp.
Mean Value T-Statfl) Mean Value T-Statfl)
CLSHARE 14.50 12.02 . 8.31** 13.40 11.71 9.02**
SASHARE 17.83 14.62 4.56** 16.31 14.64 3.91”
CLEXP .282 .153 6.39** .261 .152 6.40**
SAEXP .229 .153 404* * .196 .152 2.87 **
LGEXP .188 .153 2.03* .166 .152 0.98

 

 

(1) Paired sample t-test of whether sample mean is greater than expected value under
benchmark model. '
** (*) Significant at the .01 (.05) level based on one-tail test.

 

 

 

 

 

 

 

 

65

Table 4.2 compares the observed sample values for auditor experience to the
expected values based on a random selection model. The results of the paired
sample t-test indicate that the choice of Big 6 auditor is not done randomly. Rather,
IPO companies select auditors with relatively greater levels of experience. The levels
of experience using both sales-based and client-based measures are greater than
would be expected under random selection.3

4.2.2 Market Share Weighted Selection Model

The market share weighted selection model differs from the random selection
model in that it controls for differences in size among the Big 6 audit firms. The
market share weighted model assumes that Big 6 firms are selected in proportion to
their overall market shares. This model would be expected to be more descriptive
than the random selection model if the larger of the Big 6 have an advantage over
smaller Big 6 firms. For example, the smaller Big 6 firms may have fewer offices,
placing them at a competitive disadvantage in certain geographic regions.

Table 4.3 presents a comparison of the IPO sample experience measures to
their expected values if auditor selection was done randomly in proportion to each
firms overall share of the market. The expected value of the experience measures
increases when weighted by each audit firm’s overall market share compared to the
random selection model. However, the differences between the sample values and
expected values remain statistically significant based on a one-sample t-test.4 This
provides further evidence that auditor selection is not made randomly by IPO

companies.

 

 

 

 

 

66

_ Table 4.3
Comparison of IPO Sample Experience Measures
To Expected Values Under Random Selection
Weighted by Total Market Share

 

 

 

 

 

 

 

 

(n=489) (n=662)
Three~digit Two-digit
SIC Level SIC Level
Sample Exp. Sample Exp.
Mean Value T—StatKI) Mean Value T-Statfl)
CLSHARE 14.50 12.76 5.29** 13.40 12.49 4.31**
SASHARE 17.83 15.09 3.78** 16.31 14.97 2.91**
CLEXP .282 .206 3.80** .261 .197 3.76**
SAEXP .229 .163 3.47** .196 .164 2.13*
LGEXP .188 .156 1.78* .166 .153 .93
(1) One sample t—test of Whether sample mean is greater than expected value under

benchmark model.
** (*) Significant at the .01 (.05) level based on one-tail test.

 

 

 

67
4.2.3 Industpj Weghted Selection Model

The final benchmark of the demand for experience by IPO companies is the
random selection model weighted by industry market share. Under this model, the
likelihood that a given audit firm is chosen is equal to the firm’s market share in the
industry of the IPO company. This is a stringent benchmark since the distribution
of clients among CPA firms within industries displays substantial levels of
concentration. If the IPO sample mean exceeds the expected value under this
model, this would suggest that IPO companies have a demand that exceeds the
demand for experience by existing publicly-traded companies.

Table 4.4 presents a comparison of the IPO sample experience measures to
the expected values under the industry weighted selection model. The client-based
measures under the industry weighted model suggest that the demand for auditor
industry experience by IPO companies is not greater than the demand for experience
by publicly-traded companies. The sales-based measures indicate that the demand
for sales-based experience is significantly less that the demand for experience by
publicly-traded companies.

The sales-based measures are not likely an appropriate basis of comparison.
As pointed out by Eichenseher and Danos [1981], sales-based measures are
dominated by large-client relationships. This can be observed from the expected
values forthe dichotomous measures in Table 4.4. 28 percent of companies select
the auditor with the largest industry market share based on clients, but 43 percent
of the sales dollars are audited by the audit firm with the largest industry market

share based on sales dollars audited.

 

 

 

 

68

Table 4.4 ,
Comparison of- IPO Sample Experience Measures
To Expected Values Under Random Selection
Weighted by Industry Market Share .

 

 

 

 

 

 

(n=489) (n=662)
Three-digit Two—digit
SIC Level SIC Level
Sample Exp. Sample Exp.
Mean Value T-Statfl) Mean Value T-Statfl)
CLSHARE 14.50 14.66 -.48 13.40 13.59 -.90
SASHARE 17.83 28.44 -14.63* * 16.31 23.63 -15.91**
CLEXP .281 .284 -.15 .261 .253 .47
SAEXP .229 .434 -10.79* * .196 .372 -11.73**

 

(1) One sample t-test of whether sample mean is greater than expected value under

benchmark model.
** (*) Significant at the .01 (.05) level based on one-tail test.

 

 

 

L; "ta

 

69

It is not necessarily surprising that the demand for auditor industry experience
by IPO companies does not exceed the demand by existing companies. This study
models the demand for auditor industry experience as a function of the information
risk of the offering. Studies such as Francis and Wilson [1988] model the demand
for audit quality as a function of agency costs. As Francis and Wilson note, these
agency costs generally increase with firm size. Existing publicly-traded companies are
much larger, on average, than IPO companies. Ettredge and Greenberg [1990] find
that auditor changes are, on average, associated with increase auditor industry
experience for the successor auditor compared with the predecessor auditor.5
4.2.4 Summag of Test Results - Demand for Auditor Experience

The first research hypothesis predicted that IPO companies would have a
positive demand for auditor industry experience. The overall results suggest that
companies going public do have a positive demand for auditor experience. The.
demand for experience by IPO companies exceeds the expected levels of experience
under models in which selection was made randomly among the Big 6 firms.
However, the demand for experience by IPO companies does not exceed the demand
for experience by existing publicly-traded companies.

The results of these tests are summarized in Figure 4.1. The results of these
tests are consistent with observed levels of concentration in the audit market (e.g.,
[Eichenseher and Danos 1981]). However, very little knowledge exists as to how
observed levels of auditor concentration have developed. As new publicly-traded
entities, IPO companies represent an important component of industry concentration

levels.

 

 

 

 

 

 

 

 

70

 

 

 

Model Model Description Result of Tests
Random Auditor choice is made Demand for auditor industry
selection randomly among the Big 6 experience by IPO companies
firms. exceeds demand predicted by
random selection model.
Random Auditor choice is made Demand for auditor industry
selection - randomly in proportion to

weighted by total
market share

each Big 6 firm’s total market
share.

experience by IPO companies
exceeds demand predicted by
random selection model
weighted by total market share.

 

 

Random
selection -
weighted by
industry market
share

 

Auditor choice is made
randomly in pr0portion to
each Big 6 firm’s share of the
market in the IPO company
industry.

 

Demand for auditor experience
by IPO companies does not
exceed demand for experience
under random selection model
weighted by industry market
share.

 

Figure 4.1

Summary of Results
Demand for Auditor Experience

 

 

 

 

71

The results of the tests of the demand for auditor experience indicate that
auditor experience is valuable in attracting new clients. This is not an obvious
conclusion. Danos and Eichenseher [1986] found that the industry market shares
held by the four largest firms in each industry decreased over the period 1950 to
1980. They attribute this result to increased competition. Auditor industry expertise
was not an important consideration in auditor changes in a study by Eichenseher and
Shields [1983].

The tests in this section indicate that IPO companies have a positive demand
for auditor industry experience. A model of the determinants of the demand for

auditor experience is tested in the following section.

4.3 Determinants of the Demand for Auditor Experience

4.3.1 Model Development

 

In section 3.1.2, it was hypothesized that the demand for auditor experience
would be related to the information risk of the firm. The variables chosen to proxy
for information risk were company age, company size, and membership in a
specialized industry. The following are the hypothesized relationships for those
variables:

HZA: The industry experience of the. auditor selected by the IPO
company will be negatively associated with the size of the IPO
company.

H23: The industry experience of the auditor selected by the IPO I

company will be negatively associated with the age of the IPO
company.

 

 

 

 

72
HZC: The industry experience of the auditor selected by the IPO

company will be positively associated with IPO company
membership in a specialized industry.

Additional control variables included in the regression model are the
reputation of the investment banker and the percentage of ownership retained in the
offering. In addition, an indicator variable is added for whether the IPO occurred
before or after the mergers that resulted in the Big 6.

Menon and Williams [1991] find that the choice of a Big 6 auditor is positively
associated with the use of a ranked investment banker (RBANK). Balvers,
McDonald and Miller [1988] argue that the investment banker either implicitly or
explicitly approves the choice of auditor. They suggest that a high reputation
investment banker will want to signal its quality through choice of a high quality
auditor. Accordingly, it is expected that high reputation bankers would prefer
auditors with industry experience.

A dichotomous indicator variable is used for investment banker reputation.
Eighteen investment banking firms are classified as ranked using the classification
scheme in Balvers et al. Other criteria can be used to classify investment bankers
as ranked; several studies use multichotomous classification Schemes (e.g., Menon
and Williams [1991]). Both Menon and Williams and Beatty [1989] indicate that
their results are robust to alternative classifications of investment banker reputation.

Leland and Pyle [1977] developed a signaling model where the entrepreneur
signals his or her private information as to the value of the firm by the percentage

ownership retained in the company. A high level of retained ownership is a positive

 

 

 

 

 

73
signal of the future prospects of the firm. In the Datar, Feltham and Hughes [1991]

model of auditor selection at the time of an IPO, the entrepreneur can substitute
auditor quality for the percentage of ownership retained to communicate private
information about the value of the firm. Because these signals are substitutes, they
hypothesize a negative relationship between auditor quality and the percentage of
retained ownership. If auditor experience is a dimension of auditor quality, this
implies a negative relationship between auditor experience and retained ownership.

The mergers that reduced the Big 8 accounting firms to the Big 6 resulted in
an upward increase in average industry market shares. Accordingly, an indicator
variable is added to indicate whether the IPO occurred before or after these mergers.
This indicator variable is included only in the models for the continuousmeasures
of experience since the effect of the mergers is more pronounced on these variables.

The full model of the cross-sectional demand for auditor industry experience
in the Big 6 audit market is as follows (predicted sign in parentheses - firm indicator

subscript suppressed for convenience):

(-) (-) (+) (+) (-)
AUDEXP = a + B,SIZE + fizAGE + B3SPEC + ,B,RBANK + BSREOWN + (1)
(+)
,BéPERIOD + 6
Where:
AUDEXP = Industry market share of the audit firm.
SIZE = IPO company size, measured by total assets prior to the IPO, plus the
offering proceeds.
AGE = Number of years the IPO company has been in existence.
SPEC = 1 if IPO company industry accounting is complex,
0 otherwise.
RBANK = 1 if a ranked investment banker is used, 0 if non-ranked.
REOWN = Percentage of ownership retained in the IPO.
PERIOD = 1 if IPO is after the mergers which created the Big 6,

0 otherwise.

 

 

 

 

74

A preliminary test of this model was conducted using the choice of a Big 6 or
non-Big 6 auditOr as. the dependent variable for the full sample of IPOs. The logistic
regression model provides a partial replication of Menon and Williams [1991] using
auditor levels rather than auditor changes. The test also offers a benchmark basis
of comparison for variable performance in tests of the model. The results of the

preliminary test of the model are reported in Table 4.5.

 

 

 

 

 

75

Table 4.5
‘ Logistic Regression Model
IPO Auditor Selection - Preliminary Test

 

 

 

 

 

Predicted ’ (n =660)

Variable Sign Coefficient Chi-square
Intercept 1.613 1639* * *
Proceeds + .039 11 .32* * *
Revenue + -.000 0.91
Age - -.063 908* * *
Banker + .868 5.01 * *
Reown - -.008 1.10
Model Chi-square = 69.09 (with 5 degrees

of freedom)* * *

 

** (***) Significant at the .05 (.01) level based on two-tail test.

Choice of a Big 6 auditor was positively associated with the size of the
offering. Surprisingly, there was no significant relationship between auditor choice
and company size measured by revenues. There was a negative relationship between
auditor choice and the age of the firm. This suggests that start-up entities select
auditors with higher reputations due to the greater information risk associated with
these firms. Consistent with both Simunic and Stein [1987] and Menon and Williams
[1991], there was a positive relationship between the reputation of the investment

banker and choice of a Big 6 auditor.

 

 

 

 

 

76
4.3.2 Model Results

The auditor experience model was estimated using ordinary least squares
regression for the continuous measures of experience, and using logistic regression
for the dichotomous measures. Preliminary tests indicated that the relationship
between the experience measures and IPO company age and the percentage of
retained ownership were insignificant. Inclusion of the variables AGE and REOWN
results in a loss of observations due to missing data. Accordingly, these variables
were drOpped from the regression models.

The lack of [significance for the variable AGE fails to support hypothesis 2B.
IPO company age was significantly associated with the decision to select a Big 6
auditor in Table 4.5. It is not clear why IPO companies with short operating
histories would prefer Big 6 auditors, but not audit firms with industry experience.

Table 4.6 contains the results of the regression model for the continuous
experience measures. Reported sample sizes are less than for the tests in Section
4.2 due to missing observations for the revenue variable. Chi-square values from
White’s test reported at the bottom of Table 4.6 indicate the presence of
heteroskedasticity in three of the models. Accordingly, the reported significance
levels are based on the z-statistics using the asymptotically consistent covariance

matrix.

 

 

 

77

Table 4.6
Auditor Experience Model
Continuous Variables

 

Estimated coefficient
(z-statistic)

Dependent Experience Measure

 

 

 

 

 

(n=428) (n=585)
Independent Predicted
variables Sign Clshare3 Sashare3 Clshare2 Sashare2
Intercept 11.129 15.777 9.693 11.110
(9.719)*** (6.103)*** (14.263)*** (7.157)***
Specialized + 4.011 1.559 1.493 1.557
Industry (2.594)* * * (.711) (2.300) * * (1.315)
Revenue (Ln) - 0.072 0.083 0.110 0.243
(.592) (.348) (1.735)** (1.778)”
Banker + 0.037 -0.032 -0.028 1.075
(.046) (-.018) (-.062) (.998)
Period + 3.124 1.723 3.618 3.196
(4.344) * * * (1.012) (7.883) * * * (2.920)**'*
Adjusted R-
Squared 0.070 -0.006 0.109 .022
F—Value 9.088** * .398 18.857“ * 4201*
Chi-Squared 20.664* "‘ 8.292 61.345* * * 34.186* "‘ *

 

*** (**) (*) - Significant at the .01 (.05) (.10) level. Z-statistics are one-tailed.

 

 

 

 

 

78

The overall explanatory power of all models is low. Consistent with the
_ results in Eichenseher and Danos [1981], the demand for auditor industry experience
is generally related to company membership in a specialized industry.6 Hypothesis
2A had predicted that auditor experience would be negatively related to IPO
company size. The results for the continuous measures generally support the
opposite relation. There was little evidence of a relationship between investment
banker reputation and auditor industry experience. As expected, experience levels
were higher in the post-merger period.

Table 4.7 reports the results from the logistic regression for the dichotomous
experience measures. The results are generally consistent with those for the
continuous measures. The models lack overall explanatory power. The specialized
industry variable is significant, but only for sales-based measures. The revenue

variable is highly significant in the opposite direction for the model SAEXP2.

 

 

79
Table 4.7

Auditor Experience Logistic Regression Model

Dichotomous Measures

 

 

 

 

 

 

 

Estimated coefficient
(Chi-squared statistic)
Dependent Experience Measure
Independent variables Predicted
Sign Clexp Saexp Lgexp
Panel A:
Three-digit measures
(n=428)
Intercept -0.866 -1.370 —1.386
(8.813)*** (17.801)*** (16.884)***
Specialized Industry + 0.327 0.485 -0.089
(1.240) (2.611)* (.062)
Revenue (Ln) - -0.019 0.018 -0.007
(.371) (.264) (.039)
Banker + -0.025 0.116 -0.077
(.011) (.067) (.081)
Model '
Chi-Squared 1.660 2.925 0.271
Panel B:
Two-digit measures
(n: 585)
Intercept -1.157 -2.550 2.134
(18.247)*** (43.424)*** (35.883)***
Specialized Industry + 0.196 0.447 0.578
(.819) (3.75)* * (5.669)***
Revenue (Ln) - 0.005 0.103 0.044
(.032) (6.985)*** (1.45)
Banker + 0053 -0.023 -0.198)
(.067) (.010) (.663)
Model
Chi~squared 0.862 13.600* * * 7.196*

 

*** (**) (*) - Significant at the .01 (.05) (.10) level. Test statistics are one-tailed where

appropriate.

 

 

 

 

80

4.3.3 Summaxy on Determinants of the Demand for Auditor Experience

The demand for auditor experience does not appear to be easily modeled.
A corollary can be drawn to studies of the choice of a Big 6 firm. Most of the
explanatory power in selection models such as Simunic and Stein [1987] and Menon
and Williams [1991] derive from size-dependent measures. Francis and Wilson
[1988] found only a weak association between agency costs and the change to a Big
6 auditor.

The demand for auditor experience was hypothesized to be negatively
associated with IPO company size and age, and positively associated with IPO
company membership in a specialized industry. The results of these tests are

summarized in Figure 4.2.

 

 

 

 

 

 

 

 

 

 

. Hypothesized
Hypothesis/ relationship to
Variable auditor experience Results
HZA: IPO company Negative Not supported. Results somewhat
size support the opposite finding of a
positive relationship between
company size and the demand for
auditor experience.
H23: IPO company Negative No evidence of a relationship
age between IPO company age and the
demand for auditor experience.
H2O: Specialized Positive Results generally support the
industry existence of a positive relationship
between the demand for
experience and membership in a
specialized industry.
Figure 4.2

Summary of Results
Determinants of Demand for Auditor Experience

 

 

 

 

81

Although it had been hypothesized that smaller IPO cempanies would have
a greater demand for auditor experience, the results were more consistent‘with the
opposite relation. This is consistent with studies that have found the demand for a
Big 6 auditor is positively related to company size [Simunic and Stein 1987; Menon
and Williams 1991]. No evidence was found of a relationship between the demand
for auditor experience and IPO company age.

Hypothesis 2C was generally supported. The demand for auditor experience
is positively associated with IPO company membership in a specialized industry. This
result is consistent with Eichenseher and Danos [1981] finding that observed auditor
concentration levels are positively related to industry regulation. This study extends
their result by demonstrating that this relationship exists for new entrants into the
audit market for publicly-traded companies.

The inability to develOp a better explanatory model of auditor experience is
also consistent with the results in F eltham, Hughes and Simunic [1991]. They were
unable to find an association between selection of a Big 6 auditor and measures of
firm risk. The lack of model explanatory power is not inconsistent with the
information quality theory of Titman and Trueman [1986]. In their model, high-
quality firms select high-quality auditors. However, firm quality is unobservable, at

least ex ante.

 

 

 

-T—¥

 

82
ENDNOTES '

It is necessary to control for industry because the sample experience measures are
based on audit firm shares of the total audit market, rather than the within Big 6
market. The Big 6 share of the market varies by industry.

The actual expected value for the variable CLE)G’ would be slightly higher due to
ties in a few industries.

Similar results for the continuous measures are obtained using the nonparametric
Wilcoxon signed ranks test, except the variables SASHARE3 and SASHARE2 are
no longer statistically significant at conventional levels. Similar results are obtained
for all dichotomous measures using a binomial test.

The tests in Table 4.3 are based on a one-sample t-test comparing the sample value
to the expected value under the market weighted model. Limitations in the
construction of the expected value in the market weighted model make it difficult to
perform a paired-sample t-test. For comparative purposes, the reported results for
the random selection model in Table 4.2 based on a paired-sample t-test do not
differ significantly from the results that would have been reported using a one-
sample t-test.

A preferable method for addressing the demand for auditor experience relative to
existing publicly-traded companies would be to match IPO companies on size with
existing companies. Preferably, these companies would not be recent IPO
companies. ' ‘

Another method to address the demand for auditor experience would be to focus on
auditor changes. If auditor industry experience is a dimension of quality, firms that
incur the cost of changing auditors prior to the IPO should demand higher levels of
auditor experience.

The ad hoc classification of industries as specialized was based on the existence of
specialized industry accounting or auditing standards. The standards generally relate
to industries as defined at the three-digit SIC level. In the regression model at the
two-digit SIC level, industries were classified as specialized if at least one of the
three-digit industries within that two-digit classification were specialized. The results
were substantially unchanged if the definition of specialized industries was based
solely on the three-digit classifications.

 

 

 

 

 

 

CHAPTER 5 - OFFERING UNDERPRICING RESULTS

5.1 Introduction

‘ Hypothesis three in Section 3.1.3 predicted that underpricing would be
negatively associated with the industry experience of the auditor.

H3: The underpricing of the IPO will be negatively associated with
the industry experience of the auditor.

This hypothesis is empirically tested in this chapter. In section 5.2, a model of
underpricing is developed. Preliminary tests of the model on the full sample of IPOs
using membership in the Big 6 as the measure of audit quality are included in section
5.3. The results of the tests of the relationship between underpricing and auditor
experience are reported in section 5.4. Additional specification tests are described

in section 5.5. Section 5.6 is a summary and conclusion on the results of the

underpricing tests.

5.2 Underpricing Model
A model of IPO underpricing is developed in this section to conduct empirical
tests of the relationship between offering underpricing and auditor experience. This
model is based primarily on the models in Balvers et al. [1988] and Beatty [1989].
Control variables for investment banker reputation (RBANK), age of the
client (AGE), the percentage of ownership retained in the IPO (REOWN), and
membership in a specialized industry (SPEC) are included in the model. These

variables were described in the previous chapter. Additional control variables

 

 

 

 

 

84

include IPOIND, a measure of the market return for newly issued securities, and the
exchange (EXCH) of-the offering company. A further description of the control
variables and the reason for their inclusion in the underpricing model is discussed
in this section.

An important factor associated with underpricing of the IPO is the reputation
of the investment banker. In the models in Beatty and Ritter [1986] and Balvers,
McDonald and Miller [1988], the reputation of the investment banker reduces the
uncertainty associated with the offering. Both Balvers et al. and Beatty [1989] found
that underpricing is less with the use of a high reputation investment banker. An
empirical link between investment banker reputation and underpricing is not
surprising since the investment banker sets the price of the offering. Beatty and
Ritter [1986] find that investment banking firms that, on average, set offering prices
either too high or too low tend to lose market share in subsequent periods.

Beatty [1989] argued that the age of the IPO company is related to the ex
ante uncertainty of the offering since a longer operating history provides investors
with more information concerning managerial decision-makin g. As Beatty notes, age
also may be associated with actual client risk. Beatty found that underpricing is
negatively associated with the age of the IPO company.

Similarly, Beatty includes the percentage of ownership retained in the IPO
[REOWN] as another variable to control for information risk. In the Leland and
Pyle [1977] signaling model, REOWN is a signal of management’s private
information concerning the company’s future prOSpects. This signal from

management reduces the uncertainty regarding the future cash flows of the firm.

 

 

 

85

In section 3.1.2, it was argued that information risk is greater for companies
in specialized industries. Accordingly, underpricing is predicted to be greater for
companies in specialized industries.

IPOIND is a market index of securities on the CRSP NASDAQ tape that
have traded for less than 20 days (Balvers et. al. [1988]). This variable attempts to
control for market effects present in the market for newly issued securities. Ritter
[1984] has shown that underpricing is greater in "hot” periods, when the IPO market
is active. Table 3.2 in Chapter 3 indicated some clustering of the sample, as 368 of
the 841 offerings occurred in 1991.

The variable IPOIND is constructed as follows. A daily market return is
computed as the average return on all securities on the CRSP NASDAQ tape that
have traded for less than 20 days. The monthly index for each day is equal to the
cumulative return for the past 20 days of the daily index.1 Defining the’monthly
index relative to the date of each IPO is a technical refinement over the approach
in Balvers, McDonald and Miller. The monthly index in Balvers et al. is calculated
. on a calendar basis; IPOs are assigned a value of the index based on the month of
offering. In this study, the monthly index is calculated on a daily basis so that the
index is always based on returns for the 20 days prior to the IPO.

The exchange in which the IPO is offered is included to‘ control for
differences in liquidity across exchanges. The majority of offerings involve
NASDAQ companies, although an increasing number of IPOs involve NYSE
companies. Within the NASDAQ market, firms may trade over-the-counter (OTC)

or through the National Market Service (NMS). NMS firms must meet certain

 

 

 

86

capital requirements, and market makers in NMS stocks provide real-time reporting
of transactions. Baker and Edelman [1992] find that spread narrows and volume
increases when firms switch from the OTC to NMS listings, consistent with greater
liquidity for NMS offerings. Previous studies have not controlled for listing
differences within the NASDAQ market.

Table 5.1 contains descriptive statistics for the sample by the exchange in
whichthe IPO is offered. The table suggests that NMS offerings are more similar
to AMEX offerings than OTC offerings. This is consistent with the finding in
Edelman and Baker [1990] that NMS firms do not significantly increase their
liquidity when they switch to the AMEX. Accordingly, the exchange variable
distinguishes OTC listings from other offerings. The exchange variable is a
dichotomous variable coded as 1 for NYSE, AMEX and NASDAQ-NMS offerings,

and 0 for NASDAQ-OTC offerings.

 

 

z

87

Table 5.1
Descriptive Statistics by Exchange
IPO Sample 1988-1991

 

 

Mean (standard deviation)
Percent
with Percent
Number (Millions) Ranked with
of Offering Offer Under- Investment Big 6
Offerings Proceeds Price Pricing Banker Auditor
Nasdaq - 278 6.90 6.01 .200 .054 .705
OTC (7.79) (3.63) (.331) (.226) (.456)
Nasdaq - 427 30.02 11.24 .109 .480 .911
NMS (34.90) (3.68) (.143) (.500) (.285)
AMEX 36 36.49 11.47 .063 .694 .972
(31.24) (4.56) (.095) (.467) (.165)
NYSE 100 160.34 17.41 .046 .970 .940
(164.40) (6.55) (.095) (.171) (.237)
All Offerings 841 38.01 10.26 .130 .407 .849
(77.17) (5.46) (.225) (.492) (.358)

 

 

 

 

 

 

 

 

88

The full underpricing model is as follows (predicted sign in parentheses; individual

firm subscript deleted for convenience):

5.3

-) (-) (+) (-)
UNDPRICE = a + p,RBANK + ,BZAGE + ,B,IPOIND + B4REOWN

(-) (+) (-)
+ BSEXCH + BéSPEC + B7EXP + e (2)

Where:
UNDPRICE = The initial return on the newly traded security.
RBANK = 1 if a ranked investment banker is used,

0 if non-ranked.
AGE 2 The number of years the IPO company has been in existence.
IPOIND = A market index of the return on securities traded

less than 20 days.
REOWN = The percentage of ownership retained in the IPO.
EXCH = 1 if offering is NYSE, AMEX or NMS, 0 if OTC.
SPEC = 1 if IPO company industry accounting is complex, 0 otherwise.
EXP = The industry experience of the audit firm.

Preliminary tests of the model are conducted in the following section.

Preliminary Model Tests

As a preliminary test of the underpricing model, equation 2 was estimated

using ordinary least squares regression on the full sample of IPOs. The following

modifications to the model were made in conducting these empirical tests. In the

preliminary tests using the full sample, the dichotomous Big 6/non-Big 6 reputation

variable was used as the measure of audit quality. In addition, an auditor-investment

banker interaction term was included since Balvers, McDonald and Miller [1988]

suggest such an interaction term is important. The specialized industry variable is

 

 

89

excluded from this test because this variable was not measured for IPOs involving

non-Big 6 auditors.

There are three reasons to estimate this model. First, as a replication of
previous research and as a benchmark for evaluating the significance of the
experience variable. Second, to evaluate performance of control variables for
inclusion in the model. Third, to assess the importance of the variable EXCH which
has not been incorporated in previous tests of offering underpricing.

The results of the preliminary regression model are included in Table 5.2.
Three separate models are estimated. Model 2A includes AGE, REOWN, and an
auditor-investment banker interaction term. In model 2B, AGE and REOWN are ‘
excluded because they do not achieve significance in Model 2A, and missing data for
these variables results in a loss of observations. The auditor-investment banker
interaction term is also excluded because it is highly collinear with the investment
banker variable. Lastly, in model 2C the exchange variable is added to model 2B.

This model is estimated separately to isolate the effect of the exchange variable on

other variables.

 

 

90

Table 5.2
IPO Underpricing Model
Preliminary Test

 

 

 

 

Estimated coefficient
(z-statistic)
Independent variables Predicted (n= 680) (n=838) (n =838)
Sign Model 2A Model 2B Model 2C
Intercept 0.245 0.201 .221
(6.245)* * * (7 .7 19)** * (7.838)***
Banker -0.132 -0.095 -0.065
(-2.361)** * (-7.046)** * (-5.788)***
Big 6 Auditor -0.075 -0.046 -0.032
(—1.880)** ,(-1.661)** (-1.125)
Banker * Big 6 + 0.031
(0.536)
IPOIND + 0.482 0.450 0.469
(3.489)*** (3.559)*** (3.767)***
Age -0.0009
(0596)
REOWN . -0.0011
(0242)
Exchange - 0067
(-3.179)* **
Adjusted R-Squared 0.083 0.067 0.080
F-Value 11.219*** 20.874*** 19.142***
Chi-Squared 3916* * * 3494* * * 37 .58** *

 

* (**) (***) Significant at the .10 (.05) (.01) level; one-tail tests where appropriate.

 

 

91

In Model 2A, the reputation of both the investment banker and auditor have
a significant negative relationship with underpricing. The magnitudes of the
coefficients for these variables are larger than, but consistent with the values
reported in Balvers, McDonald and Miller [1988]. Specification tests using only
NASDAQ offerings indiCate that the higher reported values are largely due to the
inclusion of NYSE and AMEX offerings. Balvers et al. excluded NYSE and AMEX
offerings from their sample.

The banker-auditor interaction term was insignificant. The coefficient of .031
is approximately half the value of .058 reported by Balvers et al., One possible
reason for the lack of significance for this variable is that it is highly correlated with
the investment banker reputation variable.

Also as expected, there is a significant positive relationship between
underpricing and IPOIND. Neither the variable AGE or REOWN were significant.
Beatty [1989] found these variables to be significantly related to underpricing.
However, his results were based on a sample of 2215 that also included best efforts
offerings.

The primary result of model 2B is that deletion of the interaction term
reduces the coefficient for investment banker reputation, while greatly increasing the
significance of this variable. This result is consistent with the effect of a highly
collinear variable (Belsley, Kuh and Welsch [1980]).

In Model 2C, the exchange variable is significant as predicted. More
importantly, inclusion of this variable results in a reduction of the significance of the

investment banker and auditor reputation variables. This is an important finding

 

92

since it suggests that the magnitude of the relationship between underpricing and
auditor and investment banker reputations may have been overstated in previous
studies that treated NASDAQ as a homogeneous market.

The results for Model 2C were substantively unchanged under different
exchange classifications.2 Because the exchange variable is highly correlated with size
variables, different size measures were added to the model. The results were
substantially unchanged when either revenue or offering proceeds were added to the
model to control for IPO company size.

The model R-squared values are relatively low, but consistent with the range
of values reported in previous studies. The chi-squared values from White’s [1980]
test indicate the presence of heteroskedasticity in each of the models. The reported
significance levels are based on the z-statistics using White’s asymptotically consistent
covariance matrix. The reported results are generally consistent with previous
empirical studies. This suggests that the model and data are appropriate for

conducting empirical tests.

5.4 Research Results

Tests of hypothesis three of the relationship between auditor experience and
underpricing were conducted using the experience measures described in section 3.2.
The sample consists of IPOs with Big 6 auditors for which an industry experience
measure could be calculated.

Consistent with the results reported in the previous section, preliminary tests

of the model indicated no significant relationship between underpricin g and company

 

 

 

93

age or the percentage of retained ownership. These variables were dropped from the
model beCause’they result in a number of missing observations.

The results of the regression model for the three-digit experience measures
are included in Table 5.3. The chi-squared values reported at the bottom of the
table indicate the presence of heteroskedasticity in each of the models. The reported
significance levels are based on the z-statistics using the asymptotically consistent
covariance estimates. Consistent with the preliminary results for the full sample,
there is a significant relationship in each model between underpricing and the control
variables for investment banker reputation, the IPO return index, and the exchange
of the offering.

Surprisingly, underpricing was lower in specialized industries. Because many
of these industries are regulated, one conjecture is that there is less uncertainty
about future earnings for these companies due to the presence of regulation. Teets
[1992] found that earnings response coefficients are smaller for electric utilities than
for nonregulated companies, indicating that unexpected earnings are less permanent
for utilities. He suggested that this is consistent with regulators buffering utilities
from changes in the operating environment (p. 275).3

The relationship between underpricing and the experience measures is
negative in four of the five models, but is statistically significant only for the variable
SAEXP3. The relationship between underpricing and the continuous experience
measures were very small. The direction of the coefficients for the dichotomous
measures are at least suggestive of a relationship between underpricing and auditor

experience.4

 

 

 

94

 

 

 

 

 

 

 

2 382 oogotogxm

Aosmufimsv EEQEoOQ 3383mm

axis

.aeenoane 22.? 8...: ans =3-on .62: a: E: S. 2: a Banana - C on;
£545 1%.; :38 :28 Renew Baseméo
***b:u.h azimfinmh scramwmwb .433“me ECIEmmN ©36>inm
So. QB. $0.0 Sod $3 Banana e333
....Am~e.e-v txaeasv igni ......8me.l.-v infini bases
weed- weed- $0.0- weed- ES- + Seesaw
Lani ASN.3 5&3 ASN.3 Emmi
2.3- 08.0- 2.0.0- 35. 2.3- - oweeeoxm
tags .2898 :qus 1:38 :6qu
$2 82 :3 :3 ammo + A254:
. 232893 :2?qu ......:S.m-v :RSE :LNSE
. SS- ES- ES- Rod- Rod- - been
so: 1.65.3 Gene 8%; sea
83. 78.0- :3- 88.- 88. - ooeoconxm
588.3 1.333 :Lsmeé ......:s..a :LRQV
a; 83 mm; a; a; 685::
98mg 96am Q55 822% SEED swam
35.5.55 SEED? E25333:—

 

, 3.5232 8520an “$5.32;
. .352 waging: 9:

 

Wm oEuH

 

 

 

 

 

 

 

 

95

The results for the underpricing model using the twqdigit experience
measures are reported in Table 5.4. Use of two-digit measures increases the sample
size from 449 to 614 observations. The increase in sample size is associated with an
increase in the statistical significance of the control variables. The chi-square statistic
also indicates the presence of heteroskedasticity in each of these models. Reported
significance levels are for the z-statistics based on the asymptotically consistent
covariance estimates.

When experience is measured using two-digit SIC codes, the coefficients for
all of the experience variables are very small, and inconsistent as to direction. The
results at the two-digit SIC level provide little support for the existence of a

relationship between underpricing and auditor industry experience levels.

 

 

 

96

 

3252.5. 22? was as. 32-25 32.2 83 $3 8. 2: a 28$:me - C A13

 

 

 

.15.: 1.3.8.3 1.08.8 .1308 .1mw3m 8.33-5
***Nm.w.w ***voo.w ***th.w ***vfio.w ***©Ofi.© 03m>nm
go. so. 83 God 83 853m-.. 333..
1 13:3 1833 .2433 162.3 1.33.3 9265
_ 53. 98.0- Rod- 08d- 08.? + 8535st
.32. 3 .33“. 3 LE. 3 .25.... 3 .69.. 3
Rod- 33.0. 20.0- 48.0- Rod- - 3.53m
.1533 .1833 1.3.23 1.28.3 1.3.2.9
. 33 3.0 was 82 :33 + 925.:
.1883 .1§m.3 1.23.3 1.32.3 1.8%3
wood. wood. $0.0. $0.0- $0.0- - .328
38.3 A83 9.8.3 833 $2.:
. $8.0 $8.0- 880.0 88.- 28. - 85.5%.
1.23.8 153.8 1.838 1.88.8 1.50.8
Ed Ed :3 Ed mad 38.35
98mg axoam 965 223m SEED swam
@806on 83st? E02835

 

 

8332 oosotomxm

65385-3 288508 322235
€823

 

3...;on oouoieaxm— “35-958
.352 33:52: 9:
, Wm «35.

 

 

 

 

97

The overall lack of significance for the experience variables is inconsistent
with the third research hypothesis. 3 One possible explanation is that the choice of
auditor does not significantly affect offering underpricing. Although Balvers,
McDonald and Miller [1988] and Beatty [1989] find that choice of a Big 6 auditor
is associated with significantly lower underpricing of the offering, evidence reported
in section 5 .3 suggests that the relationship between auditor reputation and offering
underpricing is reduced after controlling for the exchange of the offering. The lack
of association between auditor reputation variables and offering underpricing is also
consistent with previous studies of the economic consequences of auditor choice. In
section 2.2.4, two studies were cited in which the relationship between choice of a

Big 6 auditor and capital market activity was not statistically significant.

5.5 Specification Tests

A number of procedures were performed to further investigate the
underpricing results. These tests relate to multicollinearity, outliers, the definition
of experience, industry clustering, and the changing composition of the Big 6.

Multicollinearity does not appear to be a problem in the model. The largest
condition index of approximately 5 is well below the threshold of 30 suggested by
Belsley, Kuh and Welsch (p.105) as indicating harmful levels of multicollinearity.
The highest variance inflation factors are approximately 1.30 for the banker and
exchange variables. These variables were significant in each model. The experience
variables were substantially uncorrelated with other variables. The variance inflation

factors for the experience variables were close to 1.

 

 

 

 

98

Underpricing models can be influenced by a few extreme observations.5 A
few offerings enjoy substantial increases in price in the first day of trading. To limit
the effect of these observations, underpricing was arbitrarily truncated at a maximum
of .50 for the underpricing models at the three-digit level. In the truncated models,
adjusted R-squared values increased to the .09 range. The control variables
increased in significance, with the exception of the exchange variable which became
insignificant.6 The estimates and significance of the experience variables did not
change significantly, and remained insignificant.

The strongest reported results for the experience measures are for the
dichotomous measures at the three-digit level. These are the measures that had
been expected to best capture auditor experience. In some instances, an audit firm
is classified as the market leader even though its market share is not substantially
larger than that of the firm with the next largest market share. In several industries,
ties result in more than one firm being classified as the most experienced. To
address this issue, the classification of the market leader was limited to audit firms
with more than 25 (35) percent of the clients (sales dollars). Results did not change
substantially when this definition of experience was used.

Approximately 40 percent of the offerings at the three-digit level came from
five industries. Dummy variables were added to the underpricing model for each of
these industries to assess whether underpricing was related to IPO company industry.
Only one of the dummy variables was marginally significant, and inclusion of the

dummy variables had little effect on the other regression coefficients.

 

 

99
5.5.1 _ Effects of Big 6 Mergers

 

An additional specification issue concerns the effects of the mergers of Ernst
and Whinney with Arthur Young, and Deloitte, Haskins and Sells with Touche Ross.
The combined market shares of these firms increased dramatically in many industries
after the mergers. However, while the combined market shares of these firms may
translate into industry experience that is valued in the marketplace, it is unlikely that
this would occur immediately. To address this issue,- the underpricing model was
estimated separately for the pre-merger and post-merger period. In addition, the
extent of underpricing by individual Big 6 firms was examined.

In the separate period regression tests, the primary focus was on the effect of
auditor experience in the period prior to the mergers of the Big 6. The magnitude
of the coefficients for the experience variables did not suggest the existence of a
significant relationship between underpricing and auditor experience in the pre-
merger period. However, the sample size is less than 200 in this period.

To further examine whether the effect of the mergers had any relationship
with offering underpricing, underpricing levels were calculated by individual Big 6
firm. This was done for two reasons. First, the merged firms of Deloitte & Touche
and Ernst & Young were involved in the largest number of IPOs in the 1990-1991
period. As previously indicated, the experience measures for these firms dramatically
increased as a result of the mergers. Second, Beatty [1989] indicates that there are
reputational differences among the Big 6. Mean and median underpricing for the

Big 6 firms for the pre-merger and post—merger period are reported in Table 5.5.

100
Table 5.5

Underpricing by Audit Firm

 

 

 

 

 

1988-1989 1990-1991
Audit Firm Mean Mean
n (Std. Dev) Median 11 (std. dev.) Median
Arthur Andersen 39 .085 .025 68 .118 .087
(.149) (.168)
Coopers and 42 .086 .029 57 .114 .056
Lybrand (.140) (.146)
KPMG Peat 47 .070 .021 84 .119 .073
Marwick (.111) (.169)
Price Waterhouse 23 .074 .023 35 .075 .029
(.137) (.196)
Deloitte, Haskins 33 .076 .022
and Sells (.133)
Touche Ross 24 .194 .062
(.317)
Deloitte and Touche 82 .198 .098
. (.291)
Arthur Young 26 .087 .048
(152)
Ernst and Whinney 23 .059 .013
(139)
Ernst and Young 120 .150 .072
(.278)
Total 257 .088 .026 446 .137 .075
(.163) (.229)

 

 

 

 

 

 

101

The highest mean and median levels of underpricing in the 1990-1991 period
are for the two newly-merged firms of Deloitte & Touche and Ernst and Young.
The underpricing levels for Touche Ross are also high in the pre-merger period.

To assess the effect of the merged firms on the experience variables, a dummy
variable was added to the model if the auditor was one of the two merged firms in
the 1990-1991 period. This is an ad hoc procedure to the extent that it is known ex
post that these firms had high levels of underpricing. However, an a priori concern
was the effect of including merged entities in the calculation of experience measures.
Also, Beatty [1989] suggests that the largest firms differ in their reputational capital.
The model is reported in Table 5.6 for the variables CLEXP3 and SAEXP3.

The dummy variable for the merged firms is positive and statistically
significant. The reported values and significance levels of the experience variables
also increase in relation to the values reported in Table 5.3. Qualitatively similar
results are obtained when the model is estimated with individual firm dummies for
the two firms, and when the underpricing model is estimated for the post-merger
period only. This provides evidence that the results for the auditor experience
variables are reduced by the effects of the increased market shares of the merged
firms.

The significance of the merged firm dummy variable suggests that the mergers
may have lowered the reputational capital of the (merged firms. Beatty [1989]
included dummy variables for the 20 firms with more than 10 offerings during the
period 1975-1984. Although Beatty had a much larger sample size, none of the

indicator variables in his study was statistically significant.7

 

 

 

102

Table 5.6 _
IPO Underpricing Model
Merged Firm Dummy Model

 

 

 

 

 

 

(n =449)
Estimated coefficient
Independent variables Predicted (t-statistic) (z-statistic)
Sign Clexp3 Saexp3
Intercept 0.179 0.181
(8.750)*** (6.000)***
Experience 0032 -0.041
(-1.381)* (-2.137)**
Banker -0.074 -0.075
(-3.284)*** (-5.019)***
IPOIND 0.291 0.281
(1.970)“ (1.698)“
Exchange 0042 -0.041
(-1.684)** (—1.312)*
Specialized industry 0062 -0.061
(-2.275) * * (-4.204) * * *
Merger variable 0.073 0.070
(3.236)*** (2.412)”
Adjusted R-Squared 0.086 .088
F-Value 8.035 * * * 8245* * "‘
Chi-Squared 27.486 32.323 *

 

 

*** (**) (*) - Significant at the .01 (.05) (.10) level. One-tail tests used where appropriate.
Based on z-statistic from asymptotically consistent covariance matrix where model is
heteroskedastic.

 

 

 

103

The results in Table 5.6 are important in (that they apparently demonstrate
differences in audit quality both Within and between Big 6 firms. The results suggest
that quality varies within firms as a function of industry experience, and between
firms as a function of overall firm reputational capital. Because industry experience
and overall firm reputation may be related, studies that do not control for differences
in reputational capital may fail to find hypothesized relationships between measures

of audit quality and auditor industry experience.

5.6 Summary and Conclusion for Underpricing Tests

The underpricing tests do not support the hypothesis that the general
construct auditor experience is associated with lower underpricing of the offering.

However, four (of five) measures at the three-digit SIC level had the hypothesized

negative coefficient. The dichotomous measures of experience at the three-digit SIC
code level each had the predicted negative sign, and the sales-based measure SAEXP
was statistically significant at conventional levels. These results do not rule out the
existence of a relationship between auditor industry experience and underpricing.
Further, as reported in Table 5.6, both the sales-based and client-based dichotomous
measures at the three-digit level were significant when an indicator variable is added
for audit firms that had been involved in mergers.

The significance of the experience measure and merged firm indicator variable
appears to suggest differences both within firms based on industry experience, and
between firms based on overall reputational capital. The significance of the merged

firm variable suggest that the reputational capital of the firms involved in these

 

 

 

 

 

 

 

104

mergers may have declined following the mergers. This subject deserves further

' investigation as additional data becomes available.

The underpricing results can also be compared to underpricing tests that use
membership in the Big 6 as the measure of quality. Previous studies by Balvers,
McDonald and Miller [1988] and Beatty [1989] suggest that Big 6 firms are
associated with lower underpricing of the offering. However, the tests described in
section 5.3 suggest that the relationship between this quality measure and
underpricing is not statistically significant when the exchange of the offering is
controlled for. Previous research on offering underpricing has not controlled for

these exchange differences.

 

 

105
ENDNOTES

Since underpricing is a one day return, the use of a one-day index would seem
appropriate. However, the number of observations for each daily index is generally
small (an average of 25-30 observations). Comparison of the monthly and daily
indices supported a slightly stronger relation between the monthly index and
underpricing.

The alternative exchange classifications were as follows:

Nasdaq (OTC) l Nasdaq (OTC) 1
Nasdaq (NMS) 2 Nasdaq (NMS) 2
AMEX 3 NYSE and AMEX 3
NYSE 4

As a regulated monopoly, electric utilities represent an extreme form of price
regulation. It is an empirical issue whether there is less uncertainty about earnings
for other regulated industries.

An economic interpretation of the coefficients may be helpful. The mean offering
proceeds for offerings involving Big 6 auditors as reported in Table 3.2 is 42.56
million. A one percent reduction in underpricing is associated with a reduction in
underpricing of $425,000.

This researcher is unaware of any underpricin g research which addresses this issue.

It is not surprising that the exchange variable becomes insignificant when
underpricing is truncated at .50, since most extreme underpricing observations
involve OTC offerings.

It should also be noted that OTC offerings are more likely to involve non-Big 6
auditors. The preliminary underpricing model 2C in Table 5.2 was reestimated,
limiting underpricing to a maximum of .50. Although the magnitudes of all the
coefficients were reduced, the statistical significance of each variable was unchanged.

Beatty reports coefficients of .063 for Peat Marwick and -.O47 for Price Waterhouse.
These coefficients represent large underpricing effects, but were not statistically
significant. For comparison, the coefficient for the Big 6 reputation variable in
Balvers et al. [1988] was -.043. One speculation is that Beatty’s use of a longer time
period without a time-varying intercept, or theinclusion of best efforts offerings,
results in larger standard errors.

 

 

 

 

CHAPTER 6 - AUDITOR COMPENSATION RESULTS

6.1 Introduction
Hypothesis four in section 3.1.4 predicted that auditor experience would be
positively associated with increased auditor compensation.

H4: Auditor industry experience will be associated with higher audit
fees.

This hypothesis is empirically tested in this chapter. The model of auditor
compensation is described in section 6.2. Preliminary tests of this model are
included in section 6.3 using membership in the Big 6 as the measure of audit
quality. Section 6.4 reports the results of the empirical tests of the research

hypothesis. Section 6.5 is the summary and conclusion.

6.2 Auditor Compensation Model

The model of auditor compensation is based on Beatty [1989]. In his model,
IPO cash compensation is a function of the offering proceeds, and pre-offering
revenues and equity' of the IPO company. Beatty’s model is expanded in this study
to include the specialized industry and exchange variables that were described in
Chapter 5, as well as the auditor experience variables.

As in Beatty [1989], the total cash compensation in the IPO is used as the
proxy for audit fees. Total cash compensation represents most of the costs of the
IPO other than the underwriter fee, and includes legal fees, printing costs and the

auditor fee. For the portion of his sample for which the audit fee is available, Beatty

106

 

 

 

 

107

reported a correlation between total cash compensation and estimated auditor
compensation of .77. Beatty concluded that total cash compensation is a good
surrogate for auditor cash compensation.

As described in section 3.3.1 in Chapter 3, data for this study was gathered
from Going Public: the IPO Rewrter (Going Public) and the DiSClosure, Inc. SEC
D’33 database (D33). Only total cash compensation is available from Going Public.
Both total cash compensation and estimated auditor compensation are available on
D33. The correlation between these two variables for the period 1990-1991 of .64
is comparable to the value of .77 reported by Beatty.

The three variables incorporated from Beatty’s model (revenues, equity and
offering proceeds) are all size variables designed to proxy for the marginal cost of
the audit and are predicted to be positively associated with IPO cash compensation.
In addition, it is likely that many of the cash expenses in the IPO vary directly with
the offering proceeds. The model in Beatty is simpler than previous fee models such
as those suggested by Simunic [1980] and Palmrose [1986]. Beatty’s model does not
control for such factors as entity complexity and risk. Nonetheless, Beatty argued
that his model is a reasonably proxy for the marginal cost of the audit.

The specialized industry variable is added to test the prediction in Danos and
Eichenseher [1986] that industry specialization leads to economies of scale in
specialized industries. The competing hypothesis is that specialized industries are
associated with higher audit fees because of greater accounting and auditing

complexity. Accordingly, no prediction is made as to the direction of this variable.

 

 

 

 

108

- The exchange of the IPO-company is also added to the model. The inclusion
' of this variable is suggested by the use of total IPO cash compensation as a proxy for
audit costs. Certain IPO costs are expected to be larger for NYSE, AMEX and
NASDAQ-NMS offerings that are more widely distributed.

The auditor experience variables are also added to the model to test the
research hypothesis that audit fees are positively related to the experience of the
auditor. The full cash compensation model is as follows (predicted sign in

parentheses; individual firm subscript deleted for convenience):

(+) (+) (+) (+) ('0
CASH COMP = a + ,BlEXP + szRoc + 13,an + B4EQUITY + BSSPEC
(+)
+ 36EXCH + e (3)
Where:
CASH COMP 2 Cash expenses paid in the IPO.
EXP 2 Experience level of the auditor
PROC = Offering proceeds.
REV = Sales of the IPO company.
EQUITY = Book value of equity.
SPEC = 1 if the company is in a specialized industry,
0 otherwise
EXCH = 1 if the company is listed on NYSE AMEX or NMS

0 otherwise

Preliminary tests of the model are conducted in the following section.

6.3 Preliminary Model Tests
As with the underpricing model, preliminary tests of the compensation model
were conducted on the full sample of IPOs using the Big 6/non—Big 6 classification

as the measure of audit quality. The results of these tests are reported in Table 6.1.

 

 

 

 

 

109

As in the tests of the offering underpricing model, the indicator variable for
specialized industry is excluded because this variable was not measured for offerings
with non-Big 6 auditors. Because the exchange variable is a modification of Beatty’s

model, the cash compensation model is estimated with (Model 3A) and without

(Model 3B) the exchange variable to isolate the effect of this measure.

 

 

 

110
Table 6.1

. IPO Cash Compensation Model

Preliminary Test

 

Estimated coefficient
(z-statistic)

 

 

 

 

(n=741)
Predicted
Independent variables Sign Model 3A Model 3B
Intercept 392.435 348.307
(14.506)** * (14.506)***
Proceeds + 0.0048 0.0046
(16.110)*** (15.440)***
Rev + 0.00009 0.00008
(2.941)*** (2.937)***
Equity + 0.00035 0.00035
(.667) (.661)
Big 6 Auditor + 104.846 60.311
(3.041)*** (1.884)**
Exchange + 130.511
‘ (3.596) * * *
Adjusted R-Squ ared 0.441 0.448
F-Value 146.687*** 121.120***
Chi-Squared 2367* * * 21 .74* * *

 

 

* (**) (***) significant at the .10 (.05) (.01) level based on a one-tailed test.

 

 

 

 

111

As with the underpricing models, the chi-square value from ,White’s test
indicates the presence of heteroskedasticity. The reported significance levels are
based on the z-statistics using the asymptotically consistent covariance matrix.

In model 3A, use of a Big 6 auditor is associated with a significant increase
‘in IPO cash compensation. As expected, each of the additional three variables is
positively associated with IPO cash compensation. In model 3B, the exchange
variable is positive and significant. The addition of the exchange variable results in
a reduction of the magnitude and significance of the auditor coefficient.

The model R-squared values of .44 are significantly higher than the .18
reported by Beatty [1989]. Additional tests indicate that the results are affected by
the data source used. The 1988—89 data is from Going Public, which is the primary
data source used by Beatty. Estimation of the cash compensation model for the
1988-1989 period resulted in a reported R-squared of approximately .14. The R-
squared value for the 1990-91 subperiod was approximately .50. Data for this period

was primarily gathered from the D33 database.

6.4 Research Results

Hypothesis four predicts that auditor experience will be associated with
higher levels of auditor compensation. Empirical tests of this model are conducted
using the experience measures described in section 3.2. Table 6.2 reports the results

of this model for the three-digit experience measures.

 

 

 

112

Table 6.2
IPO Cash Compensation Model
Three-digit Experience Measures

 

 

(N 2408)
Estimated coefficient
(t-statistic)

Experience Measure

 

 

 

 

Independent
variables Clshare Sashare Clexp Saexp Lgexp
Intercept 319.115 368.877 351.949 389.077 398.909
(4.986)*** (6.785)*** (7.020)*** (7.766)*** (8.001)***
Experience 6.236 1.721 164.172 41.524 -0.214
(1.890)* * (1.197) (3.086)* * * (.736) (-.003)
Proceeds 0.0045 ' 0.0045 0.0045 0.0045 0.0045
(12.157)*** (12.321)*** (12.376)*** (12.293)*** (12.258)***
Revenue 0.00008 0.00009 0.00008 0.00009 0.00009
(2.858)* * * (2.950)* * * (2.698)* * (2.887)* * * (2.996)** *
Equity 0.00033 0.00030 0.00032 0.00031 0.00031
(2.441)** * (2.246)** (2.418)* * * (2.342)* * * (2.345)***
Specialized ~74.026 -52.841 461.148 -54.880 309%
Industry (-1.101) (-.797) (-.931) (-.825) (-.768)
Exchange 126.113 131.870 140.296 133.721 133.249
(2.212)” (2.311)” (2.482)*** (2.341)*** (2.332)***
Adjusted R-
Squared 0.430 0.427 0.438 .425 .425
F-Value 52.103*** 51.473*** 53.831*** 51.212*** 51.052***
Chi-Squared 32.401 31.706 28.556 33.389* 31.395

 

*** (**) (*) - Significant at the .01 (.05) (.10) level. One-tail tests used where appropriate.

 

 

 

 

113

Modelperformance and estimates of the control variables are consistent with
those reported in Table 6.1. The variable for membership in a specialized industry
(SPEC) was negative in all regressions, but. was not statistically significant at
conventional levels. The chi-square statistic from White’s test did not indicate
significant levels of heteroskedasticity.

The coefficient for the continuous and dichotomous measures of experience
based on number of clients were positive and statistically significant. The coefficients
for the measures based on client sales were positive, but were not statistically
significant. (The results suggest that client-based industry experience is associated
with a statistically significant fee premium in the audit market.

It is interesting to compare the results of the compensation model to the
results of the offering underpricing model. The only significant experience variable
in the underpricing model was SAEXP, the dichotomous sales-based measure. In
contrast, the fee premium is larger for client-based measures. This suggests that the
value of auditor industry experience consists of factors other than its effect on
underpricing of the offering. However, caution is warranted in drawing this
conclusion. The results may be attributable to unknown omitted variables. Also, the
underpricing tests may have lacked power to detect a relationship between auditor
experience and underpricing.

The results for the dichotomous measures can also be contrasted with the
results for the Big 6 indicator variable in Table 61. The coeffiCient for client
experience of 164.17 is much larger than the Big 6 premium of 60.31 reported in

Table 6.1. This suggests that at least a portion of the fee premium for Big 6 firms

 

 

 

 

 

 

114

is due to industry experience. To test this implication, the variable CLEXP was
added to the compensation regression model for the full sample of IPOs. The
variable CLEXP is as previously defined, and is equal to zero for all non-Big 6 firms.
The results of the cash compensation model for the full sample of IPOs is reported
in Table 6.3. The results are estimated with and without the experience variable.

When the dichotomous measure of client experience is added to the
regression model, the coefficient for the Big 6 variable is reduced by a third, from
60.31 to 39.58. The variable for client experience is significant, while the Big 6
variable is not. Because the client experience variable was already found to be
significant for IPOs with Big 6 auditors, it would not be appropriate to interpret this
result as indicating that experience commands a fee premium, and membership in
the Big 6 does not. However, the result is consistent with a portion ofthe fee
premium for the Big 6 representing the effects of industry experience.

White’s chi-square test indicates the presence of heteroskedasticity in the
model. The reported results are based on the z-statistics using the asymptotically

consistent covariance matrix.

 

 

 

115

Table 6.3
IPO Cash Compensation Model
Ell] Sample of IPOs With
Experience Variable

 

(N =741)
Estimated coefficient
(z—statistic)

 

 

 

 

 

Independent Without With
variables Experience Experience
Variable Variable
Intercept 348.307 347.529
(14.506)*** (14.997)***
CLEXP 116.433
(2.230)* *
Big 6 60.311 39.578
(1.884)** (1.178)
Proceeds 0.0046 0.0046
~ (15.440)*** (15.447)***
Revenue 0.00008 0.00008
(2.937)* * * (2.697)* * *
Equity 0.00035 0.00035
(.669) (.698)
Exchange 130.511 133.115
(3.596)* * * (3.670)* * *
Adjusted R-Squared 0.448 0.452
F-Value 121.120*** 102.593***
Chi-Squared 29.044* * 39490“ *

 

 

*** (**) (*) - Significant at the .01 (.05) (.10) level based on a one-tailed test.

 

 

 

 

116

The results for the compensation model using experience measures at the two-

digit SIC level are reported in Table 6.4. The specialized industry variable was

deleted since it was not significant in the three-digit model. Excluding this variable.

from the three-digit model had little effect on other coefficients.

Although the coefficient for the variable CLEXP decreased significantly
compared to the three-digit model, it was still statistically significant. Compared to
the models at the three-digit level, the variable CLSHARE lost significance, while
the variable SASHARE gained significance. The results at the two-digit level are
somewhat consistent with a fee premium for auditor experience.

The chi-squared statistics at the bottom of Table 6.4 indicate the presence of
heteroskedasticity in each of the models. Use of the asymptotically consistent
covariance matrix did not effect the reported significance levels for the experience
variables. Accordingly, the reported results are the t-statistics from the ordinary least
squares regression to maintain comparability to the compensation models at the
three-digit level reported in Table 6.2.

The lack of significance for the variable SAEXP2 is consistent with the results
in Palmrose [1986]. She was unable to find a fee premium for auditor experience
based on sales at the two-digit level. The results reported here suggest that both the
use of experience at the two-digit level, and the use of sales-based measures, may

have contributed to her inability to find a relationship between auditor experience

and audit fees.

 

 

 

 

 

117

_ Table 6.4 ,
IPO Cash Compensation Model
Two-digit Experience Measures

 

(N =556)
Estimated coefficient
(t-statistic)

Experience Measure

 

 

 

 

Independent
variables Clshare Sashare Clexp Saexp Lgexp
Intercept 382.333 379.566 401.798 412.574 411.608
(6.577)*** (8.525)*** (10.097)*** (10.567)*** (10.556)***
Experience 2.787 2.612 60.761 29.144 38.845
(.798) (1.647)** (1.397)* * (.607) (.756)
Proceeds 0.0044 0.0044 0.0044 0.0044 0.0044
(15.436)*** (15.595)*** (15.615)*** (15.618)*** (15.584)***
Revenue 0.00009 0.00009 0.00009 0.00009 0.00010
(3.296)* * * (3.287)* * * (3.228)* * * (3.226)*** (3.395)***
Equity 0.00030 0.00028 0.00029 0.00029 0.00029
(2.520)* * * (2.369) * * * (2.490)* * * (2.461)* * * (2.449)“ *
Exchange 109.187 106.286 112.182 110.452 110.989
(2.388)* * * (2.329)* * * (2.462)* * * (2.419) * * * (2.433)* * *
Adjusted R-
Squared 0.472 0.474 0.473 .472 .472
F-Value 100.268*** 101.060*** 100.770*** 100.165*** 100.243***
Chi—Squared 34.463* * 34.573* * 29.657 * * 33.713* * 31.624* *

 

*** (**) (*) - Significant at the .01 (.05) (.10) level. One-tail tests used where appropriate.

 

 

 

118

6.4.1 Compensation Regression with Audit Fees

One concern with the results for the'cash compensation models is the use of
total IPO cash compensation as a proxy for audit fees. The data for the 1990-1991
period were drawn primarily from the Disclosure, Inc. D’33 database. This database
includes an estimate of audit fees included in cash compensation. Table 6.5 includes
an estimate of the compensation model at the three-digit SIC level using the
estimated audit fees, rather than the total compensation.

When estimated audit fees are used, the model adjusted R-square values
decrease, as do the individual coefficient values. This result is not unexpected since
many IPO cash expenses other than audit fees are likely directly related to the size
of the offering. The coefficient for membership in a specialized industry changes
signs, but remains insignificant. There is little evidence of lower fees due to
economies of scale in specialized industries. The exchange variable continues to be
significant. This is consistent with either greater audit effort or greater audit risk for
offerings on major exchanges.

As in Table 6.2, the variable CLEXP is significant. The continuous sales-
based measure SASHARE is also significant. The variable of CLSHARE is negative,
and very small. The overall results are supportive of a fee premium for auditor

experience.

 

 

 

119

. Table 6.5
IPO Audit Fee Model
Three-digit Experience Measures

 

 

 

 

 

 

 

 

(N=265)
Estimated coefficient
(t-statistic)
Experience Measure

Independent
variables Clshare Sashare Clexp Saexp Lgexp
Intercept 89.541 68.571 71.845 79.609 83.731

(3.756)*** (3.568)*** (3.975)*** (4.495)*** (4.750)***
Experience 0364 0.780 33.705 18.690 0.872

(-.345) (1.780)* * (2.107)* * (1.069) (.044)
Proceeds 0.0005 0.0005 0.0005 0.0005 0.0005

(5.442)* * * (5.526)*** (5.455) * * * (5.480)* ** (5.430)” *
Revenue 0.00001 0.000009 0.000006 0.000008 0.00001

(1.297)* (1.168) (.980) (1.079) (1.269)
Equity 0.00007 0.00005 0.00006 0.00006 0.00007

(1.487)* (1.130) (1.343)* (1.371)* (1.445)*
Specialized 16.567 13.421 13.570 13.319 15.106
industry (.780) (.649) (.658) (.640) (.469)
Exchange 45.933 47.918 49.918 46.771 46.348

(2.346)* * * (2.464)* * * (2.566) * * * (2.398)” * (2.371)“ *
Adjusted R-
Squared 0.199 0.208 0.212 .202 .199
F-Value 11.965*** 12.614*** 12.884*** 12.183*** 11.940***
Chi-Squared 38.885 27.548 27.040 29.143 24.382

* ** (* *) (*) - Significant at the .01 (.05) (.10) level. One-tail tests used where appropriate.

 

 

 

 

 

120
6.4.2 Effects of Big 6 Mergers

In section 5.5.1 of Chapter 5, results were reported that suggested that
underpricing is greater for the Big 6 firms that were formed as a result of mergers.
If these mergers resulted in a loss of reputation capital, they may also have had an
effect on the fees for these firms. Table 6.6 reports mean and median fee residuals
for the members of the Big 6 for the period 1990-1991. Fee residuals are reported
using both total cash compensation and the estimated accounting fee.

The residuals for total cash compensation are based on Model 3B in Table
6.1. The mean fee residual of 67.56 is comparable to the Big 6 coefficient of 60.311.
The fee residual based on estimated accounting fees is from the regression model for
CLEXP in Table 6.5. The positive mean residual is largely due to the exclusion of
the experience variable in calculating the fee residual. As in Beatty, mean residuals
are larger than median residuals.

The results do not provide strong evidence of lower fees for the merged firms
of Deloitte & Touche and Ernst and Young. The mean residuals are lower than the
Big 6 average, but the medians are higher than the Big 6 average. There also

appears to be some evidence of the fee premium for Price Waterhouse documented

by Simunic [1980].

 

 

 

 

 

 

121

Table 6.6

1990-1991

 

Fee Residual by Audit Firm

 

IPO Cash Compensation

Estimated Audit Fee

 

 

 

 

 

Audit Firm Mean Mean

It (Std. Dev) Median n (std. dev.) Median

Arthur Andersen 63 102.74 8.27 62 45.50 6.17
(454.96) (172.76)

Coopers and 55 37.74 «48.03 49 13.10 -28.87
Lybrand (444.65) (169.88)

Deloitte and Touche 80 27.05 18.34 76 10.64 -5.33
(339.29) (91.40)

Ernst and Young 116 34.13 7.27 114 10.32 -7.79
(404.72) (107.13)

KPMG Peat 83 57.52 -16.96 78 -9.20 -40.88
Marwick (456.55) (114.91)

Price Waterhouse 35 278.30 38.74 32 48.20 20.09
(722.29) (136.22)

Total 432 67.56 1.81 411 15.26 -9.51
(452.43) . (129.39)

 

 

 

 

 

 

 

 

 

122

6.5 Summary and Conclusions for Auditor Compensation Tests

Hypothesis four predicted that auditor experience would be associated with
higher audit fees. The general pattern of results suggests the presence of a fee
premium for auditor experience. Significant relationships between auditor
experience and compensation were found using both total IPO compensation and
estimated audit fees. However, the relationship between auditor experience and
auditor compensation is not significant in all models. This suggests that researchers
should be careful in their selection of experience measures. Although no clear
pattern of significance exists, the dichotomous client-based measures of experience
was significant in all models. It was argued in section 3.2.1 of Chapter 3 that this
measure best captured auditor experience.

The results of this study can be compared to Palmrose [1986] and Ettredge
and Greenberg [1990]. Palmrose failed to find a fee premium for industry
experience in audit fees. One possible explanation for her inability to find a fee
premium is her use of sales-based experience measures calculated at a two-digit SIC
code level. Ettredge and Greenberg [1990] found that fee reductions were larger
with the change to a, more experienced auditor. However, first-year fee reductions
are not necessarily related to recurring audit fees, or the fees in a unique context,
such as IPOs.

The evidence of a fee premium for auditor experience indicates that
experience is valued by clients. In Chapter 4, evidence was presented that suggested
that auditor experience was positively demanded by IPO companies. The results in

this chapter suggest that auditor experienceis demanded, even though it is costly.

 

 

 

 

CHAPTER 7 - SUMMARY, CONCLUSIONS, AND
SUGGESTIONS FOR FUTURE RESEARCH
This research investigated the relationship between auditor industry
experience and auditor selection, offering underpricing, and auditor compensation
for initial public offerings during the period 1988-1991. Section 7.1 summarizes and
discusses the results of the study. These results are compared to the results of
previous studies on audit quality in section 7.2. Additional results of research
interest are summarized in section 7.3. Some of the limitations of the study and

suggestions for future research are discussed in section 7.4.

7.1 Summary of Results

Four research hypotheses about auditor experience were tested in this study.
The research hypotheses and the results of the empirical tests of the hypotheses are
included in Figure 7.1
7.1.1 Demand for Auditor Experience

The first research hypothesis predicted that IPO companies would demand
auditor experience. The results of the tests of this hypothesis are discussed in
Chapter 4. These tests indicated that auditor experience is demanded by IPO
companies, but not to a greater extent than by existing publicly-traded companies.

Tests of hypothesis one were conducted by comparing the auditor experience
measures for the IPO companies to expected values under three random selection
benchmarks. The demand for auditor experience by IPO companies exceeded the

expected experience values under a random selection model, and under a model in

123

 

 

 

 

124

 

5:83— :8853— he .CaEEzm

 

fiseaa
.525on 88898
3 no a 5m E85580 988885 52586 .88 :08 88.5 _EB 8880558
88898 3 «o v 88 Emoanmm E83585 8:255 on EB 88898 553:: 88634 ”.5

 

.585

892: 88 888 8x88 8:2 9 26 @888 88
8:588 88898 5“ 3:52 :2: 8838 oEom
.38an5

3.83885 @989 2385 80 .38. o5 :8
-825 8 52588 m we e do.“ 285580 388885

20995 oEom

.588 85 Co 8:888
@585 85 EB 888058 b82882
8 EB OE 2: .«o 88:98:: 2: “a:

 

583586
88:38 =a com E85580 380885 58585
88t898 S we o 5% 88me5 3825:me

8:896

8:895 52

850995 “oz

55:9: 8838mm 8 E QEanEoE
389:8 On: FEB 888058 38:50; on
EB 5868 85 .«o 88:88 D585 2:. “oer

.cASwLEOo On: 85
mo owe 2: EB 888058 28288: on
EB 5568 85 Co 88:898 @522: 2: ”am

.BceaEOo On: of
mo 85 85 «EB @8858 38288: on
EB 8:98 85 mo 88to98 D585 8:9 ”SE

 

88388
88:20:93 8388 85 888 83.8%
a 8 Ho: :5 .88898 @888 528980 OE

notoamsm

88898 @585 SE98 8% @868
8:58 a 82 EB 528858 OE mm

 

 

«85:50

 

$158M

 

$8529»: :88an

 

 

 

 

 

125

which Big 6 auditors are chosen in relation to their share of the total audit market.
These two results indicate that auditor experience is positively demanded by IPO
companies. However, the IPO sample measures did not exceed the expected values
for auditor experience under an industry weighted selection model. This latter result
indicates that the demand for auditor experience by IPO companies does not exceed
the demand for experience by existing publicly-traded companies.

Existing literature has indicated that there are substantial levels of auditor
concentration within industries (e.g., Eichenseher and Danos [1981]). However,
comparatively little is known about the elements of the audit market that lead to
observed concentration levels. Since IPOs represent new entrants to the audit
market for publicly-traded companies, understanding the demand for audit
experience by IPO companies is an important element of understand industry
concentration in the audit market.

7.1.2 Determinants of the Demand for Auditor Experience

The second set of research hypotheses made predictions about variables
expected to be associated with auditor experience. Based on the information quality
model in Titman and Trueman [1986], the demand for auditor experience was
predicted to be negatively related to the size and age of the IPO company, and
positively related to IPO company membership in a specialized industry.

The results of the tests of these hypotheses are also discussed in Chapter 4.
The demand for auditor experience was positively related to IPO company
membership in a Specialized industry, but was not related to IPO company age or

size. The auditor selection models had low levels of explanatory power.

 

 

 

 

 

 

 

 

 

126

The positive relationship between auditor industry experience and specialized
industries. is consistent with an increased demand for auditor experience in these
industries due to increased information risk. However, the underpricing tests in
Chapter 5 of this study found that underpricing is lower in specialized industries.
This latter result is consistent with lower information risk in specialized industries.
These two results are not necessarily inconsistent if information risk is a
multidimensional construct. For example, there could be greater uncertainty about
accounting information, but less uncertainty about the overall level of cash flows for
specialized industries.

The positive relationship between the demand for auditor experience and
membership in a specialized industry is also consistent with the finding in
Eichenseher and Danos [1981] that auditor concentration within industries is related
to industry regulation. Eichenseher and Danos speculated that the increased
concentration levels in regulated industries are due to scale economies. However,
the tests of auditor compensation in Chapter 6 of this study did not detect the
existence of lower audit fees in specialized industries.

Little relationship was found between auditor experience levels and IPO
company age. It is possible that company age is not an important element of a
company’s information environment, although Beatty [1989] has found that IPO
underpricing is negatively related to company age. The tests also may have lacked
power to detect the hypothesized relationship.

There was some evidence that the demand for auditor experience was

positively related to IPO company size. The opposite relationship had been

 

 

 

 

 

 

127

hypothesized. Although less information exists for small firms (Freeman [1987]),
IPO company size may not be an important element of the IPO company
information environment. Larger firms may also receive greater benefit from
selection of an auditor with industry experience.
7.1.3 Offering Underpricing

Hypothesis three predicted that auditor experience would be associated with
lower underpricing of the offering. There was some support for this hypothesis,
although the relationship between underpricing and most measures of auditor
experience was not significant. The dichotomous measure based on sales at the three
digit SIC level (SAEXP3) was statistically significant. In addition, when an indicator
variable for merged firms was included, both the sales-based and client—based
dichotomous measures at the three digit SIC level were significant. It was argued
in section 3.2.1 that these dichotomous measures may best capture the perception
of auditor experience. The results for these two dichotomous measures suggest that
auditor experience is associated with lower underpricing.
7.1.4 Auditor Compensation

Hypothesis four predicted that auditor experience would be associated with
increased auditor compensation. The results were generally consistent with this
hypothesis, although not all experience measures were statistically significant.

A significant relationship between auditor experience and total IPO cash
compensation was found for experience measures at the two-digit and three—digit SIC
level. Evidence of a relationship between auditor experience and compensation was

also found when the tests were performed using estimated auditor compensation,

 

 

 

 

 

128

rather than total IPO cash compensation. Consistent with the notion that
dichotomous client-based measures best capture experience, this measure was
significant in each compensation model.

Evidence of a premium in audit fees for auditor experience is an important
finding. Palmrose [1986] hypothesized that auditor experience would be associated
with higher audit fees, but did not find a significant relationship between experience
and fees. More recent evidence has tended to suggest that auditor experience is
associated with lower audit fees (Ettredge and Greenberg [1990]; Deis and Giroux
[1992b]). The existence of a fee premium is a potential determinant of the demand
for auditor experience.

There are several possible explanations for the existence of a fee premium in
IPOs, but not other settings. First, previous studies have relied upon the use of one
experience measure. These measures may not have been the most appropriate for
capturing experience. Second, IPOs are a unique setting in which auditor experience
may have greater value. Third, the nature of the size or variance of audit fees in the
IPO setting may facilitate the detection of a relationship between audit fees and

auditor experience.

7.2 Comparison to Previous Audit Quality Research

The results reported in Figure 7.1 generally support auditor industry
experience as a dimension of audit quality. However, the pattern of results is not
overwhelmingly conclusive as to the importance of auditor experience. To provide

some perspective, these results are compared to the results of previous empirical

 

 

 

 

 

 

 

 

129

research on audit quality. Figure 7.2 summarizes the results of this research and
previous audit quality research in IPO and non-IPO settings. Figure 7.3 provides a.
more detailed comparison of the results of this study to the results of previous audit

quality research in IPO and non-IPO settings.

 

 

 

 

 

 

 

 

 

 

130
Previous IPO Previous non-

Dimension of Audit Quality IPO Audit
Audit Quality Current Study Research Quality Research
Demand for
Audit Quality Supported Supported Supported
Determinants of
Demand for Weak Support Mixed Support Mixed Support
Audit Quality .
Economic Effects
of Auditor Some support Supported (1) Not supported
Choice
Auditor
Compensation Supported Supported (1) Supported

 

 

 

 

 

 

 

(1) Results in current research suggest that relationship is less strong after
controlling for offering exchange effects.

FigUre 7.2
Comparison of Research Results to
Previous Audit Quality Research

 

 

 

 

 

131

 

8.883: 3:80 «:03: 888.5
0: 838% 8.883— .? camiwmfiev 8:52:
5:888

 

. .58:
88% :8 88¢ ”Ema: 858883

88888 8% a 0288: 0 mm:

.280:
5888 88: 888 8388: o wmm

88:88
8:: a 0380:
5:988 808:0me

:0._85:0:800
88:3

 

.258: a a

88:36 89:08 0 Em 3 8:68
883800 :0: 888E: 3:88:85
. :0: 588:8 0: 8088: 8082

:58:
8:5 :08 22:27: 582. c :8
a 0: 8888 0: :088: 80:88 235

.280:
380m ”Ema: .8 8 582:3 5:88
e mi 8 HEB 8B0: 8 88:08:83

80503:

808898 22B
:0B0_ 5: 88:88::
8:: 8888 0:83

5:800 £026.
:0 580:5 0:80:08:

 

.258: EEB
:8 888% 5500 8:08 _EB

888058 :28»; 8888 5:854:

.250: 880

:8 8858885 @288 8:88
8:88 :08 80888: 85:08
0: 8:20: 8:888:00 0 8m

.68: .3 a 588:0
05: SE 0: 0828: :0: 888 5:636:

.58:
588E? :08 :08sz 8888:

80:83 808508: 8.8 888:0
m0 890 .85 0: 888: 888 5:83:

85:08 888825
a 8 92582808
88800 On:

0: 888: 888098
5:88 :08 0:88Q

5:80
£084 :0» UcwEoQ
:0 5:888:88:

 

65::

885885 8 5083 888800
82: 25:: :o 85 508

0 m5 68888 5:888 c Em

:58: a no 8280 :88“
0 mm: 88: 88880 8088800 88,:

:o 8.5: 8:88: 29:03 o :8

80:08:80
5:00:08 80:88:98:

880
:83: :0: :888Q

 

 

533.8 8:55 8.3. :50

 

8:8QO On: m=£>8m

 

raw—5m «88:0

 

b580, H0 50855::—

 

 

 

 

 

 

132

Figure 7.2 indicates that the results of the current study are consistent with
most of the results of previous audit quality research. There are two main areas of
difference. Previous IPO and non-IPO research has been able to explain a greater
portion of the demand for audit quality than the current study. Also, previous IPO
research has found stronger evidence of a relationship between offering underpricin g
and audit quality.

Comparison of the results of this study with previous research into the
determinants of the demand for audit quality indicates more similarities than
differences. Much of the explanatory power in previous models relates to the type
of offering and reputation of the investment banker. Simunic and Stein [1987] and
Menon and Williams [1991] have argued that these variables should be associated
with a demand for increased auditor credibility. However, these two variables are
not clearly linked to'a theory of audit quality.

As in this study, previous research has not been successful in linking the
demand for audit quality to specific theories. Feltham, Hughes and Simunic [1991]
were unable to find a relationship between the demand for audit quality and firm
risk in an IPO setting. Little relationship was found between the demand for audit
quality and agency costs in a non-IPO setting (Francis and Wilson [1988]).

Previous research (Balvers, McDonald and Miller [1988]; Beatty [1989]) has
found a significant relationship between auditor reputation and measures of auditor
quality. However, the-results reported in Table 5.2 suggest that the reduction in
underpricing associated with use of a Big 6 auditor is less significant than reported

in previous research. The results in Table 5.3 and Table 5.6 suggest that a

 

 

 

 

133

relationship may exist between dichotomous measures of experience and
underpricing. The magnitudes of the relationships between dichotomous measures
of auditor experience and underpricing are comparable to the relationship between

underpricing and use of a Big 6 auditor reported in Table 5.2.

7.3 Other Results

In addition to the main research hypotheses, several other reported results are
of research interest. These results include the correlation among the various
measures of experience, the effect of the exchange of the offering on offering
underpricing and compensation models, the effect of mergers on the reputational
capital of the Big 6 firms, and the replication of studies of auditor selection, auditor
compensation and offering underpricing.
7.3.1 Measurement of Auditor Experience

As discussed in section 3.2.1, auditor experience measures can be categorized
along three dimensions: (1) sales-based or client-based measures, (2) the SIC code
level used (two, three, or four digit), and (3) dichotomous or continuous measures.
Absent a clear theory as to what constitutes industry experience at the firm level, any
definition may have empirical validity. The tests in this study were performed using
multiple experience measures that capture all three dimensions. Previous studies
have generally tested only one experience measure.

As reported in Table 3.1, the correlation between the measures of experience
varies substantially. Reported results in the empirical tests also vary substantially,

dependent upon the experience measure used. Ultimately. researchers studying the

 

 

 

 

 

 

134

role of audit firm experience need to deve10p a theory of the demand for different
types of experience. For example, the dichotomous measures may best be
categorized as signals of audit quality. An argument based on signaling would
appropriately use dichotomous experience measures. Prior to development of a
theory of the demand for auditor experience, researchers should justify their choice
of experience measures.

7.3.2 Exchange Effects

 

This study introduced the exchange of the offering as an explanatory variable
to IPO research. Although research in finance has identified the exchange as an
element of liquidity, and capital markets research has recognized differences between
exchanges (Grant [1980]; Atiase [1987]), accounting research has not recognized
differences within the NASDAQ market.

I The exchange of the offering was statistically significant in models of offering
underpricing and IPO compensation. The exchange of the offering is also correlated
with the reputation of the auditor and investment banker. Controlling for the
exchange of the offering reduced the reported coefficients for auditor reputation in
the underpricing and compensation models. This result suggests that the relationship
between audit quality and offering underpricing and auditor compensation is likely
to be less significant than reported in previous research.

7.3.3 Effect of Mergers on Firm Reputational Capital
The mergers among the Big 8 accounting firms that occurred in late 1989 and

early 1990 were an important change to the structure of the audit market. Because

 

 

 

 

135

these mergers are recent, little empirical evidence exists as to the impact these
mergers have had on the audit market.

The results reported in Table 5.5 and 5.6 indicate that average underpricing
is higher for offerings that involve the two Big 6 audit firms formed in these mergers.
This provides preliminary evidence that the reputational capital of these firms was
reduced as a result of these mergers. Additional research is needed to address the
effect of these mergers on the audit market.

7.3.4 Replication of Previous Research

Previous research into the role of audit quality and auditor selection, offering
underpricing and auditor compensation has generally examined each dimension of
audit quality independently. In this study, all three dimensions of audit quality were
examined. The results of these tests were largely consistent with the results of
previous research. The replication of three dimensions of audit quality in one study
is an important contribution. Because the dimensions of audit quality are related,
it is desirable to consider each of the dimensions in the same setting.

Although the results of previous research were largely supported, some
important findings were not replicated, or were changed due to the inclusion of other
variables. As described in section 7.3.2, controlling for the exchange of the offering
reduced the significance of the auditor reputation variable in models of offering
underpricing and auditor selection. In addition, as shown in Table 5.2, the
interaction between auditor and investment banker reputation was not significantly

related to underpricing in this study. This interaction variable was significant in

 

 

 

136

Balvers et al. [1988]. No ready explanation exists for the differences in results,

except that the tests were performed on different samples.

7.4 Limitations and Suggestions for Future Research

This study investigated whether auditor industry experience is a relevant
aspect of audit quality. To address this issue, a context was chosen (IPOs) in which
previous research has demonstrated the importance of audit quality. This research
considered multiple dimensions of audit quality, and also considered multiple
experience measures. Limitations of this research relate to the measurement of
experience and other data constraints.

7.4.1 Measurement of Audit Firm Experience

Absent a theory of what constitutes auditor experience, this study considered
multiple experience measures. However, as with most research, the research
measures of experience are only proxies for the underlying construct of audit quality.
Noise in the measure of the construct may impede the ability of the research to
detect hypothesized relationships.

Future research may wish to directly investigate the process by which firms
establish reputations for technical expertise in a given industry. Surveys of company
officials and financial analysts is one possible approach to this issue.

7.4.2 Data Limitations

The data also impose constraints on the research. These constraints relate

to both the time period studied and the data sources utilized. The time period

studied was one in which the Big 8 audit firms were reduced to the Big 6 as the

 

 

 

 

137

result of mergers. These mergers may have introduced additional noise into the
process by which firms establish their industry reputations.

Most of the time period under study was not a particularly active time for
IPOs. This limited the size of the potential data set for testing hypotheses, especially
for the subperiods before and after the mergers.

Research extensions both backward and forward in time may be helpful. A
relationship between auditor industry experience and auditor selection and offering
underpricing may be more likely to be found in the period before the mergers that
resulted in the Big 6. Data sets used by previous researchers could be readily
modified to conduct these tests. Research using current data as it becomes available
may provide further evidence on the role of experience in the current market. The
current market structure is most relevant for auditors, clients and investors in
understanding the effects of auditor industry experience. Recent data will also be
helpful in addressing whether audit firms’ reputational capital changed as a result of

merger activity among the largest audit firms.

 

 

 

 

 

 

 

LIST OF REFERENCES

 

 

LIST OF REFERENCES

American Institute of Certified Public Accountants. AICPA Professional Standards.
Commerce Clearing House, Chicago, Illinois (1989).

Antle, R. "The Auditor as an Economic Agent." Journal of Accounting Research
(Autumn 1982): 503-527.

Atiase, R. "Market Implications of Predisclosu re Information.” Journal of Accounting
Research (Spring 1987): 168-176.

Baker, H. and R. Edelman. "The Effects on Spread and Volume of Switching to the
NASDAQ National Market System." Financial Analysts Journal (J an-Feb
1992): 83-88.

Balvers, R., B. McDonald, and R. Miller. "Underpricing‘of New Issues and the
Choice of Auditor as a Signal of Investment Banker Reputation." 1h;

Accounting Review (October 1988): 605-622.

Beatty, R. "Auditor Reputation and the Pricing of Initial Public Offerings." m
Accounting Review (October 1989):. 693-709.

Beatty, R. and J. Ritter. "Investment Banking, Reputation, and the Underpricing of
Initial Public Offerings." Journal of Financial Economics (1986): 213-232.

Belsley, D., E. Kuh and R. Welsch. Regression Diagnostics: Identifying Influential
Data and Sources of Collinearity. John Wiley and Sons (1980).

Carpenter, C. and R. Strawser. "Displacement of Auditors When Clients Go Public."
Journal of Accountancy (June 1971): 55-58.

Clarkson, P. and R. Thompson. ”Empirical Estimates of Beta When Investors Face
' Estimation Risk." The Journal of Finance (June 1990): 431-453.

Copley, P. "The Determinants of Municipal Audit Fees: Some Additional Evidence."
Research in Governmental and Nonprofit Accounting (1989): 3-23.

138

 

 

 

 

 

139

Copley, P. "The Association Between Municipal Disclosure Practices and Audit
Quality." Journal Of Accounting and Public Policy (Winter 1991): 245-266.

Cushing, B., and J. Loebbecke. "Comparison of Audit Methodologies of Large
Accounting Firms." Studies in Accounting Research No.26 (1986). American
Accounting Association, Sarasota, Florida.

Danos, P. and J. Eichenseher. "Audit Industry Dynamics: Factors Affecting Changes

in Client-Industry Market Shares." Journal of Accounting Research (Autumn
1982): 604-616.

Danos, P. and J. Eichenseher. "Long-Term Trends Toward Seller Concentration in
the US. Audit Market." The Accounting Review (October 1986): 633-650.

Datar, S., G. Feltham, and J. Hughes. "The Role of Audits and Audit Quality in

Valuing New Issues." Journal of Accountingand Economics (March 1991):
375-399.

DeAngelo, L. "Auditor Size and Auditor Quality." Journal of Accounting and
Economics (December 1981) pp. 183-199.

Deis, D., and G. Giroux. "Determinants of Audit Quality in the Public Sector." _Th__e_
Accounting Review (July 1992a): 462-479.

Deis, D. and G. Giroux. "Determinants of Audit Fees and Audit Hours Using a
Direct Quality Measure." Louisiana State University working paper (1992b).

Edelman, R., and H. Baker. "Liquidity and Stock Exchange Listing." The Financial
Review (May 1990): 231-249.

 

Eichenseher, J., and P. Danos. "The Analysis of Industry-Specific Auditor

Concentration: Towards an Explanatory Model." The Accounting Review
(July 1981): 479-492.

Eichenseher, J. and D. Shields. "The Correlates of CPA-Firm Change for Publicly-

Held Corporations." Auditing A Journal of Practice and Theory (Spring
1983): 2337.

_ Ettredge, M., and R. Greenberg. "Determinants of Fee Cutting on Initial Audit
‘ Engagements." Journal of Accounting Research (Spring 1990): 198-210.

Ettredge, M., P. Shane, and D. Smith. "Audit Firm Size and the Association
Between Reported Earnings and Security Returns." Auditing: A Journal of
Practice and Them; (Spring 1988): 29-42.

 

 

 

 

 

140

Feltham, G., J. Hughes, and D. Simunic. "Empirical Assessment of the Impact of
Auditor Quality on'the Valuation of New Issues." Journal of Accounting and
Economics (December 1991): 375-399.

Francis, J. "The Effect of Audit Firm Size on Audit Prices: A Study of the Australian
Market." Journal of Accounting and Economics (August 1984): 133-151.

Francis, J ., and D. Simon. "A Test of Audit Pricing in the Small-Client Segment of
the US. Audit Market." The Accounting Review (January 1987): 145-157.

Francis, J. and D. Stokes. "Audit Prices, Product Differentiation, and Scale
Economies: Further Evidence from the Australian Market." Journal of
Accounting Research (Autumn 1986): 383-393.

Francis, J ., and E. Wilson. "Auditor Changes: A Joint Test of Theories Relating to
Agency Costs and Auditor Differentiation." The Accounting Review (October
1988): 663-682.

Freeman, R. "The Association Between Accounting Earnings and Security Returns

for Large and Small Firms." Journal of Accounting and Economics (July
1987): 195-228.

Grant, E. "Market Implications of Differential Amounts of Interim Information."
Journal of Accounting Research (Spring 1980): 255-268.

Hoel, P., S. Port, and C. Stone. Introduction to Statistical Theory. Houghton Mifflin
Company, Boston, Massachusetts (1971).

Imhoff, E. "A Comparison of Analysts’ Accounting Quality Judgments Among CPA
Firms’ Clients." Auditing: A Journal of Practice and Theory (Spring 1988):
182-191.

Klein, B., R. Crawford, and A. Alchian. "Vertical Integration Appropriable Rents

and the Competitive Contracting Process." Journal of Law and Economics
(1978) 21: 297-326.

Leland, H., and D. Pyle. "Informational Asymmetries, Financial Structure, and
Financial Intermediation." The Journal of Finance (May 1977): 371-387.

Menon, K. and D. Williams, "Auditor Credibility and Initial Public Offerings," 111;:
Accounting Review (April 1991) pp. 313-332.

Nichols, D., and D. Smith. "Auditor Credibility and Auditor Changes." Journal of
Accounting Research (Autumn 1893): 534-544.

 

 

 

 

141

Palmrose, Z.., "Audit Fees and Auditor Size: Further Evidence," Journal of
Accounting Research Spring (1986) 'pp. 97-110.

Palmrose, Z., "An Analysis of Auditor Litigation and Auditor Service Quality," _T_hg
Accounting Review (January 1988) pp. 55-73.

Rhode, J ., G. Whitsell, and R. Kelsey. "An Analysis of Client-Industry
Concentrations for Large Public Accounting Firms." The Accounting Review
(October 1974): 772-787.

Ritter, J. "The Long-Run Performance of Initial Public Offerings." The Journal of
Finance (March 1991): 3-27.

 

Rock, K. "Why New Issues Are Underpriced" Journal of Financial Economics
(January/February 1986): 187-212.

Rubin, M. " Municipal Audit Fee Determinants." The Accounting Review
(April 1988): 248-261.

Shields, D. "Small CPA Firm Product Differentiation in the Small Business Market."
AccountirgL Organizations and Socieg (1984) 9: 61-90.

 

Simon, D. "The Audit Services Market: Additional Empirical Evidence." Auditing:
A Journal of Practice and Theory (Fall 1985): 71—78.

 

Simunic, D. "The Pricing of Audit Services: Theory and Evidence." Journal of
Accounting Research (Spring 1980): 161-190

Simunic, D. and M. Stein, Product Differentiation in Auditing: Auditor Choice in the
Market for Unseasoned New Issues, The Canadian Certified General
Accountants’ Research Foundation, Vancouver, BC. (1987).

 

Teets, W. "The Association between Stock Market Responses to Earnings
Announcements and Regulation of Electric Utilities.” Journal of Accounting
Research (Autumn 1992): 274-285.

Titman, S. and B. Trueman. "Information Quality and the Valuation of New Issues,"
Journal of Accounting and Economics (June 1986) pp. 159-172.

Ward, D., R. Elder, and S. Kattelus. "Further Evidence on the Determinants of
Municipal Audit Fees." Michigan State University working paper (1993).

Watts, R. and J. Zimmerman. Positive Accounting Theory. Prentice-Hall, Inc.,
Englewood Cliffs, NJ, 1990.

 

 

 

 

142

White, H. "A Heteroskedasticity-consistent Covariance Matrix Estimator and A
Direct Test for Heteroskedasticity." Econometrica (1980) 48: 817-838.

 

Williams, D. "The Potential Determinants of Auditor Change." Journal of Business,
Finance and Accounting (Summer 1988): 243-261.

 

 

 

 

 

APPENDICES

 

 

 

 

Description of Appendices Contents

The appendices contain the market share measures developed from the Compact

Disclosure database that were used as a proxy for audit firm experience. Market share was

measured using both number of clients and sales dollars audited. These measures were

calculated for three-digit and two-digit SIC classifications. Market share for the years 1988-

1989 was measured using the January, 1988 version of the database. This data is contained

in Appendix A. Market share for the period 1990-1991 was measured using the January,

1990 version of the database. This data is contained in Appendix B.

Appendix A includes the period when there were eight large firms referred to as the

Big 8. Appendix B includes the post-merger period in which the Big 8 were reduced to the

Big 6. These audit firms are listed using initials in the appendices. The following are the

audit firms included in each appendix, along with the initials used to refer to these firms.

Arthur Andersen
Coopers and Lybrand
Deloitte and Touche
Ernst and Young

KPMG Peat Marwick

Appendix A Appendix B
AA Arthur Andersen AA
AY Arthur Young CL
CL Coopers and Lybrand DT
DHS Deloitte, Haskins and Sells EY
EW Ernst and Whinney PM
PM KPMG Peat Marwick PW

PW Price Waterhouse
TR Touche Ross

143

Price Waterhouse

 

 

 

APPENDIX A
BIG 8 MARKET SHARES
JANUARY, 1988

 

 

 

Appendix A.1
Client-Based Market Shares
Three-Digit SIC Level
January, 1988

$.12 AA _AX £1; D___HS BE PM. M B T___0tal
104 11.5 5.8 19.2 5.8 7.7 3.8 1.9 3.8 59.6

 

 

131 15.5 7.0 11. 4 5.5 4.8 12.5 7.4 6.3 70.5
138 12.1 7. 6 6. 4 6.4 5.7 6.4 7. 0 3.8 55.4
152 14.8 11. 1 11. 1 7.4 7.4 3.7 18. 5 11.1 85.2
153 15.4 0. 0 7.7 11.5 7.7 15.4 3.8 11.5 73.1
201 14.3 14. 3 7.1 3.6 7.1 25.0 7.1 0.0 78.6
208 19.2 3. 8 7.7 0.0 11.5 19.2 23.1 3. 8 88.5
232 10.7 17. 9 3.6 7.1 21.4 3.6 7.1 10. 7 82.1
245 14.8 0.0 7.4 7.4 18.5 14.8 7.4 3.7 74.1
251 15.6 0.0 6.3 18.8 18. 8 9.4 3.1 3.1 75.0
264 16.1 0. 0 9.7 9.7 3. 2 19.4 6. 5 9.7 74.2
271 10.7 17. 9 3. 6 10.7 17. 9 7.1 14. 3 0.0 82.1
283 14.3 9.8 12. 8 6.0 8.3 11.3 6.0 9.0 77.4
284 14.3 8.2 16.3 10.2 2.0 8.2 4.1 8. 2 71.4
291 12.1 9.1 12. 1 6.1 12. 1 15.2 18.2 12. 1 97.0
307 12.7 5.1 7. 6 7.6 8. 9 8.9 15.2 8.9 74.7
331 12.7 3.6 16. 4 7.3 21 8 5.5 12.7 7.3 87.3
342 29.6 0. 0 11. 1 7.4 14.8 11.1 11.1 0.0 85.2
344 7.5 12. 5 2. 5 5.0 10.0 5.0 10. 0 7. 5 60.0
349 6. 3 9.4 12. 5 3.1 9.4 21.9 0. 0 12. 5 75.0
353 16. 7 3.7 14.8 3.7 9 3 13.0 20. 4 7.4 88.9
354 23.5 2.9 20.6 8.8 11. 8 2.9 2.9 5.9 79.4
355 5.1 17. 9 5.1 10.3 10.3 15.4 7.7 5.1 76.9
356 12.8 9. 0 3.8 7.7 17. 9 7.7 9.0 7.7 75.6
357 17.9 14.4 11.0 9.6 7. 6 8.9 8.9 7.9 86.3
358 12.1 12.1 91 6.1 15. 2 15.2 0. 0 6.1 75.8
362 8. 3 8.3 13. 9 5.6 16. 7 11.1 11.1 0.0 75.0
363 13. 8 3.4 13.8 17.2 6. 9 20.7 3.4 6.9 86.2
364 32.3 3.2 0.0 3.2 12. 9 16.1 9.7 9.7 87.1
365 14.3 3.6 7.1 10.7 14.3 3.6 3.6 7.1 64.3
366 12.3 8. 2 7.3 9.5 11. 8 10.0 7.3 9.5 75.9
367 13.8 10. 8 7. 8 7.8 9 0 12.6 8.4 7.2 77.2
369 12. 2 0.0 14. 3 6.1 14. 3 6.1 12.2 6. 1 71.4
371 9. 8 4.9 13.1 11.5 16.4 8.2 9.8 11. 5 85.2
372 12. 5 6.3 15.6 3.1 12.5 21.9 6.3 9.4 87.5
381 7. 3 ' 7.3 12.2 9.8 12.2 4.9 9.8 9.8 73.2
382 12. 5 8. 3 11.1 4.9 8.3 11.8 9.7 9.0 75.7
383 10.0 10. 0 0.0 16.7 3. 3 20.0 3.3 6.7 70.0
384 12.6 12.6 6. 7 4.2 12. 6 13.4 7. 6 7.6 77.3
386 22.9 0.0 11 .4 2.9 2.9 8.6 17 1 8.6 74.3
399 11.8 0.0 5.9 8.8 8. 8 20.6 8.8 2.9 67.6
401 11.8 2.9 8.8 17.6 14.7 17.6 17.6 0.0 91.2

144

 

145

Appendix A.1
Client-Based Market Shares
Three-Digit SIC Level
January, 1988

S_I_C. _A_A _Al 91. ____DH8 1331/ PM. E)! 113 T.__o_ta_l
421 23.0 11.5 4.9 3.3 16.4 18.0 0.0 6.6 83.6
451 24.6 18.0 1. 6 1.6 8.2 27.9 1.6 6.6 90.2

481 26.8 8.1 26. 8 4.1 4.1 8.1 3.3 1.6 82.9

 

 

 

483 9. 3 18.6 9.3 11.6 7.0 11.6 2.3 7 0 76.7
489 9. 3 5.6 5 6 22.2 3.7 13.0 5.6 11. 1 75.9
491 31. 3 1.5 17 2 27.6 0.7 8.2 9.7 0.0 96.3
492 40.2 9.8 7. 4 12.3 4.9 10.7 4.1 4.1 93.4
493 44.0 0.0 14.0 16.0 2.0 4.0 16 0 0.0 96.0
506 10.1 6.3 11. 4 5.1 11.4 6.3 5.1 5.1 60.8
508 11.7 6.3 9 4 5.5 4.7 9.4 7. 8 3. 9 58.6
509 11.1 3.7 18. 5 0. 0 3.7 7.4 11 1 11. 1 66.7
512 15.0 2.5 15. 0 12. 5 12.5 2.5 7.5 2. 5 70.0
514 20.5 4.5 4 5 2.3 6.8 9.1 2 3 18 2 68.2
531 12.8 2.1 12 8 6.4 19.1 8.5 12. 8 14.9 89.4
541 7.9 4.8 6 3 6.3 12.7 22.2 6. 3 17.5 84.1
571 4 0 0.0 16.0 0.0 12.0 8.0 32. 0 12 0 84.0
573 16.1 3.2 6.5 6.5 19.4 6.5 9.7 6 5 74.2
581 10.5 5.3 8.3 7.5 9.0 15.8 6.0 12 0 74.4
594 14.6 4.9 2.4 9.8 7.3 17.1 12.2 7.3 75.6
596 15.6 9.4 6.3 3.1 9. 4 12.5 , 3.1 12. 5 71.9
602 10.6 3.1 7.3 6 4 14. 3 23.7 4.6 4. 4 74.3
612 3.7 5.0 3.7 16 9 8 7 33.1 3.3 10. 3 84.7
614 8. 3 8.3 4.2 15.3 13 9 18.1 8.3 11.1 87.5
615 11 7 4.3 6.4 9.6 7.4 19.1 11.7 17.0 87.2
616 13.9 18.1 2. 8 9.7 5.6 25.0 2.8 0.0 77.8
621 10 6 4.7 10 6 10.6 9 4 15.3 5.9 9 4 76.5
631 1 4 2.8 14 8 2.8 20 4 16.9 4.9 14. 1 78.2
632 16 7 8.3 5. 6 0.0 22.2 19.4 8.3 5. 6 86.1
633 4.9 3.7 17 1 3.7 20.7 22.0 6.1 11.0 89.0
641 9. 5 2.4 9.5 4.8 11.9 26 2 0.0 7.1 71.4
651 11. 5 6.4 5.1 7.7 7.7 7. 7 2.6 7.7 56.4
655 6.8 1.4 8.1 13.5 6.8 14 9 6.8 6.8 64.9
671 6.2 2.5 6.8 8.0 9.3 21.6 5.6 3.1 63.0
679 8 0 5.3 7.1 6.2 8.0 11.5 5.3 5.3 56.6
701 15. 5 8.6 5.2 5. 2 1.7 10.3 3.4 10.3 60.3
737 12.1 * 9.7 9.7 10 0 6.9 11.8 9.3 9.0 78.5
739 ' 11.2 9.5 7.1 7.1 8. 7 12.5 6.0 9.3 71.4
781 9.6 9.6 2 7 0.0 15.1 9.6 9.6 8.2 64.4
794 6.7 3.3 13 3 3.3 6.7 0.0 3.3 6.7 43.3
799 81 5.4 13.5 2.7 5. 4 5.4 8.1 2.7 51.4
806 14 8 22.2 11.1 0.0 29 6 7. 4 7.4 7.4 100.0

807 7.7 7.7 3.8 7.7 11.5 19. 2 3.8 7.7 69.2

S_IC
808
891
899

Avg.

“E7
13.2
20.2

13.8

AX
10.7
13.2

3.8

6.9

146

Appendix A.1
Client-Based Market Shares
Three-Digit SIC Level
January, 1988

9L D_H_§ EVE P_M_
3.6 14.3 10.7 14.3
9.4 7.5 7.5 11.3
7.7 4.8 4.8 8.7

9.3 7.7 10.6 12.7

my
0.0
5.7

14.4

7.9

1H
13.2
5.8

7.5

MI.
78.6
81.1
70.2

76.4

 

 

 

147

Appendix A.2
Sales-Based Market Shares
Three-Digit SIC Level
January, 1988

S_I_C. & AX _C_L QH§ w m _P_W E M11
104 23.3 20.1 11.9 28.1 5.2 7.9 0. 0 0.3 96.6
131 2.9 31.0 14.2 0.2 6.0 0.6 43.0 0.2 98.0
138 67. 7 2.4 2.2 2.9 0.2 0.6 23.4 0.0 99.4
152 4.1 12.5 58.5 1.4 4.7 0.0 12.3 2.4 95.8
153 16. 4 0.0 49.3 1.7 8.4 11.1 0.3 1.8 89.0
201 8.5 24.6 1.5 0.3 11.1 17.4 4. 4 0.0 68.0
208 5. 3 22.5 1.0 0.0 26.0 1.4 43.8 0.0 100.0
232 12. 3 23.9 0.4 16.2 32.7 0.8 1. 7 8.4 96.5
245 8. 7 0.0 11.4 5.9 30.5 11.2 26.7 1.0 95.4
251 14.9 0.0 5.2 14.1 15.1 20.7 8.2 1. 3 79.5
264 6.7 0.0 45.1 2.2 3. 2 6.8 19.7 15. 5 99.2
271 2. 7 8.8 6.3 12.1 30. 3 2.1 26.9 0.0 89.2
283 22. 8 0.6 6.2 8.0 6.6 16. 9 26. 8 2.6 90.5
284 12.8 1.7 21.7 44.1 0.1 6.0 1. 1 4.2 91.8
291 4.0 1.1 12.6 2.1 31.5 5.9 41. 4 1.4 100.0
307 4.9 0.8 3.5 1.2 1. 8 2.5 77.7 2.2 94.6,
331 8. 3 1.2 7.8 9.6 45. 7 1. 0 22.2 3.7 99.5
342 13. 4 0.0 3.8 25.9 18.9 34. 9 1.6 0.0 98.5
344 1. 8 4.3 0.0 3.7 9.9 5. 9 73.1 0. 3 99.0
349 14. 8 28.9 12.8 0.1 5.4 20. 5 0.0 14.5 97.0
353 15.3 0.6 5.9 9.2 1. 6 7.2 57. 4 2.2 99.5
354 23. 0 0.3 7.3 13.6 49.0 0. 2 0.1 2.9 96.4
355 2. 8 7.0 3.1 3.6 3. 3 54.2 22. 7 0.7 97.4
356 12. 7 5.8 10.7 3.3 34.1 2.3 25.3 2.6 96.8
357 10.8 10.9 10.3 1.3 5.1 4. 2 55. 8 0.7 99.1
358 1.6 32.7 4.1 25.3 7.3 10. 3 0. 0 5.1 86.3
362 0. 2 10. 2 6.4 3.9 7. 8 54.4 16. 3 0.0 99.2
363 19. 6 0. 0 2.2 20.8 32. 8 15.5 5.5 2.9 99.3
364 48. 3 30. 9 0.0 0.0 3.7 5.1 5. 7 5.7 99.4
365 0. 6 0.0 9.2 0.1 0.1 0.1 19.0 69.8 99.0
366 14.5 19.3 3.5 13.5 11. 3 17.7 5.6 10.7 96.0
367 13.0 26.4 6.2 2.5 29.2 12.8 5.8 1.5 97.5
369 10. 3 0.0 26.4 0.1 18.2 19.6 13. 1 5.6 93.3
371 0. 3 0.6 23.0 67.7 1. 5 1.8 2. 4 2. 2 99.5
372 19.1 3.5 0.8 18.5 19.8 3.6 24.0 10. 6 99.9
381 1. 8 4.5 8.7 2.9 8. 6 _ 8.1 39.1 1.0 74.7
382 11. 5 2.2 16.6 15.7 14.0 1. 3 29.8 6. 8 97.8
383 22.5 5.0 0.0 35.9 0.5 17.0 0.2 12. 1 93.1
384 14.3 9.1 2.7 0.8 38. 5 8. 0 9. 8 13.5 96.7
386 5.4 0.0 0.1 0.3 3.0 40.5 43.5 0.2 93.0
399 5.5 0.0 1.5 3.2 1 7 25.1 8.8 0.0 45.8

401 3.3 0.2 21.9 15.4 18. 6 19.3 21.3 0.0 99.9

 

148

Appendix A.2
Sales-Based Market Shares
Three-Digit SIC Level
January, 1988

SI_C & A_Y 2L _DHS _Efl m M E _Total

421 31.4 5.9 3.7 9.2 21.6 22.3 0.0 1.1 95.1
451 38.9 27.3 0.0 0.1 13.1 10.9 1.2 0.9 92.5
481 22.6 4.9 65.3 0.4 0.1 0.9 2.1 3.4 99.7
483 2.6 43.0 38.6 2.0 1.8 4.7 0. 2 5.4 98.3
489 7. 5 1.2 1.7 34.2 10.7 7.9 29.0 4.6 96.8
491 34. 2 1.2 14.0 39.9 0.0 2.7 7.4 0.0 99.4

492 34.7 7.6 ‘ 5.2 17.0 2.4 16.5 7.6 8.1 99.0
493 43.5 0.0 15.2 17.2 3.7 1.8 18. 5 0.0 99.9

506 5.3 48.9 5.5 0.4 2.8 5.8 10.9 1.6 81.2
508 6. 2 2.1 12.4 1.8 6.2 6.4 57.6 1.7 94.4
509 95. 3 0.1 2.0 0.0 0.0 0.2 0.7 0.5 98.9
512 6. 8 0.0 12.9 64.4 8.4 0.6 5.3 0. 1 98.5
514 17. 7 2.3 3.7 4.1 3.5 12.9 2. 5 49.0 95.6
531 11.8 7.4 6.2 0.4 7.7 11.5 16. 2 38.1 99.4
541 4.4 1.1 17.6 7.4 16.9 27. 0 9. 2 12.7 96.3
571 5.6 0.0 10.3 0.0 17.2 0. 6 18. 6 7.2 59.6
573 6. 4 0.0 11.1 0.8 8.9 15. 4 45. 8 1. 3 89.8
581 19.1 16.9 2.6 10.1 8.0 9.2 2. 7 25. 0 93.5

594 32.1 4.2 1.1 5.4 3.0 6.4 12. 2 27.0 91.4
596 31.1 0.7 3.2 0.0 11.4 4. 6 1. 5 6.9 59.4

602 7.4 2.7 6.1 3.8 22.4 35.1 17. 0 2. 6 97.0
612 2.5 6.6 0.9 14.6 3.0 44.4 8.9 15 6 96.5
614 0. 8 3.8 0.0 43.5 13.3 4.7 9.4 24.2 99.8
615 13 8 7.7 33.3 6.6 4.6 8 7 5.9 17.8 98.4
616 22.0 11.1 0.1 12.5 22.7 12 9 0.6 0.0 81.9
621 22.7 11.2 23.9 26.2 1.1 6. 8 0.4 2 0 94.1
631 0.1 1.6 13.2 0.7 27.3 41.3 2.5 10. 8 97.5
632 2. 4 6.8 14.1 0.0 42.6 30.8 0 8 1. 2 98.6
633 15. 0 3.7 19.9 3.0 11.8 12.8 15 9 16 6 98.7
641 32.3 0.0 27.3 17.2 1.4 6.8 0.0 12.9 98.0
651 1.2 0.9 0.0 0.6 0.1 2 9 0.2 71.0 77.0
655 7. 0 1.9 0.8 9.6 6 5 37. 5 7.0 20.0 90.4
671 10 0 3.5 2 2 3.5 13 8 36.9 26.5 0.2 96.6
679 21.9 2.3 13 5 4.5 8.8 15.4 4.6 17.0 87.9
701 33.1 1.4 1 1 7.8 0.2 1.0 0.3 30.6 75.5
737 6. 0 4.9 44. 8 9.9 2.9 8 3 8.6 12.8 98.2
739 21 7 5.7 30.2 0.9 8.4 17. 5 7. 4 5.4 97.3
781 25.2 5.3 0.1 0.0 19.8 1.1 29. 1 1.9 82.4
794 20.6 5.7 7. 9 1.9 7.4 0.0 2 1 6.6 52.3

799 21.0 0.9 19. 2 1.5 3.1 4 0 44. 0 0.3 94.0
806 22.6 5.3 21.0 0.0 33. 3 l6 1 0.1 1.7 100.0
807 3.8 6.1 34.8 7.9 34.4 3.8 0.0 2.2 92.9

 

 

 

g;
808
891
899

Avg.

4.9
34.6
26.0

15.5

_A_Y_
7.9
12.3
10.3

7.6

149

Appendix A.2
Sales-Based Market Shares
Three-Digit SIC. Level
January, 1988

EL RES .E_W _P_M
0.0 43.4 20.7 2.1

22.0 1.7 5.7 7.7
6.1 7.0 2.8 1.2

11.7 10.1 12.3 11.8

P_W

0.0
2.5
11.3

15.2

IR;
19.4

4.0
31.5

8.5

19121
98.5
90.3
96.3

92.7

 

 

 

150

Appendix A.3
Client-Based Market Shares
_ Two-Digit SIC Level
January, 1988

519. AA A1 9L ____DHS BE EM. BE 13 T. _0_t_a.l_

 

 

10 11.9 4.5 17.9 9.0 9.0 3.0 3.0 6.0 64.2
13 15.0 7.4 9.8 6.1 5.1 11.0 6.4 5.4 66.2
15 . 20.0 6.2 9.2 7.7 7.7 10.8 9.2 9.2 80.0
17 9.4 9.4 0.0 3.1 12.5 15.6 0.0 9.4 59.4
20 10.8 8.9 10.1 3.8 8.9 13.3 13.9 5.1 74.7
22 17.1 7.9 3.9 6.6 21.1 7.9 7.9 3.9 76.3
23 14.1 8.5 4.2 4.2 12.7 5.6 4.2 9.9 63.4
24 23.1 1.9 5.8 9.6 15.4 9.6 5.8 3.8 75.0
25 13.8 3.4 5.2 8.6 19.0 6.9 6.9 3.4 67.2
26 18.2 2.6 14.3 10.4 3.9 14.3 10.4 6.5 80.5
27 9.8 7.5 6.8 9.8 9.8 9.0 10.5 5.3 68.4
28 13.7 7.4 11.3 8.1 10.2 10.9 7.7 6.7 76.1
29 14.6 7.3 12.2 7.3 12.2 9.8 19.5 9.8 92.7
30 10.9 5.5 8.2 9.1 9.1 10.0 13.6 9.1 75.5
31 17.2 0.0 6.9 6.9 6.9 10.3 10.3 13.8 72.4
32 13.2 14.7. 10.3 4.4 10.3 2.9 11.8 5.9 73.5
33 13.2 5.5 15.4 7.7 17.6 6.6 9.9 5.5 81.3
34 13.4 7.3 10.4 4.3 12.2 10.4 9.1 7.3 74.4
35 16.2 11.5 10.3 8.2 10.3 10.3 8.6 7.5 82.9
36 14.3 7.5 8.0 8.6 11.4 10.5 7.7 8.0 75.9
37 12.9 8.1 12.1 8.1 16.1 12.9 8.1 8.9 87.1
38 12.8 9.7 8.6 5.8 9.2 11.5 9.2 8.9 75.7
39 10.3 2.3 9.2 8.0 12.6 17.2 5.7 3.4 69.0
40 11.8 2.9 8.8 17.6 14.7 17.6 17.6 0.0 91.2
42 22.7 12.1 4.5 3.0 15.2 18.2 1.5 6.1 83.3
44 20.0 4.0 8.0 8.0 20.0 24.0 8.0 4.0 96.0
45 23.9 16.4 3.0 3.0 ~ 7.5 28.4 1.5 7.5 91.0
48 19.1 9.5 18.2 10.5 4.5 10.0 4.1 4.5 80.5
49 34.9 4.6 11.5 18.4 2.9 11.2 8.6 2.0 94.2
50 10.6 6.7 10.0 4.6 7.9 8.2 7.9 6.7 62.6
51 13.9 4.6 11.3 6.0 12.6 5.3 5.3 9.9 68.9
52 14.3 7.1 3.6 10.7 10.7 17.9 14.3 10.7 89.3
53 10.9 3.1 12.5 10.9 14.1 7.8 14.1 12.5 85.9
54 6.6 3.9 6.6 5.3 13.2 18.4 6.6 14.5 75.0
56 6.7 0.0 8.3 1.7 6.7 18.3 10.0 25.0 76.7
57 10.3 1.7 12.1 3.4 17.2 6.9 19.0 8.6 79.3
58 10.5 ‘ 5.3 8.3 7.5 9.0 15.8 6.0 11.3 73.7
59 18.5 5.6 4.0 5.6 4.8 19.4 6.5 8.9 73.4
60 10.3 3.1 7.4 6.3 14.1 24.0 4.5 4.5 74.1
61 7.5 7.3 4.2 14.4 8.8 26.9 5.8 9.8 84.6

62 8.1 3.6 9.0 9.0 8.1 16.2 8.1 9.0 71.2

 

151

Appendix A.3
Client-Based Market Shares
Two-Digit SIC Level
January, 1988

E AA AX Q]: _DHS .E_W_ RM BV_V TR _TOtal

63 5.6 4.3 16.3 3.4 24.0 24.0 7.3 13.3 98.3
64 9.5 2.4 9.5 4.8 11.9 26.2 0.0 7.1 71.4
65 8.3 3.3 6.1 10.0 7.2 11.1 4.4 7.2 57.8
67 6.5 4.1 6.8 7.5 9.2 16.4 5.1 4.1 59.9
70 15.3 6.8 5.1 6.8 0.0 10.2 5.1 10.2 59.3
72 9.4 6.3 3.1 6.3 3.1 9.4 6.3 15.6 59.4
73 12.2 8.9 7.9 8.7 7.7 11.9 8.3 8.6 74.2
78 11.1 7.8 2.2 1.1 13.3 10.0 10.0 7.8 63.3
79 6.4 3.8 11.5 5.1 5.1 2.6 7.7 5.1 47.4
80 10.6 10.6 6.1 9.1 14.4 13.6 3.8 9.8 78.0
89 17.0 7.0 7.6 6.4 6.4 11.1 11.7 8.2 75.4

Avg. 13.4 6.2 8.5 7.3 10.7 12.9 8.0 8.0 75.1

 

 

152

Appendix A.4
Sales-Based Market Shares
Two-Digit SIC Level
January, 1988

§l§ _A_A; Al 9L DHS _Efl PM. Pl! 113 T___0tal

10 12.3 3.8 19.3 40.8 7.4 1.5 13.6 1.0 99.7
13 11.3 27.9 12.9 0.5 5.0 0.6 40. 9 0.2 99.2
15 20. 6 5.6 45.3 1.3 5.6 6. 7 5. 2 1.7 92.0
17 3. 4 10.5 0.0 32.6 11.0 12. 3 0. 0 11.2 80.9
20 12. 1 12.2 9.6 2.2 12.8 14.8 23.2 6.2 93.2
22 15.1 23.9 1.8 9.3 20.0 18.1 6.8 0.6 95.5
23 16.0 14.4 1.1 18.0 20.0 7.0 1.8 6.9 85.3
24 71.6 0.0 1.3 1.6 14.7 2. 3 4. 4 0.3 96.2
25 18.6 3.4 3.4 6.8 21. 1 16. 0 10. 4 1.8 81.5
26 28.7 1.3 27.1 9.8 4. 2 6.6 16.1 5.3 99.0
27 15.7 12.1 4.7 11.0 25. 9 3. 4 16.2 3.0 92.0
28 15.2 0.9 11.3 26.4 6. 5 14. 0 13.8 2.1 90.3
29 6.6 1.6 17.6 2.9 12. 1 7.5 49.6 2.0 100.0
30 3.6 4.9 12.0 3.3 4.4 4.3 62.4 1.3 96.2
31 7. 4 0.0 10.3 1.0 8.8 4.4 35.0 3.7 70.6
32 28. 3 19.8 15.7 1.7 12.7 1.2 11.8 2.1 93.3
33 8.3 2.3 20.7 11.8 34.0 1.0 17.3 2.7 98.2
34 6.2 4.1 18.4 5.5 14. 1 11.1 28.8 11.0 99.1
35 11.9 9.1 6.7 4.6 7. 2 22.1 34.6 1.4 97.7
36 15.4 16.1 4. 5 8.5 16. 4 17.1 6. 5 13.6 98.1
37 2.7 2.5 11. 0 20.2 53. 7 1.8 5. 8 2.2 99.9
38 7.7 2.2 4.9 4.3 9. 5 25.3 38.0 3.3 95.3
39 4.4 8.1 8.4 4.2 10. 2 23.1 12.5 0.2 71.1
40 3. 3 0.2 21.9 15.4 18.6 19.3 21.3 0.0 99.9
42 37.5 5.3 3.3 8.2 19.3 20. 8 0.3 1.0 95.6
44 33.0 0.6 1.1 12.1 36.9 6. 7 7.5 2.1 99.9
45 38.8 27.1 0.1 0.1 13.0 11. 1 1.2 1.1 92.5
48 21.6 7.2 63. 8 1.4 0.4 1.3 3.1 0.7 99.5
49 38.3 3.4 10.8 28.6 1.6 6.6 7.4 3.0 99.6
50 46. 4 14.0 5.5 2.5 5. 0 4.6 11.9 3.2 93.2
51 7. 3 33.6 4.8 11.8 16. 1 4. 7 4.0 14.9 97.2
52 31.9 2.3 1.1 16.0 15. 0 16. 5 5. 3 11.4 99.6
53 11. 0 6.9 7. 0 3.5 7. 0 10.5 18. 5 34.7 99.1
54 4. 4 1.1 17. 7 7.3 16.8 27.0 9.2 12.7 96.2
56 10. 3 0.0 8. 3 1.2 7.1 39.6 6. 9 18. 5 91.9
57 3. 5 0.0 46.7 0.3 6.8 6.0 23. 5 ' 1. 8 88.7
58 19. 1 17.0 2.6 10.2 8.0 9. 2 2.7 24.7 93.5
59 36.7 1.5 1.5 2.6 2. 4 28.3 5. 8 11. 9 90.6
60 7.3 2.6 6.0 3.7 22. 6 35.1 16. 9 2. 6 96.8
61 3. 9 5.7 4. 7 25.2 8.0 22.7 8.4 18. 8 97.6

62 22. 0 10.9 23. 5 25.5 1.1 6.9 2.0 2.1 94.0

 

153

Appendix A.4
Sales-Based Market Shares
Two-Digit SIC Level
January, 1988

E AA fl C_L _DHS EV! m M TR _Total

63 8.4 3.0 16.8 1.9 19.5 25.1 10.1 13.4 98.2
64 32.3 0.0 27.3 17.2 1.4 6.8 0.0 12.9 98.0
65 2.3 1.0 1.8 2.6 1.6 20.6 2.2 50.4 82.6
67 12.5 9.0 3.7 4.4 16.3 43.3 4.5 1.2 95.1
70 42.9 0.3 1.4 10.1 0.0 1.3 0.4 39.6 96.0
72 16.5 17.1 5.3 0.3 0.4 9.5 5.1 26.4 80.7
73 12.1 2.9 23.4 3.7 3.4 8.1 8.2 5.4 67.2

78 42.5 3.3 0.1 2.4 12.7 0.7 21.3 5.7 88.7
79 20.2 1.1 17.8 2.1 3.3 3.6 40.9 1.1 90.2
80 18.1 12.1 14.9 8.3 26.5 11.4 0.7 6.4 98.3
89 26.6 10.4 7.1 6.7 2.9 1.7 10.8 29.8 95.9

Avg. 18.3 7.4 11.9 8.9 12.1 12.1 13.8 8.5 93.1

APPENDIX B
BIG 6 MARKET SHARES
JANUARY, 1990

Appendix 31
Client-Based Market Shares
Three-Digit SIC Level
January, 1990

S_IC_ & _Cl- DI E_Y 11M P_W M
104 8.5 23.4 12.8 8.5 6.4 0.0 59.6
131 16.7 12.4 11.0 11.0 15.2 7.1 73.3
138 14.7 11.9 8.3 12.8 11.0 3.7 62.4
201 14.8 11.1 11.1 18.5 25.9 7.4 88.9
232 15.4 0.0 23.1 30.8 7.7 7.7 84.6
264 14.8 11.1 14.8 7.4 14.8 7.4 70.4
271 16.0 4.0 16.0 28.0 12.0 12.0 88.0
283 17.9 17.1 10.3 13.7 16.2 6.0 81.2
284 11.1 17.8 8.9 8.9 17.8 4.4 68.9
291 13.3 10.0 16.7 23.3 13.3 20.0 96.7
307 12.7 7.0 14.1 14.1 11.3 14.1 73.2
331 14.0 14.0 16.0 28.0 8.0 12.0 92.0
344 6.3 0.0 18.8 21.9 9.4 12.5 68.8
349 6.9 17.2 13.8 20.7 13.8 0.0 72.4
353 17.4 10.9 13.0 8.7 15.2 23.9 89.1
354 22.6 19.4 12.9 19.4 3.2 3.2 80.6
355 7.9 10.5 15.8 18.4 13.2 7.9 73.7
356 15.9 5.8 14.5 23.2 13.0 8.7 81.2
357 16.0 10.5 16.8 23.8 12.1 9.4 88.7
358 7.4 11.1 3.7 40.7 22.2 0.0 85.2
362 10.0 6.7 10.0 23.3 13.3 13.3 76.7
363 14.8 14.8 14.8 14.8 25.9 3.7 88.9
364 32.0 0.0 12.0 12.0 16.0 8.0 80.0
366 14.8 7.1 21.9 18.4 12.2 6.6 81.1
367 14.7 10.7 10.7 25.3 14.0 8.0 83.3
369 11.3 11.3 9.4 17.0 11.3 5.7 66.0
371 13.5 17.3 26.9 23.1 3.8 7.7 92.3
372 20.7 13.8 17.2 10.3 24.1 3.4 , 89.7
381 7.7 20.5 20.5 15.4 0.0 5.1 69.2
382 13.7 9.7 17.7 16.1 12.9 9.7 79.8
383 11.5 0.0 23.1 11.5 19.2 3.8 69.2
384 10.9 7.6 16.0 22.7 16.8 7.6 81.5
386 18.5 11.1 14.8 7.4 11.1 14.8 77.8
394 6.9 3.4 17.2 17.2 13.8 6.9 65.5
399 10.3 6.9 20.7 10.3 24.1 3.4 75.9
401 12.5 9.4 15.6 15.6 21.9 12.5 87.5
421 21.7 4.3 4.3 41.3 23.9 0.0 95.7
451 26.1 2.2 8.7 30.4 23.9 2.2 93.5
481 19.2 29.8 6.7 16.3 7.7 3.8 83.7

155

Appendix B.1
Client-Based Market Shares
Three-Digit SIC Level
January, 1990

81$ AA C_L m‘ H EM PAM M
483 10.8 10.8 16.2 27.0 13.5 0.0 78.4
489 11.1 6.7 28.9 13.3 11.1 4.4 75.6
491 33.3 16.7 29.7 1.4 7.2 8.7 97.1
492 41.8 5.5 20.0 x 12.7 10.0 4.5 94.5
493 45.5 10.9 16.4 3.6 3.6 18.2 98.2
506 8.2 8.2 13.1 18.0 13.1 9.8 70.5
508 9.2 10.1 12.8 5.5 17.4 7.3 62.4
509 7.4 11.1 22.2 7.4 11.1 7.4 66.7
512 12.8 12.8 17.9 17.9 2.6 15.4 79.5
514 16.3 2.3 18.6 11.6 16.3 2.3 67.4
531 7.9 10.5 26.3 26.3 10.5 13.2 94.7
541 12.7 7.3 23.6 16.4 23.6 5.5 89.1
571 4.0 12.0 16.0 16.0 4.0 24.0 76.0
573 14.3 14.3 17.9 32.1 3.6 7.1 89.3
581 7.3 7.3 24.5 15.5 15.5 3.6 73.6
594 17.6 2.9 20.6 11.8 17.6 8.8 79.4
596 19.2 7.7 11.5 19.2 7.7 7.7 73.1
602 9.8 7.8 12.5 15.1 26.5 5.0 76.8
612 1.6 2.4 26.8 14.4 36.0 3.2 84.4
614 11.1 3.7 18.5 37.0 11.1 3.7 85.2
615 14.1 4.7 20.3 10.9 15.6 14.1 79.7
616 13.1 3.3 14.8 19.7 27.9 3.3 82.0
621 9.1 11.7 22.1 16.9 13.0 3.9 76.6
631 2.6 14.9 16.7 18.4 21.9 5.3 79.8
632 9.7 6.5 6.5 29.0 22.6 9.7 83.9
633 6.3 17.7 11.4 27.8 25.3 5.1 93.7
641 ' 11.4 8.6 11.4 14.3 22.9 0.0 68.6
651 10.0 5.0 21.7 10.0 10.0 5.0 61.7
655 3.3 10.0 23.3 13.3 11.7 5.0 66.7
671 9.0 9.0 14.0 10.0 21.0 7.5 70.5
679 5.3 6.0 12.0 9.8 8.3 6.0 47.4
701 15.9 0.0 15.9 6.8 15.9 2.3 56.8
737 12.1 10.1 16.2 17.0 10.9 9.7 76.1
739 11.3 8.0 12.5 16.7 14.0 4.8 67.3
781 10.6 1.5 7.6 24.2 13.6 7.6 65.2
799 p 13.5 13.5 8.1 10.8 5.4 8.1 59.5
806 18.5 11.1 3.7 51.9 7.4 7.4 100.0
809 4.0 16.0 12.0 28.0 4.0 8.0 72.0
891 14.6 12.5 18.8 22.9 10.4 6.3 85.4

Avg. 13.6 9.7 15.7 18.1 14.1 7.3 78.5

156

Appendix B.2
Sales-Based Market Shares
Three-Digit SIC Level
January, 1990

§I_C AA 91.: m H EM. BM 1321
104 36.3 20.6 21.7 7.9 11.7 0.0 98.3
131 16.0 9.8 0.1 21.0 0.4 52.6 99.9
138 60.0 25.0 1.1 1.0 0.3 12.5 99.9
201 32.6 2.6 20.6 25.7 13.5 3.1 98.1
232 34.8 0.0 10.1 48.5 1.3 2.7 97.4
264 6.9 44.2 17.5 2.8 7.0 21.0 99.4
271 3.6 8.0 12.3 38.0 2.5 34.5 98.8
283 29.2 14.8 7.7 7.3 23.1 17.0 99.1
284 10.2 23.2 44.9 1.2 19.0 1.2 99.7
291 4.3 12.0 3.3 18.3 6.1 56.1 100.0
307 9.1 4.0 3.3 4.4 3.5 69.2 93.5
331 7.4 7.9 11.9 40.2 2.5 29.9 99.8
344 7.9 0.0 6.0 8.1 7.0 70.1 99.0
349 11.3 14.6 23.0 27.2 20.2 0.0 96.4
353 48.6 4.0 6.0 0.8 4.6 35.9 99.8
354 21.7 6.3 12.5 57.7 0.2 0.5 99.0
355 3.2 4.9 4.4 8.4 48.0 29.2 98.0
356 12.0 16.8 5.9 24.9 7.4 31.3 98.3
357 4.5 10.4 2.1 20.3 4.7 57.4 99.4
358 2.8 7.0 6.0 50.4 18.7 0.0 84.9
362 0.1 3.7 5.4 21.5 55.5 13.2 99.5
363 12.1 1.4 11.4 46.2 17.2 10.1 98.3
364 50.0 0.0 6.2 33.7 4.4 4.9 99.3
366 9.1 6.0 27.7 33.3 18.7 4.2 99.0
367 10.7 8.2 3.0 59.9 13.0 4.6 99.4
369 20.1 3.6 6.4 19.8 24.4 23.6 97.9
371 0.4 33.3 60.3 2.9 0.3 2.7 99.9
372 20.4 0.6 39.7 16.8 2.5 19.9 100.0
381 3.2 28.4 20.9 10.1 0.0 12.8 75.4
382 11.6 16.2 23.2 12.4 1.6 33.1 98.1
383 31.3 0.0 51.5 4.9 10.4 0.1 98.2
384 12.2 5.2 14.4 39.5 17.9 7.2 96.5
386 2.4 0.1 0.3 1.1 48.3 47.5 99.9
394 0.7 3.5 3.5 26.9 36.4 26.5 97.5
399 5.5 1.0 28.7 2.4 53.9 8.4 99.8
401 3.7 21.8 16.0 21.2 26.5 10.8 100.0
421 42.9 2.3 2.2 35.2 16.5 0.0 99.2
451 46.8 0.0 1 0 39.8 10.4 1.3 99.3

481 19.0 69.1 0.6 7.9 0.3 2.6 99.5

157

Appendix B.2
Sales-Based Market Shares
Three-Digit SIC Level

January, 1990

.SI_C AA. C_L RI 131 m KW. M
483 6.5 29.3 3.9 54.7 3.6 0.0 97.9
489 6.0 1.1 22.8 12.6 33.2 23.7 99.4
491 33.4 10.7 44.2 1.1 2.7 7.4 99.6
492 34.8 5.5 35.9 9.7 12.9 1.0 99.8
493 42.7 9.2 18.4 4.5 1.5 23.7 99.9
506 6.6 1.6 1.0 50.1 7.3 13.8 80.5
508 10.5 24.7 12.5 11.8 26.7 9.3 95.5
509 4.0 45.0 12.3 8.6 8.4 10.6 89.0
512 4.6 9.9 48.6 29.0 0.5 6.3 98.8
514 13.1 4.2 56.4 3.5 18.4 1.9 97.5
531 8.6 4.9 36.9 19.6 12.0 17.9 99.9
541 6.7 18.2 26.0 18.3 21.5 6.8 97.4
571 1.2 5.0 39.7 16.9 0.3 15.9 79.2
573 4.4 14.5 8.6 10.3 14.5 45.2 97.5
581 6.4 6.0 39.0 33.2 11.8 1.2 97.5
594 31.3 2.4 39.0 7.1 7.6 , 6.5 93.9
596 35.0 2.4 3.0 8.0 19.6 3.3 71.3
602 12.0 6.1 5.2 19.2 32.7 23.2 98.3
612 0.8 0.6 29.4 11.0 46.6 8.6 97.0
614 11.2 0.0 65.8 14.2 5.1 3.4 99.7
615 15.5 44.7 15.3 9.2 5.3 8.2 98.1
616 21.5 0.0 11.6 46.9 8.3 0.5 88.9
621 9.5 15.9 19.8 44.2 4.1 2.0 95.4
631 0.6 10.5 14.4 22.6 45.3 3.6 97.0
632 1.7 17.3 1.0 48.1 29.6 0.8 98.5
633 19.4 16.9 16.1 23.3 11.0 12.5 99.2
641 35.1 20.9 26.6 1.0 13.9 0.0 97.5 '
651 0.8 0.0 92.5 1.2 4.0 0.2 98.7
655 10.8 3.7 31.2 10.0 31.1 1.8 88.5
671 13.9 5.1 22.0 29.2 13.2 11.6 95.0
679 20.5 4.2 23.8 9.0 17.7 12.0 87.2
701 49.2 0.0 45.5 0.3 1.9 0.0 97.0
737 6.0 49.8 20.0 9.4 6.3 7.2 98.7
739 17.5 35.6 7.3 18.9 11.0 5.9 96.2
781 7.0 0.1 0.3 56.7 6.7 22.2 93.1
799 16.1 9.2 1.3 21.6 4.2 45.2 97.5
806 22.7 21.1 1.3 36.7 17.9 0.3 100.0
809 0.3 4.3 10.9 38.6 6.9 28.0 89.0
891 17.5 21.6 19.0 26.5 8.3 3.9 96.8

Avg. 15.8 11.8 18.9 21.1 14.0 14.8 96.4

158

Appendix B.3
Client-Based Market Shares
Two-Digit SIC Level
January, 1990

SI_C AA _C_L QT .13.! _P_M PAM T_0t§l
10 10.6 22.7 16.7 9.1 4.5 1.5 65.2
13 16.0 12.2 10.0 11.6 13.8 6.0 69.6
15 20.3 8.5 11.9 13.6 11.9 6.8 72.9
17 16.1 3.2 16.1 12.9 16.1 3.2 67.7
20 7.7 9.2 14.1 16.9 16.2 14.8 78.9
22 15.6 4.7 15.6 29.7 9.4 3.1 78.1
23 11.3 0.0 19.7 19.7 11.3 4.2 66.2
24 27.3 6.8 6.8 15.9 11.4 6.8 75.0
25 18.8 6.3 8.3 27.1 6.3 4.2 70.8
26 18.8 13.0 17.4 7.2 13.0 10.1 79.7
27 12.9 8.9 13.9 17.8 12.9 8.9 75.2
28 14.0 16.3 12.0 15.5 17.1 8.1 82.9
29 18.4 7.9 13.2 23.7 13.2 21.1 97.4
30 13.1 9.1 15.2 14.1 12.1 12.1 75.8
31 21.4 7.1 21.4 7.1 10.7 10.7 78.6
32 13.3 8.3 13.3 26.7 6.7 13.3 81.7
33 13.8 13.8 14.9 25.3 10.3 9.2 87.4
34 11.4 10.7 15.0 21.4 9.3 8.6 76.4
35 16.0 10.1 14.8 22.3 13.0 9.5 85.6
36 15.1 8.2 14.9 19.7 13.6 6.9 78.5
37 16.4 12.9 19.8 22.4 12.9 5.2 89.7
38 12.4 9.2 17.9 17.6 13.3 8.4 78.7
39 9.9 7.4 17.3 16.0 18.5 4.9 74.1
40 11.1 8.3 13.9 13.9 22.2 13.9 83.3
42 19.6 5.9 3.9 39.2 23.5 2.0 94.1
45 25.9 3.7 13.0 27.8 22.2 1.9 94.4
48 15.4 20.2 14.4 17.6 10.6 ‘ 3.2 81.4
49 35.4 10.6 22.0 9.7 10.3 8.9 96.9
50 10.6 8.8 14.1 12.0 12.7 8.8 66.9
51 12.9 9.4 16.5 16.5 6.5 7.9 69.8
53 7.7 9.6 30.8 19.2 9.6 13.5 90.4
54 11.5 6.6 21.3 16.4 21.3 4.9 82.0
56 12.1 10.3 25.9 8.6 17.2 12.1 86.2
57 8.8 14.0 19.3 24.6 5.3 14.0 86.0
58 7.2 7.2 24.3 15.3 15.3 3.6 73.0
59 19.8 5.7 16.0 11.3 16.0 7.5 76.4
60 9.4 7.7 12.7 15.0 26.9 4.8 76.5
61 6.2 3 0 23.6 17.3 28.9 5.1 84.1

62 6.5 103 21.5 14.0 14.0 4.7 71.0

159

Appendix B.3
Client-Based Market Shares
Two-Digit SIC Level
January, 1990

$9. AA ELL DI E_Y_ M. m M
63 4.5 14.5 12.8 23.1 24.4 6.2 85.5
64 11.4 8.6 11.4 17.1 22.9 0.0 71.4
65 7.0 7.7 20.3 11.9 11.9 5.6 64.3
67 7.4 8.0 12.9 10.3 15.8 7.2 61.6
70 14.3 0.0 16.3 6.1 14.3 2.0 53.1
72 10.0 3.3 20.0 10.0 13.3 3.3 60.0
73 11.9 8.8 14.1 16.1 12.6 7.8 71.3
78 9.4 1.2 12.9 21.2 12.9 8.2 65.9
79 10.0 12.9 11.4 11.4 2.9 7.1 55.7
80 12.4 8.3 14.9 24.0 11.6 5.0 76.0
89 16.0 9.3 16.0 18.7 10.7 9.3 80.0

Avg. 13.7 8.8 15.9 17.2 13.9 7.3 76.9

160

Appendix B.4
Sales-Based Market Shares
Two-Digit SIC Level
January, 1990

E & C_L ll" El P_M. BE M
10 22.2 25.3 8.2 16.8 2.1 25.1 99.7
13 21.0 11.6 0.2 18.7 0.4 48.0 99.9
15 24.2 35.8 3.2 13.4 7.3 8.3 92.3
17 33.7 0.4 24.9 21.3 14.9 0.1 95.2
20 13.2 5.0 15.1 27.0 15.5 23.5 99.3
22 21.4 3.1 15.1 41.6 7.2 3.8 92.1
23 29.8 0.0 13.1 32.5 11.2 2.4 89.0
24 75.0 1.3 0.8 13.8 1.9 4.1 96.8
25 25.0 2.7 5.8 31.1 10.5 9.3 84.3
26 30.1 23.8 16.4 4.9 6.9 16.8 98.9
27 16.1 5.4 11.2 40.3 5.0 18.2 96.1
28 14.6 19.3 27.1 7.1 21.7 9.7 99.6
29 5.6 11.6 3.2 19.1 5.9 54.6 100.0
30 7.4 8.2 5.6 11.0 6.4 58.0 96.7
31 6.4 8.2 5.4 5.5 6.6 36.1 68.2
32 21.2 8.7 21.5 32.3 3.7 11.5 98.9
33 7.5 27.2 10.6 29.8 3.0 21.6 99.8
34 7.1 14.1 17.9 19.2 11.3 29.4 99.1
35 15.4 6.5 4.3 15.1 25.0 32.8 99.1
36 13.7 5.5 14.7 39.2 19.2 6.5 99.0
37 3.2 14.0 30.5 31.4 1.0 5.4 85.5
38 5.9 6.2 8.4 8.3 30.9 37.0 96.8
39 4.9 2.4 18.6 15.5 43.4 13.7 98.5
40 3.0 17.9 13.1 17.5 22.6 25.8 100.0
42 42.2 2.3 2.2 34.7 17.4 0.4 99.2
45 46.6 0.1 1.4 39.5 10.4 1.3 99.3
48 ‘ 17.9 64.8 1.7 10.0 1.7 3.3 99.4
49 37.1 8.8 35.0 4.1 5.0 9.8 99.7
50 12.6 9.4 10.1 34.7 11.7 11.2 89.7
51 9.1 6.5 41.4 26.9 7.0 4.4 95.3
53 8.1 4.7 37.3 17.9 11.0 20.5 99.5
54 6.7 18.2 26.0 18.3 21.5 6.8 97.4
56 21.2 7.5 19.6 10.9 30.9 5.4 95.5
57 2.0 37.2 15.1 9.0 7.8 22.0 93.0
58 6.4 6.0 39.0 33.2 11.8 1.2 97.5
59 44.2 6.2 15.0 3.6 21.4 2.6 93.1
60 11.7 6.0 5.2 19.0 33.4 22.8 98.1
61 7.5 5.9 39.7 13.7 24.5 6.3 97.6

62 9.4 15.9 20.0 43.5 4.1 2.3 95.2

 

161

Appendix B.4
Sales-Based Market Shares
Two-Digit SIC Level
January, 1990

E M 21: m E_Y BM BE IL?“
63 11.7 14.8 14.4 24.6 24.2 8.8 98.4
64 35.1 20.9 26.5 1.1 13.9 0.0 97.5
65 3.0 2.7 64.2 3.1 21.8 1.1 96.0
67 13.8 5.1 21.1 26.3 17.0 11.3 94.6
70 49.2 0.0 45.5 0.3 1.9 0.0 97.0
72 17.9 5.3 26.2 20.3 13.0 2.3 85.0
73 12.1 40.6 13.7 11.6 8.9 11.1 97.9
78 7.4 0.0 12.7 45.7 9.2 19.4 94.4
79 16.1 8.8 2.0 21.0 3.9 43.4 95.1
80 20.0 16.2 6.2 40.6 14.4 1.0 98.5
89 22.9 19.1 16.9 23.5 9.6 5.1 97.0

Avg. 18.4 11.9 17.1 21.0 12.8 14.5 95.7

 

 

 
 
 
 

 

 

1111ij

3129300914

lllllll

ill

Ill

MICHIGAN STQTE UNIV.