Auditing: A Journal of Practice & Theory
Vol. 30, No. 1
February 2011
pp. 49–73
American Accounting Association
DOI: 10.2308/aud.2011.30.1.49
The Impact of the Big 4 Consolidation on
Audit Market Share Equality
Kimberly Dunn, Mark Kohlbeck, and Brian W. Mayhew
SUMMARY: We investigate the Big 5 to Big 4 consolidation and its impact on audit
market share equality. We extend the GAO’s 共2008兲 study on audit firm industry market
concentration to examine whether the remaining Big N firms’ market shares are more
equal after the Big 4 consolidation. We also extend the GAO study to examine audit
market shares at the city and city-industry levels. We find that while overall market
concentration increases, the Big 4 have more equal market shares than the Big 5 had
prior to the consolidation at all levels of analysis. The increase in market share equality
may explain why there has been inconsistent evidence of an association between market concentration and competition after the consolidation 共Feldman 2006; GAO 2008兲.
However, we find that the largest four clients in each market we examine are more likely
to share the same auditor after consolidation, which suggests the largest clients face
constrained choices.
Keywords: auditing; market equality; competition.
Data Availability: The data used in this study are available from public sources indicated in this paper.
INTRODUCTION
e investigate the Big 5 to the Big 4 consolidation 共hereafter, Big 4 consolidation兲 and
its impact on audit market share levels. The Big 4 consolidation resulted from Big 5
firm Arthur Andersen’s demise in 2002. We compare the equality of the Big 5’s market
share prior to consolidation in 2001, to the Big 4’s market shares after the audit market settled in
2007. We specifically focus on the equality of market shares among the Big N at the nationalindustry level, the city level, and the city-industry level.
We study how equally the Big N firms share clients before and after the Big 4 consolidation
for three reasons. First, the United States Congress expressed specific concern about audit firm
W
Kimberly Dunn and Mark Kohlbeck are both Associate Professors at Florida Atlantic University, and Brian
W. Mayhew is an Associate Professor at University of Wisconsin–Madison
We appreciate comments from Chris Hogan and two anonymous reviewers. We appreciate comments received on earlier
versions of this paper from workshop participants at Florida Atlantic University. Kimberly Dunn appreciates financial
support from the School of Accounting Executive Program at Florida Atlantic University. Brian W. Mayhew appreciates
the financial support of the Wisconsin School of Business.
Editor’s note: Accepted by Chris E. Hogan.
Submitted: July 2008
Accepted: July 2010
Published Online: February 2011
49
50
Dunn, Kohlbeck, and Mayhew
concentration. It directed the General Accounting Office 共GAO兲 to study concentration levels and
competition as part of the Sarbanes-Oxley Act of 2002 共SOX; GAO 2003, 2008兲. The congressional mandate to study the audit market speaks to the importance of the topic to policy setters.
Second, the GAO 共2003, 2008兲 studies only consider the impact of consolidation based on
overall market concentration and do not examine whether consolidation impacts how equally
divided market shares are among leading firms pre- and post-consolidation. We examine market
equality based on evidence from industrial organization research that suggests equality of market
shares among the largest suppliers, in addition to market concentration, can impact competition
共e.g., Willis and Rogers 1998; Schmalensee 1989兲. In addition, there is indirect evidence in the
auditing literature that unequal market shares can impact competition. Mayhew and Wilkins
共2003兲 show large market share differences between market leaders and the next largest firm lead
to price premiums. Such differences in market shares are consistent with market inequality. Doogar and Easley 共1998兲 also model that differences in market shares 共i.e., equality兲 can inhibit the
ability of small market share firms to compete for clients due to capacity constraints.
Third, we examine the Big 4 consolidation because it did not result from a merger like the
prior consolidations and, therefore, the impact on market equality is less clear. The demise of
Andersen led to significant competition for its clients 共Kohlbeck et al. 2008兲.1 In contrast, the Big
8 consolidation to the Big 6 共hereafter, Big 6 consolidation兲 and the consolidation from the Big 6
to the Big 5 共hereafter, Big 5 consolidation兲 both resulted from mergers among the largest firms.
Clients tend to stay with merged firms 共Healy and Lys 1986兲. Andersen’s demise also occurred
contemporaneously with the passage of SOX. SOX strengthened auditors’ independence rules and
required internal control reviews that absorbed auditors’ capacity. Both independence requirements
and a decrease in available capacity could affect market equality. The competition for Andersen’s
clients and advent of SOX can lead to either industry leaders becoming more dominant; or it can
lead to gains by industry laggards gaining relative market share. The former 共latter兲 results in less
共more兲 equal market shares.
We investigate how equally the Big N shares clients before and after the consolidation, using
three different measures. Each measure provides its own evidence on the equality of market
shares. First, we use the Gini measure of concentration, which explicitly measures market equality
among suppliers and includes an adjustment for the number of the suppliers 共Quick and Wolz
1999; Abidin et al. 2008兲. Gini provides the purest measure of equality among the Big N firms.
Second, we use the adjusted Herfindahl Index 共ADJ_HI兲 developed by Minyard and Tabor 共1991兲.
It adjusts the Herfindahl Index for the expected market shares, based on equal sharing among the
Big N audit firms. This measure is similar in purpose to the Gini measure, but it is less refined.
Finally, we develop a new measure of auditor diversification among the largest four clients in each
industry. Prior research suggests that the largest clients in many industries do not want to employ
the same auditor due to concerns over client-sensitive data 共Kwon 1996兲. This measure supplies
evidence on how equally the industry’s largest clients are divided among the Big N. We employ
both descriptive statistics and multivariate analysis at each level of analysis.
Descriptive analysis of industry-level data shows that firms with relatively low market shares
共less than 15 percent兲 gain market share relative to firms with high market shares 共greater than 25
percent兲. They appear to gain market share via Andersen’s clients, Big 4 to Big 4 switchers, and
non-Big 4 to Big 4 switchers. They also gain relatively more new clients. We find, using multivariate analysis, that market equality among the Big 4 firms increases at the industry, city, and
1
Kohlbeck et al. 共2008兲 provide evidence that some of Andersen’s clients were purchased through office-level purchases
by other Big 4 firms. However, more than half of Andersen’s clients changed before the office was purchased or were
affiliated with offices that were not purchased.
Auditing: A Journal of Practice & Theory
American Accounting Association
February 2011
The Impact of the Big 4 Consolidation on Audit Market Share Equality
51
city-industry levels following the Big 4 consolidation, based on the Gini and adjusted Herfindahl
Index. However, we find an increase in commonality of auditors among the four largest clients at
the industry and city-industry levels. Our results are consistent with increased sharing of small and
medium-size clients among the Big 4. Our multivariate analysis also suggests that high-litigation
industries have more equality of market shares than other industries. It appears that the Big 4 firms
try to avoid becoming overly concentrated in high-litigation industries.
Our research makes three primary contributions. First, we extend the GAO’s 共2008兲 report by
examining the impact of consolidations using multivariate controls for alternative explanations
and examining changes in the equality of market shares. The GAO report provides univariate
descriptive analysis of industries’ market shares, using the Herfindahl-Hirschman Index for 2002
and 2006 共GAO 2008兲. We control for alternative explanations for observed market share increases, using a multivariate approach. We extend the GAO’s findings to show that industry-level
market equality increases, even after controlling for other factors. This is important because
evidence from the industrial organization literature suggests that equality of market shares among
the leading firms impacts competition 共Willis and Rogers 1998兲.
Second, we examine the impact of consolidation on equality of city-level 共and city-industry
level兲 market shares to provide insight into the local effects of consolidation. Recent research has
examined the city-level audit market competition and audit quality, suggesting that focusing only
on the national-industry level may be incomplete 共Francis et al. 2005; Kallapur et al. 2009兲. We,
again, find more equality in market shares at the city and city-industry levels following the Big 4
consolidation.
Third, we extend the analysis of market share changes to explore whether the increased
concentration levels have constrained the largest clients from selecting different auditors than their
largest competitors. This work expands on Kwon’s 共1996兲 observation that the largest clients in
each industry typically seek different auditors. We provide evidence that the largest clients in each
industry are more likely to have the same auditor after the consolidation. This suggests that the
Big 4 consolidation limited the largest clients’ choices.
The paper proceeds as follows. First, we provide background information and related research
on auditor concentration and the Big 4 consolidation event. We then develop our research design
employed to evaluate the data. In the next two sections, we describe the sample and report our
results. The final section provides our summary and conclusions.
BACKGROUND
In this section, we provide the basis for our research. First, we discuss the concept of market
equality from the industrial organization literature and how it relates to audit markets. Second, we
provide an overview of the audit concentration research. Finally, we discuss the Big 4 consolidation and our specific research question.
Market Equality
The industrial organization literature uses market concentration levels to assess market structure and competition 共Schmalensee 1989兲. Much of this research uses cumulative market shares of
the largest firms in the industry as a measure of concentration. For example, Danos and Eichenseher 共1986兲 and Hogan and Jeter 共1999兲 use market concentration ratios based on the cumulative
market shares of the largest four auditors in the industry to assess audit market concentrations.
Market concentration ratios often mask underlying differences between industries, in terms of
market share equality, across the market share leaders. For example, compare two hypothetical
industries. Industry A’s leaders have market shares of 60, 10, 5, and 5; and industry B’s have
market shares of 20, 20, 20, and 20. Both industries have four-firm concentration ratios of 80, but
we expect more competition in the latter industry.
Auditing: A Journal of Practice & Theory
February 2011
American Accounting Association
52
Dunn, Kohlbeck, and Mayhew
Market share equality provides insight into the level of competition within industry or city
markets, as well as the ability of clients to find unique auditors. Industrial organization research
generally supports the conclusion that more-concentrated industries are less competitive 共Schmalensee 1989兲. However, recent research on the Big 4 consolidation suggests the association between concentration and competition is less predictable 共Feldman 2006; GAO 2008兲. Feldman
共2006兲 finds audit fees increase more in industries that become more concentrated after the Big 4
consolidation. In contrast, the GAO 共2008兲 finds only an economically insignificant increase in
fees associated with concentrated industries. We believe the lack of consistent evidence is due to
differences in the equality of market shares among the Big N. PepsiCo and Coca-Cola provide a
classic example of two firms in a highly concentrated industry, but who have relatively equal
market shares.2 These two companies compete rigorously. Willis and Rogers 共1998兲 provide additional evidence that market equality impacts competition. They find advertising intensity 共advertising expense/sales兲 is higher in food markets where the leading producers have less-dispersed
共i.e., more equal兲 market shares suggesting that equality can lead to more competition.
Auditor Concentration
Danos and Eichenseher produced the first large-scale U.S. studies of auditor industry concentration 共Eichenseher and Danos 1981; Danos and Eichenseher 1982, 1986兲. They find that regulated industries are more concentrated than non-regulated industries 共Eichenseher and Danos
1981兲; that the changes in the auditors’ industry market shares depend on beginning overall market
shares, industry market shares, and nature of the industry 共Danos and Eichenseher 1982兲; and that
growth firms and large clients drive increases in Big 8 market shares, and most of this growth
occurs in regulated industries 共Danos and Eichenseher 1986兲.
Hogan and Jeter 共1999兲 update Danos and Eichenseher 共1982兲 through the Big 6 mergers that
formed Deloitte & Touche and Ernst & Young. They find evidence of significant increases in
industry concentration levels in non-regulated industries, while regulated industries remain highly
concentrated. They also provide evidence that the market leaders continue to increase their market
shares, while Big 6 firms with small industry market shares lose market share over time. Doogar
and Easley’s model appears to explain the findings of Hogan and Jeter 共1999兲. Doogar and Easley
共1998兲 predict firms with smaller market shares will continue to have small market shares because
they do not have the same productive capacity to serve new clients that large market share firms
possess. Minyard and Tabor 共1991兲 and Wootton et al. 共1994兲 find overall concentration increased
after the Big 6 consolidation. Francis et al. 共1999兲 examine the effect of the Big 6 mergers at the
national-industry, city, and city-industry level. They find that Ernst & Young increased market
shares in cities where it had high pre-consolidation market shares, thereby becoming a market
leader. Deloitte & Touche increased market shares, providing them more coverage of cities and
industries, but generally did not become a market leader.
Big 4 Consolidation
The Big 4 consolidation raises concerns about the level of audit market competition 共GAO
2003, 2008兲. The consolidation also corresponded to the creation of SOX and the Public Company
Accounting Oversight Board 共PCAOB兲 and the strengthening of independence rules that limit the
services an audit firm can provide to an audit client. The consolidation and the change in independence rules reduce the number of auditors for clients to choose from—especially for larger
2
Beverage Digest reports Coca-Cola with 42.7 percent and PepsiCo with 30.8 percent of the Carbonated Soft Drink 2008
market. The combined 73.5 percent market share is very high based on typical measures of market concentration.
Auditing: A Journal of Practice & Theory
American Accounting Association
February 2011
The Impact of the Big 4 Consolidation on Audit Market Share Equality
53
companies that demand a large international audit firm. The Big 4 consolidation should increase
concentration in the audit market, but the impact on market share equality is less clear.
Prior consolidations do not provide much insight into the Big 4 consolidation. In contrast to
the Big 4 consolidation, the Big 6 and Big 5 consolidations resulted from mergers. Clients tend to
follow their auditor to the newly merged firm 共Healy and Lys 1986兲. However, Andersen’s failure
resulted in a less orderly client movement 共Kohlbeck et al. 2008兲. KPMG, Ernst & Young, and
Deloitte & Touche each purchased some former Andersen offices; and all four remaining firms
competed on the open market for clients.
The Big 4 consolidation can produce different equality of market share outcomes. A significant increase in industry-leader market shares generates less market equality. Industry leaders can
gain market share because clients demand industry leaders for quality reasons 共Craswell et al.
1995; Dunn and Mayhew 2004兲. Industry leaders can also pursue clients to maintain market
shares. And leaders may be the only firms with the production capacity to expand 共Doogar and
Easley 1998兲. SOX increased the reporting requirements to include internal controls, thereby
increasing the audit effort required for each client. The increased reporting requirements reduced
the ability of auditors to supply audits due to constrained resources. Doogar and Easley 共1998兲
assert that firms with large market shares have more capacity to adjust to increased supply requirements than small market share firms, such that market leaders tend to remain so. This same
idea applies at the industry level, where large industry market share firms are more likely to have
the necessary expertise to expand within the industry than small industry market share firms.
Francis et al. 共1999兲 document similar leader-based gains associated with the merger that formed
Ernst & Young in 1989.
Conversely, the Big 4 consolidation provides the non-industry leaders opportunities to gain
market shares relative to the leader共s兲. For example, Francis et al. 共1999兲 document the merger
creating Deloitte & Touche in 1989, as an effort of non-leaders to increase coverage across the
relevant audit markets. Non-leaders also gain when the industry-leading firms do not pursue
additional clients to avoid potential scrutiny 共e.g., political costs兲 or to diversify industry-specific
risk. The GAO’s concern over industry-level concentration adds credibility to the claim that firms
face political costs in overly concentrated industries. As a result, audit firms with lower preconsolidation market shares gain relatively greater market shares. This result produces more
equally balanced industry market shares among the remaining large audit firms.
SOX also limits the ability of audit clients to purchase nonaudit services from their auditor. If
clients value the nonaudit services greater than the audit services, auditor-concentration ratios can
change as clients obtain new auditors. Conversely, SOX may not affect auditor-concentration
ratios as audit leaders maintain their audit-specialization investment and forgo conflicting nonaudit
fees.3
Most national and international studies document an increase in market concentration from
the Big 4 consolidation 共GAO 2008; Beattie et al. 2003; Hamilton et al. 2008; Abidin et al. 2008兲.
Beattie et al. 共2003兲 and Abidin et al. 共2008兲 consider the impact of the Andersen demise in the
U.K. market, where Deloitte acquired the Andersen practice, and find a general increase in concentration levels. More importantly, Abidin et al. 共2008兲 find that market equality increases across
3
We considered the potential impact of nonaudit services when developing our research design. We observed a significant
decline in post-SOX nonaudit service fees. The cause of this decline appears twofold. First, three of the remaining Big
4 sold their consulting practices at or near the time SOX was implemented. Second, the SOX prohibition on some
services and more public awareness of the high level of nonaudit services sold to audit clients also reduced the level of
nonaudit services. To be consistent with prior research, we do not include nonaudit fees in our design. However, we
revisit the potential role of nonaudit fees in our sensitivity tests.
Auditing: A Journal of Practice & Theory
February 2011
American Accounting Association
54
Dunn, Kohlbeck, and Mayhew
the remaining firms using the same metric we employ. Ballas and Fafaliou 共2008兲 find national
concentration metrics increase in most of the EU as well as in broad industry sectors.
The Big 4 consolidation research to date shows clear evidence that overall concentration
levels have increased, but provides limited insight into whether market shares are more or less
equal after consolidation. In addition, there is little or no research examining the impact of
consolidation at the city or city-industry level.
RESEARCH DESIGN
In this section, we describe three measures of equality and the empirical methods we use to
examine auditor industry market equality. Each metric measures a different aspect of audit market
equality.
First, we consider the Gini measure of the Big N market shares at the industry level. The Gini
measure provides a measure of market share equality based on the Lorenz curve. A Lorenz curve
graphs the cumulative market share from the smallest market share to the largest market share in
an industry. The Gini measure effectively compares the curve to the 45-degree angle representing
equal market shares. We adopt the measure used by Abidin et al. 共2008兲 and Quick and Wolz
共1999兲:
Ginik,t =
冉 冊 兺 冋冉
n,k,t
2
2
nk,t
Xk,t
i−
i=1
冊 册
nt + 1
Xk,t,i ⫻ 100
2
共1兲
where n equals the number of Big N auditors in industry k and time t; Xk,t equals the mean market
share based on the audit fees for the Big N auditors within industry k, at time t; and Xk,t,i equals
the industry market share k, at time t, of Big N audit firm i based on the audit fees. The Gini
measure is calculated for each industry, with the Big N auditors ordered from smallest to largest in
terms of market share rank within each industry.4
Close inspection of the Gini measure shows that as market equality decreases Gini increases.
Market shares are ranked from smallest to largest. The smaller market shares below the median
count as negative numbers; and larger market shares count as positive numbers in the summation.
When market shares are relatively equal, the summation will be close to zero, and when they are
unequal they will approach 100. This increase results from the larger market share firms creating
relatively large positive numbers that exceed the much smaller negative numbers in the summation. It also adjusts for number of supply firms, which enables us to compare pre- and postconsolidation Gini scores.
We employ a second equality measure based on the adjusted Herfindahl Index 共ADJ_HI兲
proposed by Minyard and Tabor 共1991兲. The Herfindahl Index measures the level of competition
within an industry or market. However, it is sensitive to the number of audit firms; in that, it will
almost always increase if the number of firms decreases. The ADJ_HI is the difference between
the Herfindahl Index and the base level 共expected market share based on the number of Big N
audit firms兲. While this measure does not specifically measure equality of market shares, it enables
us to use the more widely accepted Herfindahl Index and compare it for the pre- and post-events.
The ADJ_HI is calculated as follows 共Minyard and Tabor 1991兲:
n,k,t
ADJ_HIk,t =
4
兺
i=1
冉
2
Xk,t,i
冊
−
1
nk,t
共2兲
As discussed further in the Sample section, our interest is competition among the Big N and therefore we base the Gini
and ADJ_HI measures on Big N auditors only.
Auditing: A Journal of Practice & Theory
American Accounting Association
February 2011
The Impact of the Big 4 Consolidation on Audit Market Share Equality
55
where n equals the number of Big N auditors in industry k; and X equals the industry market share
of Big N audit firm i.
Both the Gini and ADJ_HI metrics measure the equality of market shares that are less susceptible to changes in the number of suppliers, but do not assess the optimality of market shares.
Audit firms can potentially provide higher-quality audits by investing heavily in an industry and
auditing a significant portion of an industry, thereby enhancing their industry-specific expertise.
Greater concentration by one auditor leads to more unequal market shares.
Our final measure differs in nature from the first two measures. It enables us to evaluate
auditor concentration among the largest clients in the industry. Kwon 共1996兲 argues large clients
often have concerns about information transfer that could result from hiring an auditor who also
services its main competitors. A notable example resulted when Ernst & Whinney and Arthur
Young merged in 1989 to form Ernst & Young. This merger resulted in Coca-Cola and PepsiCo
having the same auditor. Immediately following the merger, PepsiCo hired a new auditor 共Kwon
1996兲.
We develop a measure based on the commonality of auditors among the largest four clients in
each industry. COMMON4 takes on one of the following four values: 1 共each of the four companies use a different auditor兲, 2 共the four companies use three different auditors兲, 3 共the four
companies use two different auditors兲, and 4 共each of the four companies use the same auditor兲.
Greater values indicate a higher level of commonality among the four largest companies in the
industry. The largest four clients represent 35 percent of 2001 industry assets, on average. These
large clients typically are the most concerned about information transfer to competitors. Based on
the desire for limited information transfer, we would not expect a change in COMMON4 after the
consolidation. However, the largest clients also require the most auditor resources and more
specialized staff that can result in production constraints on auditors with smaller market shares
共Doogar and Easley 1998兲. In such cases, we would expect large clients to share the same auditor.
We are not able to adjust the COMMON4 measure for the change in the number of Big N auditors.
However, we intentionally selected the top four clients so that it was possible for each of the four
clients to have a separate auditor both pre- and post-consolidation.
More 共less兲 equality of market shares among the Big 4 auditors following the Big 4 consolidation is consistent with lower 共greater兲 measures for Gini, ADJ_HI, and COMMON4. We run
simulations to assess how objectively these metrics measure equality based on the consolidation
from the Big 5 to Big 4. We evaluate two basic scenarios. First, we assume Andersen’s clients are
allocated pro rata, based on the remaining firms’ 2001 market shares within each industry. Under
such conditions, the industry leader obtains more clients than the industry laggards. This allocation
should not result in more equality. Our simulations show, essentially, zero change in the Gini and
ADJ_HI metric. Of course, any more extreme allocation favoring the industry leaders, results in
less industry equality. Second, we simulate the case where the remaining firms share Andersen’s
clients equally. In this case, the leading firm and lagging firms all get the same share of clients.
This case leads to more equality, as the small market share firms will gain relatively more clients
than the leading firms. As expected, we see an increase in the Gini and ADJ_HI metrics under such
assumptions. Again, any allocation that results in more than an equal share going to the small
market share firms also results in more equality. Our simulations provide comfort that these
metrics capture the equality of market shares, regardless of the number of suppliers in this setting.
We develop a multivariate model to investigate changes in these three market share equality
measures. We draw on prior research of changes in industry auditor concentration 共Hogan and
Jeter 1999兲 to develop an empirical model. We use the same empirical model for each measure,
because this set of independent variables captures the industry and city characteristics that are
likely to impact each measure. We estimate the following equation using industry-year observations for each of our measures of audit market share equality:
Auditing: A Journal of Practice & Theory
February 2011
American Accounting Association
56
Dunn, Kohlbeck, and Mayhew
Market Share Equalityk,t = ␣1 + ␣2CCR4k,t + ␣3SIZEk,t + ␣4REGk,t + ␣5LITk + ␣6BIG4t + ␧k,t
共3兲
where Market Share Equality takes on the value of Gini, ADJ_HI, or COMMON4, as defined
above.
Hogan and Jeter 共1999兲 model industry auditor concentration as a function of industry characteristics including client concentration in the industry, industry size, regulation, and litigation
risk.5 CCR4 is the four-firm client concentration ratio for each industry, calculated as the ratio of
the sales of the four largest clients in the industry to total industry sales. Greater client concentration results in a lower level of equality. Auditors cannot audit a portion of a client, so when a
few clients dominate an industry, so will a few auditors. SIZE is the natural log of the mean asset
size of the clients included in industry k. We include average client size because the larger the
average client, the more likely the industry is to be equally shared. This view is essentially the
corollary to the client concentration ratio. REG is an indicator variable if the client is in regulated
industries 共utilities, banking, insurance, and trading industries, based on Fama and French’s 关1997兴
49 industry classifications兲 and 0 otherwise. Economies of scale associated with auditing firms in
regulated industries also lead to higher inequality. Finally, LIT is an indicator variable for clients
in litigious industries 共medical equipment, pharmaceutical products, chemicals, machinery, electrical equipment, business services, computer hardware, electronic equipment, and measuring and
control equipment industries, based on Fama and French’s 关1997兴49 industry classifications兲. We
expect more equality for higher litigation risk industries, consistent with auditor risk aversion.
However, auditors may seek greater market share to increase expertise and reduce audit risk as a
way to limit litigation risk. Greater individual firm market share increases inequality across auditors. BIG4 equals 1 after the Big 4 consolidation.6 This variable captures the Big4 consolidation
event from Big 5 to Big 4.
SAMPLE
Our sample consists of observations from 2001 and 2007 to provide data immediately prior to
Andersen’s demise—and passage of SOX and a more recent year sufficiently distant from the
event—that should reflect equilibrium. We include all annual firm observations from the cross
section of Audit Analytics and Compustat from these two years with available financial, auditor,
and industry data. Our final sample consists of 47 industry observations from each year where
industries are based on Fama and French 共1997兲49 industry classifications.7
We limit our main analysis to Big N auditors for two reasons. First, we want to compare
results to the prior research that also focuses on the Big N auditors. Second, the level of competition among the largest auditors is the primary concern of the GAO and Congress 共GAO 2003,
2008兲. We also provide supplementary analysis that incorporates the non-Big N to provide some
insight into the impact on the overall audit market.
In Table 1, we provide an analysis of auditors’ industry market shares changes from 2001 to
2007, organized by the auditor’s 2001 rank within the industry.8 The summary statistics show that
the smaller auditors in the market 共2001 rank ⫽ 3, 4, and 5兲 appear to have gained the most at the
expense of the industry leaders. The third 共fourth兲 ranked auditor reported increases in 80 percent
5
6
7
8
Hogan and Jeter 共1999兲 also include a growth variable, a trend variable as a main effect, and the trend variable interacted
with regulatory industries to investigate changes in industry auditor concentration overall and conditional on the regulatory nature of the industry. We exclude these because of the reduced time frame being investigated.
Each variable is remeasured at the city level and city-industry level, as appropriate, for these additional analyses.
Our analysis is based on Fama and French’s 共1997兲49 industry code classifications. Excluded from our analysis are two
industries. We exclude group 45, as auditor codes are not available on Compustat for banks in 2001, and group 49,
which has minimal observations.
Andersen is excluded from this analysis to better understand the effect on the remaining Big N auditors.
Auditing: A Journal of Practice & Theory
American Accounting Association
February 2011
Fama and French’s
Industry
Classification
2001–07
Change
64%
40%
33%
57%
49%
⫺11%
9%
23%
⫺19%
⫺6%
37%
47%
39%
55%
32%
66%
42%
47%
45%
33%
88%
42%
34%
61%
27%
38%
29%
35%
33%
34%
5%
⫺20%
⫺11%
5%
⫺9%
⫺23%
⫺16%
1%
12%
⫺4%
⫺14%
⫺8%
5%
⫺8%
⫺27%
4%
⫺20%
4%
3%
5%
35%
⫺4%
Market
Share
Rank ⴝ 2
2001–07
Change
28%
36%
21%
17%
41%
31%
9%
⫺4%
4%
15%
9%
0%
43%
31%
33%
24%
16%
19%
⫺32%
10%
⫺27%
10%
21%
11%
33%
5%
25%
28%
⫺1%
5%
0%
⫺3%
26%
8%
28%
32%
0%
14%
Market
Share
Rank ⴝ 3
2001–07
Change
24%
20%
⫺5%
⫺6%
17%
13%
8%
13%
12%
18%
10%
17%
14%
12%
14%
21%
⫺6%
5%
8%
11%
⫺2%
6%
1%
10%
4%
13%
14%
9%
26%
17%
19%
16%
18%
17%
7%
15%
24%
⫺5%
14%
3%
5%
14%
Market
Share
Rank ⴝ 4
2001–07
Change
3%
7%
13%
4%
11%
9%
2%
2%
2%
9%
0%
58%
6%
14%
7%
7%
8%
2%
6%
14%
7%
6%
7%
11%
2%
18%
⫺2%
20%
2%
8%
17%
16%
15%
15%
16%
10%
18%
⫺1%
10%
5%
8%
⫺1%
Market
Share
Rank ⴝ 5
2001–07
Change
10%
⫺5%
10%
8%
8%
6%
8%
5%
11%
3%
6%
5%
8%
1%
4%
11%
8%
5%
⫺1%
11%
4%
9%
7%
10%
4%
6%
11%
14%
10%
9%
7%
1%
8%
7%
(continued on next page)
57
February 2011
American Accounting Association
Agriculture
Aircraft
Apparel
Automobiles and Trucks
Beer and Liquor
Business Services
Business Supplies
Candy and Soda
Chemicals
Coal
Communication
Computer Hardware
Computer Software
Construction
Construction Materials
Consumer Goods
Defense
Electrical Equipment
Electronic Equipment
Entertainment
Fabricated Products
Food Products
Healthcare
Insurance
Machinery
Measuring and Control
Equipment
Medical Equipment
Market
Share
Rank ⴝ 1
The Impact of the Big 4 Consolidation on Audit Market Share Equality
Auditing: A Journal of Practice & Theory
TABLE 1
Analysis of Auditor Industry Market Shares from 2001 to 2007
Fama and French’s
Industry
Classification
February 2011
Number of industries:
Decreasing market share
Increasing market share
Average changes:
Simple
2001–07
Change
Market
Share
Rank ⴝ 2
2001–07
Change
63%
⫺34%
20%
⫺10%
34%
48%
33%
⫺15%
⫺20%
3%
29%
10%
43%
39%
42%
28%
⫺24%
⫺15%
36%
51%
60%
36%
100%
45%
37%
33%
⫺4%
⫺8%
⫺7%
11%
⫺23%
⫺3%
Market
Share
Rank ⴝ 3
2001–07
Change
16%
18%
0%
11%
13%
11%
10%
29%
⫺3%
⫺5%
0%
40%
29%
26%
26%
20%
40%
66%
⫺22%
⫺6%
⫺8%
16%
10%
4%
25%
16%
8%
6%
7%
31%
10%
22%
6%
3%
49%
22%
28%
0%
23%
29%
6%
1%
⫺11%
20%
2%
14%
16%
21%
0%
20%
18%
21%
16%
5%
11%
3%
5%
6%
6%
9%
15%
14%
Market
Share
Rank ⴝ 4
2001–07
Change
3%
7%
12%
6%
15%
0%
3%
24%
⫺2%
4%
10%
13%
15%
3%
29%
6%
0%
14%
15%
⫺11%
13%
6%
11%
4%
6%
3%
Market
Share
Rank ⴝ 5
2001–07
Change
7%
6%
8%
0%
3%
7%
8%
3%
20%
10%
15%
1%
0%
46%
8%
6%
1%
0%
1%
2%
5%
6%
3%
9%
3%
10%
5%
4%
25
16
10
22
7
30
5
27
2
27
⫺5%
5%
8%
8%
7%
(continued on next page)
Dunn, Kohlbeck, and Mayhew
Non-Metallic and Industrial
Metal Mining
Personal Services
Petroleum and Natural Gas
Pharmaceutical Products
Precious Metals
Printing and Publishing
Real Estate
Recreation
Restaurants, Hotels, Motels
Retail
Rubber and Plastic
Products
Shipbuilding, Railroad
Equipment
Shipping Containers
Steel Works
Textiles
Tobacco Products
Trading
Transportation
Utilities
Wholesale
Market
Share
Rank ⴝ 1
58
Auditing: A Journal of Practice & Theory
American Accounting Association
TABLE 1 (continued)
Fama and French’s
Industry
Classification
Weighted by 2001 audit
fees
Weighted by 2001
number of clients
Market
Share
Rank ⴝ 1
2001–07
Change
Market
Share
Rank ⴝ 2
2001–07
Change
Market
Share
Rank ⴝ 3
2001–07
Change
Market
Share
Rank ⴝ 4
2001–07
Change
Market
Share
Rank ⴝ 5
2001–07
Change
⫺5%
7%
13%
10%
11%
⫺18%
4%
8%
10%
6%
Auditor industry market shares and rank are based on 2001 audit fees. The 2001–2007 Change columns show total market share change from 2001 to 2007. Table only provides
information for the Big 4 auditors. The data is missing when Andersen or a non-Big 4 auditor was an industry leader in 2001. Andersen was the only Big 5 auditor with public clients
in the precious metals industry.
The Impact of the Big 4 Consolidation on Audit Market Share Equality
Auditing: A Journal of Practice & Theory
TABLE 1 (continued)
59
February 2011
American Accounting Association
60
Dunn, Kohlbeck, and Mayhew
共82 percent兲 of the industries with a weighted average increase of 13 percent 共10 percent兲. The
fifth ranked auditor had similar increases 共93 percent of the industries increase with a weighted
average of 11 percent兲. In contrast, the industry leader 共rank ⫽ 1兲 increased market share in only
39 percent of the industries and had a weighted average decrease in market share of 5 percent.9
Table 2 classifies each auditor’s industry fees based on their 2001 industry market share into
one of three categories 共greater than 25 percent, 15 percent to 25 percent, and less than 15 percent兲
and then describes the change in audit fees from 2001 to 2007. The 2001 auditors with market
shares greater than 25 percent decline from 59.4 percent of the 2001 market to 46.9 percent in
2007. The auditors with less than 15 percent 共15 percent to 25 percent兲 group increased their
market share by 8 percent 共4 percent兲. The evidence is consistent with auditors in the less than 15
percent group gaining market share from relatively greater fee growth from existing clients,
gaining more Andersen clients, and capturing relatively more Big 4 switchers and new listings
than auditors in the other groups. This descriptive summary suggests a more equal sharing of
clients after the Big 4 consolidation.
ANALYSIS
Our analysis consists of three distinct approaches to investigating changes in market share
equality. First, we consider industry market share equality on a national basis. Second, we investigate market share equality on a city basis. Finally, we investigate city-industry market share
equality. We then discuss sensitivity tests.
Industry Market Share Equality on a National Level
In Table 3, we present the Gini measure by industry for 2001 and 2007. Thirty-eight out of 47
industries’ Gini measures decrease in 2007. We calculate weighted-average Ginis, based on 2001
audit fees and the number of clients. Both show a downward trend in weighted-average Ginis,
consistent with increased equality.
Table 4, Panel A reports 2001 and 2007 descriptive statistics for our market share equality
measures and the independent variables of the multivariate regression. The mean and median Gini
measure, and the mean ADJ_HI measure decrease significantly, consistent with increased equality.
In contrast, the median Herfindahl Index increased slightly, consistent with the GAO’s 共2008兲
assertion that market concentration increased. This univariate evidence is consistent with increased
overall concentration; at the same time, there is an overall increase in market share equality across
Big N audit firms.
The number of industries where the largest four clients all have different auditors decreases
from 20 to 15. All of the decrease flows to industries where two auditors audit the four largest
clients. This change in distribution results in a statistically significant increase in the median
COMMON4—despite the fact it is 2 in both years. The independent variables, with the exception
of SIZE, are not different. SIZE increases in 2007, suggesting that the average 2007 firm is larger.
We calculate but do not present Pearson and Spearman correlations for the dependent and
independent variables in Equation 共3兲. Gini and ADJ_HI are highly correlated 共0.988兲, but are
much less correlated with COMMON4 共0.103兲. In addition, the CCR4 ratio is highly correlated
with Gini and ADJ_HI. We expect this correlation, as the size of the largest clients in an industry
directly affects the level of auditor industry market share equality. LIT is negatively correlated
with the equality measures, suggesting more equality among high-litigation industries.
Table 4, Panel B documents our Equation 共3兲 estimate using ordinary least squares 共ordered
9
Similar results are obtained when we determine market share using the square root of total assets audited.
Auditing: A Journal of Practice & Theory
American Accounting Association
February 2011
($ in 000s)
>25%
2001 clients using 2001 audit fees
No longer in Audit Analytics
Continuing clients using 2001 audit fees
Audit fee growth for continuing clients
Continuing clients using 2007 audit fees
Net changes using 2007 audit fees
Former Andersen clients
Other Big 4 to Big 4
Other Big 4 to non-Big 4
Non-Big 4 to Big 4
New listings using 2007 audit fees
2007 clients using 2007 audit fees
$970,138
⫺2,355
967,783
2,032,379
3,000,162
⫺120,191
⫺83,835
⫺106,667
7,796
491,811
$3,189,076
<15%
15%–25%
58.4%
58.4%
49.4%
52.0%
$397,958
⫺160
397,798
1,130,014
1,527,812
42.7%
28.9%
39.1%
46.9%
15,548
⫺9,156
⫺62,610
4,395
418,944
$1,894,933
24.0%
24.0%
27.5%
26.5%
$292,186
⫺982
291,204
950,294
1,241,498
25.1%
16.3%
33.3%
27.9%
104,643
92,991
⫺80,449
14,740
345,899
$1,719,322
Total
17.6%
17.6%
23.1%
21.5%
$1,660,282
⫺3,497
1,656,785
4,112,687
5,769,472
32.2%
54.7%
27.5%
25.3%
0
0
⫺249,726
26,931
1,256,654
$6,803,331
61
February 2011
American Accounting Association
This table reconciles the audit fees 共per Audit Analytics兲 earned in 2001 to those earned in 2007 by audit firms at the industry level. We group audit firms within each industry based
on their 2001 industry market shares measured by audit fees. The ⬎25% group includes firm industries where the audit firm has 25 percent or more of the 2001 industry fees; the
15–25% group includes firm-industry observations where the audit firm has 15–25 percent of the 2001 industry fees; and the ⬍15% group includes firm-industry market shares less
than 15 percent. We reconcile by first removing client fees for clients who leave Audit Analytics between 2001 and 2007. We add the growth in client fees for continuing clients to
get to continuing client fees in 2007 dollars. We then account for changes among auditors. These changes all are zero sum changes; in that, a loss in fees by one category is offset
by a gain in another category. Finally, we add new listings since 2001 at their 2007 audit fee. The total gives the total audit fees in 2007 based on 2001 audit firm industry market
shares. The table shows that the industry market leaders in 2001 have relatively less market share in 2007 and that the laggards gain relative market share.
The Impact of the Big 4 Consolidation on Audit Market Share Equality
Auditing: A Journal of Practice & Theory
TABLE 2
Analysis of Big N Auditor Changes from 2001 to 2007
2001 Auditor Industry Market Share
62
Dunn, Kohlbeck, and Mayhew
TABLE 3
Industry Gini Measures 2001 versus 2007
Fama and French’s Industry
Classification
Agriculture
Aircraft
Apparel
Automobiles and Trucks
Beer and Liquor
Business Services
Business Supplies
Candy and Soda
Chemicals
Coal
Communication
Computer Hardware
Computer Software
Construction
Construction Materials
Consumer Goods
Defense
Electrical Equipment
Electronic Equipment
Entertainment
Fabricated Products
Food Products
Healthcare
Insurance
Machinery
Measuring and Control Equipment
Medical Equipment
Non-Metallic and Industrial Metal Mining
Personal Services
Petroleum and Natural Gas
Pharmaceutical Products
Precious Metals
Printing and Publishing
Real Estate
Recreation
Restaurants, Hotels, Motels
Retail
Rubber and Plastic Products
Shipbuilding, Railroad Equipment
Shipping Containers
Steel Works
Textiles
Tobacco Products
Trading
Transportation
Utilities
Big 5/2001
Big 4/2007
61.18
45.91
19.89
47.17
54.30
24.71
32.23
54.04
34.28
57.23
20.57
54.86
29.39
40.36
42.20
29.31
70.78
37.73
24.27
50.19
20.27
26.17
19.56
29.34
22.21
23.47
31.29
57.24
28.71
41.81
24.46
80.00
47.21
32.80
34.09
29.61
19.74
52.95
70.68
60.31
55.23
34.11
80.00
38.51
28.65
37.77
45.99
40.12
40.40
20.86
46.72
11.82
30.05
47.66
23.40
42.78
16.60
32.07
7.34
25.65
38.61
14.99
57.10
13.38
17.65
35.51
42.13
24.49
20.74
22.46
14.84
24.79
33.40
35.35
20.43
11.13
18.28
55.34
53.78
35.76
19.08
27.96
14.57
37.37
49.31
51.64
38.30
35.88
61.58
26.14
21.86
36.93
(continued on next page)
Auditing: A Journal of Practice & Theory
American Accounting Association
February 2011
The Impact of the Big 4 Consolidation on Audit Market Share Equality
63
TABLE 3 (continued)
Fama and French’s Industry
Classification
Big 5/2001
Big 4/2007
Wholesale
26.75
28.40
Average Across All Industries
Simple
Weighted by 2001 audit fees
Weighted by 2001 number of clients
40.08
49.55
31.61
31.29
37.64
22.53
Gini measures equality of industry market shares across the Big N auditors based on audit fees for each client.
multinomial Logit兲 with Gini and ADJ_HI 共COMMON4兲 as our market share equality measure.10
The Gini and ADJ_HI models generate adjusted R2s of 61 percent and 53 percent, respectively.
The COMMON4 model has a Pseudo R2 of approximately 11 percent.
The BIG4 coefficient is significantly negative in the Gini and ADJ_HI models. Both imply an
increase in equality following the Big 4 consolidation.11 CCR4 is positive and significant in both
models, consistent with expectations that industries with relatively large clients will tend to have
less equal auditor market shares. REG is marginally significant in the Gini model, implying less
equal market share in regulated industries.
The COMMON4 model shows a positive and significant BIG4 coefficient, suggesting that the
frequency that the largest four companies in the industry employ the same auditor共s兲 increases.
Both SIZE and LIT are negative in the COMMON4 estimation. The LIT variable indicates that
auditors are less concentrated among the four largest clients in high litigation risk industries.
Market Share Equality on a City Level
We identify 88 city markets across the United States where Big N firms have offices. All
measures of market share equality and relevant independent variables are recalculated at the city
level. We delete five city markets that do not have Big 4 auditors in 2007. Our final sample of 83
city markets is similar to the 77 used in Francis et al. 共2005兲. Sixty-three of the 83 cities experience a decrease in Gini. Weighted-average Gini across cities also shows a decline from approximately 40 to 30 from 2001 to 2007.
Table 5, Panel A provides descriptive statistics for the market share equality measures and the
independent variables. The Gini measure decreases between 2001 and 2007, consistent with increased equality among the Big 4 auditors. However, there are no differences in the other market
share equality metrics. The average SIZE also increases.
Table 5, Panel B documents our multivariate regressions at the city level. We delete the two
indicator variables, REG and LIT, from Equation 共3兲, as these two variables are not well defined
10
11
To control for differences among industries, we also estimate our models using weighted least squares, using number of
clients within the industry as a weight with similar results.
The consolidation from Big 5 to Big 4 is unique in that Andersen went out of business leaving its clients to the other
firms to pursue. Prior consolidations all involved mergers where it was more likely clients remained with the merged
firm 共Healy and Lys 1986兲. We investigate whether Andersen’s pre-consolidation market share impacted the 2007
equality of market shares. We include an indicator variable, capturing whether Andersen had the first or second leading
2001 market share, and interact it with BIG4. We find no evidence that Andersen’s pre-consolidation market position
impacts the equality of the subsequent audit market.
Auditing: A Journal of Practice & Theory
February 2011
American Accounting Association
64
Auditing: A Journal of Practice & Theory
American Accounting Association
TABLE 4
Auditor Industry Level Market Share Equality 2001 and 2007
Panel A: Descriptive Statistics—Mean (Median)
Gini
2001
2007
HI
40.07
共34.27兲
31.29 ***
共30.05兲**
ADJ_HI
0.35
共0.28兲
0.35
共0.32兲**
COMMON4
0.15
共0.08兲
0.10 *
共0.07兲
1.74
共2.00兲
2.00
共2.00兲*
CCR4
SIZE
0.35
共0.26兲
0.38
共0.29兲
7.74
共7.69兲
8.26 ***
共8.22兲***
LIT
REG
n
0.19
共0.00兲
0.19
共0.00兲
0.06
共0.00兲
0.06
共0.00兲
47
47
Panel B: Multivariate Industry Market Share Equality 2001 versus 2007
Market Share Equalityk,t ⫽ ␣1 ⫹ ␣2CCR4k,t ⫹ ␣3SIZEk,t ⫹ ␣4REGk,t ⫹ ␣5LITk ⫹ ␣6BIG4t ⫹ ␧k,t
Industry Auditor Market Share Equality Metric
Gini
Variables
(n ⴝ 94)
⫹
⫺
⫹
⫹/⫺
?
20.33**
46.02***
0.40
6.90*
⫺0.89
⫺10.29***
61.9%
COMMON4
t-statistic
Estimated
Coefficient
t-statistic
Estimated
Coefficient
␹2-statistic
2.1
9.9
0.3
1.4
0.3
4.7
⫺0.10
0.40***
0.01
0.02
0.01
⫺0.07***
53.7%
1.1
8.5
1.1
0.5
0.4
3.4
0.77
⫺0.35*
0.61
⫺0.77*
1.16***
0.7
2.0
0.4
1.9
7.1
10.8%
31.2***
22.5***
February 2011
*, **, *** Estimated coefficient is significant at the 0.10, 0.05, and 0.01 level, respectively, based on one-tailed tests for directional prediction, and two-tailed tests otherwise.
Gini and ADJ_HI models are estimated using OLS; and the COMMON4 model is estimated using an ordered multinomial Logit.
(continued on next page)
Dunn, Kohlbeck, and Mayhew
Intercept
CCR4
SIZE
REG
LIT
BIG4
Adj. R2
Pseudo R2
F-statistic
Estimated
Coefficient
ADJ_HI
The Impact of the Big 4 Consolidation on Audit Market Share Equality
Auditing: A Journal of Practice & Theory
Variable Definitions:
Market Share Equality ⫽ the value of Gini, ADJ_HI, or COMMON4;
Gini ⫽ the measure of Big N auditor equality within each industry calculated in Equation 共1兲 using audit fees;
HI ⫽ the Herfindahl Index, calculated as the sum of squares of the auditor market shares based on audit fees;
ADJ_HI ⫽ the adjusted Herfindahl Index and is calculated as the difference between the HI and the expected market shares based on the number of Big N
auditors;
COMMON4 ⫽ an indicator variable that takes one of the following four values: 1 共each of the four companies use a different auditor兲, 2 共the four companies use
three different auditors兲, 3 共the four companies use two different auditors兲, and 4 共each of the four companies use the same auditor兲;
CCR4 ⫽ the four-firm client concentration ratio for each industry calculated as the ratio of the sum of the square root of assets of the four largest clients in the
industry to the total for the industry;
SIZE ⫽ the natural log of the mean asset size of the clients included in industry k;
LIT ⫽ an indicator variable for clients in litigious industries 共medical equipment, pharmaceutical products, chemicals, machinery, electrical equipment,
business services, computer hardware, electronic equipment, and measuring and control equipment industries based on Fama and French’s 关1997兴49
industry classifications兲;
REG ⫽ an indicator variable if the client is in regulated industries 共utilities, banking, insurance, and trading industries based on Fama and French’s 关1997兴49
industry classifications兲, and 0 otherwise; and
BIG4 ⫽ an indicator variable equal to 1 if the year is 2007, and 0 if 2001.
65
February 2011
American Accounting Association
66
Auditing: A Journal of Practice & Theory
American Accounting Association
TABLE 5
Auditor City Market Share Equality 2001 versus 2007
Panel A: Descriptive Statistics—Mean (Median)
Gini
HI
2001
53.76
共51.87兲
47.37**
共46.50兲**
2007
0.50
共0.39兲
0.54
共0.44兲
ADJ_HI
COMMON4
CCR4
SIZE
n
0.30
共0.19兲
0.29
共0.19兲
2.06
共2.00兲
2.19
共2.00兲
0.68
共0.66兲
0.67
共0.67兲
6.98
共6.99兲
7.54***
共7.58兲***
83
83
Panel B: Multivariate City Market Share Equality 2001 versus 2007
Market Share Equalityk,t ⫽ ␣1 ⫹ ␣2CCR4k,t ⫹ ␣3SIZEk,t ⫹ ␣4REGk,t ⫹ ␧k,t
City Auditor Market Share Equality Metric
Gini
Variables
(n ⴝ 166)
⫹
⫺
?
15.42***
58.40***
⫺0.255
⫺5.38***
73.1%
COMMON4
t-statistic
Estimated
Coefficient
t-statistic
Estimated
Coefficient
␹2-statistic
3.1
20.5
0.5
3.4
⫺0.193**
0.72***
⫺0.01
0.01
60.2%
2.4
15.7
0.1
0.1
2.98***
⫺0.09
0.45
26.5
0.7
2.3
19.5%
150.4***
84.3***
February 2011
*, **, *** Estimated coefficient is significant at the 0.10, 0.05, and 0.01 level, respectively, based on one-tailed tests for directional prediction, and two-tailed tests otherwise.
Gini and ADJ_HI models are estimated using OLS; and the COMMON4 model is estimated using an ordered multinomial Logit.
(continued on next page)
Dunn, Kohlbeck, and Mayhew
Intercept
CCR4
SIZE
BIG4
Adj. R2
Pseudo R2
F-statistic
Estimated
Coefficient
ADJ_HI
The Impact of the Big 4 Consolidation on Audit Market Share Equality
Auditing: A Journal of Practice & Theory
Variable Definitions:
Market Share Equality ⫽ the value of Gini, ADJ_HI, or COMMON4;
Gini ⫽ the measure of Big N auditor equality within each city calculated in Equation 共1兲 using audit fees;
HI ⫽ the Herfindahl Index, calculated as the sum of squares of the auditor market shares based on audit fees;
ADJ_HI ⫽ the adjusted Herfindahl Index and is calculated as the difference between the HI and the expected market shares based on the number of Big N
auditors;
COMMON4 ⫽ an indicator variable that takes one of the following four values: 1 共each of the four companies use a different auditor兲, 2 共the four companies use
three different auditors兲, 3 共the four companies use two different auditors兲, and 4 共each of the four companies use the same auditor兲;
CCR4 ⫽ the four-firm client concentration ratio for each city calculated as the ratio of the sum of the square root of assets of the four largest clients in the city
to the total for the city;
SIZE ⫽ the natural log of the mean asset size of the clients included in city k; and
BIG4 ⫽ an indicator variable equal to 1 if the year is 2007, and 0 if 2001.
67
February 2011
American Accounting Association
68
Dunn, Kohlbeck, and Mayhew
at the city level.12 The models are well specified with the explanatory power ranging from 19
percent to 73 percent. CCR4 ratio is positively related to each of the measures, as expected. The
BIG4 coefficient suggests a significantly lower Gini measure post-Big 4 consolidation. However,
we find no significant results regarding changes in ADJ_HI or COMMON4 in 2007.
We expanded our analysis at the city level in untabled results. First, we create a variable
representing the number of Big 4 audit firms in each city 共#BIG4兲. The variable itself is negative
and significant when added to the Gini and ADJ_HI city models. That is, the number of Big 4
auditors in a city increases market share equality. #BIG4 does not affect the Gini estimation, but
it produces a significantly negative BIG4 coefficient in the ADJ_HI estimation that was insignificant otherwise. The BIG4 coefficient remains insignificant in the COMMON4 model with the
addition of #BIG4. This result is essentially mechanical due to the way we construct COMMON4
and its high correlation with #BIG4. We also reestimate our models, excluding 15 city-level
observations where only one Big N firm is present. The results strengthen.
Second, we created a variable measuring the number of clients in each city, and include its log
共LN#CLIENTS兲 in the city-level models. LN#CLIENTS is negative and significant in all three
city-level models. The more clients in the city, the equality of Big 4 market shares is greater. The
significance of BIG4 does not change in the Gini model, but again becomes negative and significant in the ADJ_HI model. Again, LN#CLIENTS has no impact on BIG4 in the COMMON4
model.
Market Share Equality on a City-Industry Level
We combine the industry-level and city-level analyses to create 649 city-industry observations. A full cross of city and industry is not possible. Many industries are not represented in each
Big 4 location in both 2001 and 2007. We recalculate all measures of market share equality and
relevant independent variables based on the city-industry level.
In Table 6, Panel A, we report descriptive statistics. The univariate analysis shows that both
the Gini and ADJ_HI are significantly less in 2007 compared to 2001, consistent with an increase
in equality. However, there is no difference in COMMON4.
Table 6, Panel B, reports our multivariate regression estimations. We again drop REG from
Equation 共3兲.13 We include LIT to capture industries with high litigation. The models’ explanatory
power ranges from 7 percent to 32 percent. As expected, CCR4 共SIZE兲 is positively 共negatively兲
related to each of the market share equality measures. We also find that LIT is negatively related
to each measure, consistent with more equality in high-risk industries and with auditors’ risk
aversion. Our city-industry level results are comparable to our industry-level and city-level results.
We find a significantly lower Gini and ADJ_HI in 2007. We document an increase in COMMON4.
Similar to the city analysis, when we add #BIG4 to the city-industry models, it is negative and
significant, but does not change the significance of the BIG4 coefficient for the Gini, ADJ_HI, or
COMMON4 models. When we add LN#CLIENTS to the city-industry models, it is negative and
significant in all three models and has no impact on the significance or sign of the BIG4 coefficient
for the Gini and ADJ_HI models, but BIG4 becomes insignificant in the COMMON4 model.
Comparison to Big 5 Consolidation
Our literature review reveals no studies on the Big 5 consolidation resulting from the merger
of Price Waterhouse and Coopers and Lybrand 共hereafter, PwC兲. This event provides a contrast to
12
13
As an alternative, we include the percentage of clients at the city level that are in regulated industries or litigious
industries as explanatory variables. Our inferences are not affected.
We find only four city industries with more than one observation coded 1 for REG. This does not generate enough
variation to include it in the model. We note that at the city level, regulated industries typically have very few firms. For
example, few cities have more than one utility company. Of course, this is part of the reason the industry is regulated.
Auditing: A Journal of Practice & Theory
American Accounting Association
February 2011
Panel A: Descriptive Statistics—Mean (Median)
Gini
HI
2001
2007
73.98
共80.00兲
67.88 ***
共75.00兲***
0.84
共1.00兲
0.84
共1.00兲
ADJ_HI
0.64
共0.80兲
0.59 ***
共0.75兲***
COMMON4
CCR4
SIZE
2.49
共3.00兲
2.53
共3.00兲
0.98
共1.00兲
0.97
共1.00兲
6.39
共6.36兲
6.97 ***
共6.93兲***
LIT
n
0.35
共0.00兲
0.35
共0.00兲
649
649
Panel B: Multivariate City-Industry Market Share Equality 2001 versus 2007
Market Share Equalityk,t ⫽ ␣1 ⫹ ␣2CCR4k,t ⫹ ␣3SIZEk,t ⫹ ⫹ ␣4LITk,t ⫹ ␣5BIG4t ⫹ ␧k,t
City-Industry Auditor Market Share Equality Metric
Gini
Variables
(n ⴝ 1,298)
⫹
⫺
⫹/⫺
?
10.86***
67.61***
⫺0.36**
⫺2.84***
⫺5.57***
31.8%
COMMON4
t-statistic
Estimated
Coefficient
t-statistic
Estimated
Coefficient
␹2-statistic
3.0
19.9
2.2
5.0
10.4
⫺0.38***
1.13***
⫺0.10***
⫺0.06***
⫺0.03***
19.7%
4.9
15.4
2.8
5.3
2.8
3.69***
⫺0.18***
⫺0.56***
0.22**
28.6
27.4
22.3
4.0
6.7%
152.2***
80.9***
*, **, *** Estimated coefficient is significant at the 0.10, 0.05, and 0.01 level, respectively, based on one-tailed tests for directional prediction, and two-tailed tests otherwise.
Gini and ADJ_HI models are estimated using OLS; and the COMMON4 model is estimated using an ordered multinomial Logit.
(continued on next page)
69
February 2011
American Accounting Association
Intercept
CCR4
SIZE
LIT
BIG4
Adj. R2
Pseudo R2
F-statistic
Estimated
Coefficient
ADJ_HI
The Impact of the Big 4 Consolidation on Audit Market Share Equality
Auditing: A Journal of Practice & Theory
TABLE 6
Auditor City-Industry Market Share Equality 2001 versus 2007
70
Auditing: A Journal of Practice & Theory
American Accounting Association
Variable Definitions:
Market Share Equality ⫽ the value of Gini, ADJ_HI, or COMMON4;
Gini ⫽ the measure of Big N auditor equality within each city-industry calculated in Equation 共1兲 using audit fees;
HI ⫽ the Herfindahl Index, calculated as the sum of squares of the auditor market shares based on audit fees;
ADJ_HI ⫽ the adjusted Herfindahl Index and is calculated as the difference between the HI and the expected market shares based on the number of Big N
auditors;
COMMON4 ⫽ an indicator variable that takes one of the following four values: 1 共each of the four companies use a different auditor兲, 2 共the four companies use
three different auditors兲, 3 共the four companies use two different auditors兲, and 4 共each of the four companies use the same auditor兲;
CCR4 ⫽ the four-firm client concentration ratio for each city-industry calculated as the ratio of the sum of the square root of assets of the four largest clients in
the city-industry to the total for the city-industry;
SIZE ⫽ the natural log of the mean asset size of the clients included in the city-industry;
LIT ⫽ an indicator variable equal to 1 if the city-industry level is in litigious industries 共medical equipment, pharmaceutical products, chemicals, machinery,
electrical equipment, business services, computer hardware, electronic equipment, and measuring and control equipment industries, based on Fama
and French’s 关1997兴49 industry classifications兲, and 0 otherwise; and
BIG4 ⫽ an indicator variable equal to 1 if the year is 2007 and 0 if 2001.
Dunn, Kohlbeck, and Mayhew
February 2011
The Impact of the Big 4 Consolidation on Audit Market Share Equality
71
Andersen’s failure to assess how a merger impacts market equality. We examine the Big 5 consolidation using the same models we use in Table 4; except we use the square root of client assets
to calculate Gini and ADJ_HI for the Big 5 consolidation because audit fee data was not publically
disclosed at that time.14 We compare 1997, the last year prior to PwC’s merger, and 2001, the last
year prior to the Andersen’s collapse. This window is the largest space between the pre- and
post-merger allocation and maximizes the chance of market equilibrium after the Big 5 consolidation.
In untabled results, we find the Big 5 consolidation does not result in a significant change in
equality of market shares under any of the measures we employ at the industry level of analysis.
In contrast, when we look at industry concentration measured by the Herfindahl Index, we find a
significant increase in concentration levels during the Big 5 共p ⬍ .01兲 and Big 4 consolidations 共p
⫽ .08兲.15 This analysis suggests that the Big 5 consolidation resulting from the merger of two
national firms did not increase equality among the leading firms and is fundamentally different
from the Big 4 consolidation.
Sensitivity Analysis
We report the results of a number of sensitivity tests in the following paragraphs. Prior
research uses the square root of clients’ assets and the square root of revenues, rather than audit
fees, in the determination of auditor concentration 共Danos and Eichenseher 1982; Hogan and Jeter
1999兲. Our results are generally consistent with those reported when we consider these alternative
measures for market share equality.
SOX became law at the same time that the Big 4 consolidation occurred. SOX specifically
banned a number of nonaudit services. Audit firms providing these services had to choose whether
to continue providing nonaudit services and resign the audit, or whether to retain the audit and
discontinue providing services. We investigate whether there is a systematic association between
the level of nonaudit fees provided to an industry in 2001 and the equality of industry market
shares in 2007. We first calculate the 2001 ratio of nonaudit fees to audit fees for each industry. We
then use cutoffs of that ratio to form indicator variables based on the median ratio and separately
for the top quarter of nonaudit fee ratio industries. The nonaudit fee variables are not significant,
nor do their inclusions change the significance of the BIG4 coefficient in any of our industry-level
models. We also interact the nonaudit fee ratio variables with BIG4 to assess the differential
impact of observations with high 2001 nonaudit fee ratios on 2007 market shares. Again, the
interactions are not significant; and the basic results do not change. We do not observe any impact
on our previously reported results at the city and city-industry level when we include similar
nonaudit fee ratio variables.
We explore the impact of the non-Big N as well. First, the Big N’s share of audit fees
decreases from 96 percent to 91 percent from 2001 to 2007. The Big N’s share of clients changes
from 86 percent to 68 percent for clients in our initial data set in both 2001 and 2007. Combined,
these changes suggest that the non-Big 4 gained a significant number of smaller clients. To
examine the impact on market equality, we remeasure our dependent variables treating all the
non-Big N as a single firm, and then reestimate the model in Table 4. We take this approach rather
than estimate market shares for literally hundreds of small audit firms, as it captures the essence of
whether these firms, as a group, capture more market share and, thus, a more equal portion of the
market. We find even stronger results than those presented in Table 5, in terms of coefficient size
14
15
As discussed in our sensitivity analyses, we still find increasing equality after the Big 4 consolidation, using measures
based on the square root of client assets.
We do not attempt to examine the Big 5 consolidation at the city level as we do not have adequate information to group
cities in 1997.
Auditing: A Journal of Practice & Theory
February 2011
American Accounting Association
72
Dunn, Kohlbeck, and Mayhew
and lower p-values. Not surprisingly, we do not see a similar impact on COMMON4, which
continues to be positive, suggesting large clients have less choice.
Hogan and Jeter 共1999兲 include industry growth in their model, analyzing changes in auditorconcentration ratios. When we include their measure of growth, it is not significant and our
inferences are not affected. We also consider interactions between the BIG4 variable and REG or
LIT, and find that neither are significant.
Finally, we adjust the composition of the various markets. For our industry analyses, we
exclude industries with fewer than ten firms, Big N market shares less than 90 percent 共greater
influence of non-Big N auditors兲, and client concentration ratios in the first and tenth deciles. We
perform the same analyses for city markets. Overall, our results weaken as our sample sizes and
power of our tests are reduced, but our results remain consistent with those reported.
CONCLUSION
We investigate the Big 4 consolidation’s impact on auditor market share equality. Increased
auditor concentration is almost a given; how any increase is shared among the Big N is unclear.
We find that changes in both the Gini measure of market equality and the adjusted Herfindahl
Index from 2001 to 2007 indicate increased equality among the remaining Big 4 auditors at both
the national-industry level and city-industry level. The Gini measure suggests more equality at the
city level. We also investigate the commonality of the auditors of the largest four companies in
each market. We document an increase in the commonality of auditors of the largest four companies over this time period at both the industry and city-industry level. Combined, we show that the
increased market equality is being driven by market dynamics outside of the largest four clients in
each market.
The increase in market share equality potentially can explain why other studies have found
little impact of the Big 4 consolidation on competition, despite an overall increase in auditor
concentration 共Feldman 2006; GAO 2008兲. Future research can evaluate whether equality of
market shares, as measured in this paper, is associated with differences in competition, audit
quality, and audit fees at the industry, city, and city-industry levels.
REFERENCES
Abidin, S., V. Beattie, and A. Goodacre. 2008. Audit market structure, fees and choice following the Andersen break-up: Evidence from the U.K. British Accounting Review 42 共3兲: 187–206. Available at:
http://papers.ssrn.com/sol3/papers.cfm?abstract_id⫽1096464.
Ballas, A., and I. Fafaliou. 2008. Market shares and concentration in the EU auditing industry: The effects of
Andersen’s demise. International Advances in Economic Research 14: 485–497.
Beattie, V., A. Goodacre, and S. Fearnley. 2003. And then there were four: A study of U.K. audit market
concentration—Causes, consequences and the scope for market adjustment. Journal of Financial
Regulation and Compliance 11 共August兲: 250–265.
Craswell, A., J. Francis, and S. Taylor. 1995. Auditor brand name reputations and industry specializations.
Journal of Accounting and Economics 20 共December兲: 297–322.
Danos, P., and J. Eichenseher. 1982. Audit industry dynamics: Factors affecting changes in client-industry
market shares. Journal of Accounting Research 20 共Autumn兲: 604–616.
——–, and ——–. 1986. Long-term trends toward seller concentration in the U.S. audit market. The Accounting Review 61 共4兲: 633–650.
Doogar, R., and R. Easley. 1998. Concentration without differentiation: A new look at the determinants of
audit market concentration. Journal of Accounting and Economics 25: 235–253.
Dunn, K., and B. Mayhew. 2004. Audit firm industry specialization and client disclosure quality. Review of
Accounting Studies 9 共March兲: 35–58.
Auditing: A Journal of Practice & Theory
American Accounting Association
February 2011
The Impact of the Big 4 Consolidation on Audit Market Share Equality
73
Eichenseher, J., and P. Danos. 1981. The analysis of industry-specific auditor concentration: Towards an
explanatory model. The Accounting Review 56 共July兲: 479–492.
Fama, E., and K. French. 1997. Industry costs of equity. Journal of Financial Economics 43: 153–194.
Feldman, E. 2006. A basic quantification of completive implication of the demise of Arthur Andersen. Review
of Industrial Organization 29: 193–212.
Francis, J., D. Stokes, and D. Anderson. 1999. City markets as a unit of analysis in audit research and the
re-examination of Big 5 market shares. Abacus 35: 185–206.
——–, K. Reichelt, and D. Wang. 2005. The pricing of national a city-specific reputations for industry
expertise in the U.S. audit market. The Accounting Review 80 共1兲: 113–136.
General Accounting Office 共GAO兲. 2003. Public Accounting Firms: Mandated Study on Consolidation and
Competition. GAO-03-864. Washington, D.C.: GAO.
——–. 2008. Audits of Public Companies: Continued Concentration in Audit Market for Large Public
Companies Does Not Call for Immediate Action. GAO-08-163. Washington, D.C.: GAO.
Hamilton, J., Y. Li, and D. Stokes. 2008. Is the audit services market competitive following Arthur Andersen’s collapse? Accounting and Finance 48: 233–258.
Healy, P., and T. Lys. 1986. Auditor changes following Big Eight mergers with non-Big Eight audit firms.
Journal of Accounting and Public Policy 5: 251–265.
Hogan, C., and D. Jeter. 1999. Industry specialization by auditors. Auditing: A Journal of Practice & Theory
18 共Spring兲: 1–17.
Kallapur, S., S. Sankaraguruswamy, and Y. Zang. 2009. Audit market competition and audit quality. Working
paper, Indian School of Business, National University of Singapore, and Singapore Management
University.
Kohlbeck, M., B. Mayhew, P. Murphy, and M. Wilkins. 2008. Competition for Andersen’s clients. Contemporary Accounting Research 25 共Winter兲: 1099–1136.
Kwon, S. 1996. The impact of competition within the client’s industry on the auditor selection decision.
Auditing: A Journal of Practice & Theory 15 共Spring兲: 53–69.
Mayhew, B., and M. Wilkins. 2003. The impact of audit firm industry specialization on fees charged to firms
going public. Auditing: A Journal of Practice & Theory 22 共September兲: 33–52.
Minyard, D., and R. Tabor. 1991. The effect of Big Eight mergers on auditor concentration. Accounting
Horizons 5 共December兲: 79–90.
Quick, R., and M. Wolz. 1999. Concentration on the German audit market—An empirical analysis of the
concentration on the German market for stock corporation audits. International Journal of Auditing 3:
175–189.
Schmalensee, R. 1989. Inter-industry studies of structure and performance. In Handbook of Industrial Organization, Vol. II, edited by Schmalensee, R., and R. D. Willig, 951–1009. Amsterdam, The Netherlands: Elsevier Science Publisher B. V.
Willis, M., and R. Rogers. 1998. Market share dispersion among leading firms as a determinant of advertising
intensity. Review of Industrial Organization 13: 495–508.
Wootton, W., S. Tonge, and S. Wolk. 1994. Pre and post Big 8 mergers: Comparison of auditor concentration.
Accounting Horizons 8 共3兲: 58–74.
Auditing: A Journal of Practice & Theory
February 2011
American Accounting Association
Copyright of Auditing is the property of American Accounting Association and its content may not be copied or
emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.
However, users may print, download, or email articles for individual use.