Singapore Management University

Institutional Knowledge at Singapore Management University

Research Collection School Of Accountancy School of Accountancy

9-2005

Client, Industry and Country Factors Affecting Choice of Big N

Industry Expert Auditors
Michael Ettredge

Soo Young KWON

Korea University

Chee Yeow LIM

Singapore Management University, cheeyeowlim@smu.edu.sg

Follow this and additional works at: https://ink.library.smu.edu.sg/soa_research

Part of the Accounting Commons, Business Law, Public Responsibility, and Ethics Commons, and the
Corporate Finance Commons

Citation
Ettredge, Michael; KWON, Soo Young; and LIM, Chee Yeow. Client, Industry and Country Factors Affecting
Choice of Big N Industry Expert Auditors. (2005). American Accounting Association Annual Meeting.
Research Collection School Of Accountancy.
Available at: https://ink.library.smu.edu.sg/soa_research/639

This Conference Paper is brought to you for free and open access by the School of Accountancy at Institutional
Knowledge at Singapore Management University. It has been accepted for inclusion in Research Collection School
Of Accountancy by an authorized administrator of Institutional Knowledge at Singapore Management University.
For more information, please email library@smu.edu.sg.
 Client, Industry and Country Factors Affecting Choice of

Big N Industry Expert Auditors

Michael Ettredge*
University of Kansas

Soo Young Kwon

Korea University

Chee Yeow Lim

Nanyang Technological University

* Corresponding author
mettredge@ku.edu

October 2008

We wish to thank workshop participants at the 4th Asian Corporate Governance

conference at Korea University, Nanyang Technological University workshop, the 2005
AAA annual conference in San Francisco, and especially Kevin C. W. Chen and Jeff
Payne.

Electronic copy available at: http://ssrn.com/abstract=820404

 Client, Industry and Country Factors Affecting Choice of

Big N Industry Expert Auditors

This study investigates client choice of industry specialist auditors from among

the Big N (Big 4 or 5) in an international (non-U.S.) setting. We investigate

client-specific, industry-level and country-level factors hypothesized to enhance

or decrease Big N clients’ demand for industry expertise. Using data for 29

countries and 14 broad industries from 1993-2005, we find that international

client choice of industry specialist Big N auditors is positively associated with

client size, client growth opportunities, and client capital intensity. The choice of

industry specialists from among the Big N is more prevalent in countries where

levels of investor protection, quality of financial reporting environment, and

national economic development are higher. Clients belonging to regulated

industries tend to select industry specialists.

Keywords: Global audit market, industry specialization, audit market shares,

audit quality, international auditing, Big N auditors

JEL Classifications: G15; L11; M41; M49

Data Availability: The data are available from the Global Vantage database.

Electronic copy available at: http://ssrn.com/abstract=820404

 Client, Industry and Country Factors Affecting Choice of

Big N Industry Expert Auditors

1. Introduction

This paper examines determinants of clients’ choices of industry expert Big N

(Big 5 or Big 4) auditors in an international setting. Prior evidence indicates

that, similar to U. S. clients, international (non-U.S.) clients desiring higher

quality audits select Big N auditors (Fan & Wong [2005]).1 We employ a

sample of international clients who have chosen Big N auditors, to investigate

characteristics of clients that take the additional quality-seeking decision of

selecting industry specialists from among the Big N. 2 Empirical evidence

suggests that industry specialist auditors have both the incentives and the

ability to provide higher quality audit services. Simunic and Stein [1987]

argue that industry-focused audit firms are more likely to invest in

technologies, physical facilities, personnel, and organization control systems

that improve the quality of audits in the firms’ focal industries. Recent

structural shifts by audit firms in the direction of greater industry focus

suggest that industry specialization plays an increasingly important role in

audit quality (Hogan & Jeter [1999]; Solomon et al. [1999]).

Industry-experienced auditors are better able to detect errors among

clients within their industry specialization than outside their specialization

(Bedard & Biggs [1991]; Wright & Wright [1997]; Owhoso et al. [2002]).

1
For simplicity, we will refer to audit clients not headquartered in the U.S. as ‘international’
clients.
2
By including in our sample only clients that purchase Big N audits, we avoid confounding
the choice of a Big N auditor with the choice of an industry specialist auditor.
Specialist auditors are more likely to comply with auditing standards than

non-specialists (O’Keefe et al. [1994]) and are less likely to be associated

with SEC enforcement actions (Carcello & Nagy [2004]). Research also

indicates that earnings quality, as measured by earnings response coefficients

and discretionary accruals, is higher for client firms audited by industry

specialists than non-specialists (Balsam et al. [2003]; Krishnan [2003]).

Moreover, financial analysts rank clients of industry-specialist audit firms

higher, in terms of disclosure quality, than clients of non-specialists (Dunn &

Mayhew [2004]). Collectively, these findings suggest that an auditor’s

industry specialization has value to clients, and that capital markets view

audits provided by industry specialists as having higher quality. Hence, the

largest audit firms use industry specialization as a differentiation strategy

(Mayhew & Wilkins [2003]).

Almost all of the evidence mentioned above is derived from U.S.

audit engagements. However, the largest audit firms attempt to market their

audit services globally on the basis of industry specialization. For example,

PwC’s Global web site provides the following statement:

The depth and breadth of our industry experience, and our

international perspective, are attributes that our clients value

highly. We invest significant resources in acquiring, refining and

sharing these capabilities to further benefit our clients … With

offices in 769 cities in 144 countries, the member firms of

PricewaterhouseCoopers offer a complete range of audit and

2
 [other] services, tailored to your specific industry, wherever you

may need them. (http://www.pwc.com/ as of July, 2008)

Existing international evidence on whether industry specialization is a

dimension of audit quality is mixed, and primarily based on Australian data

(Ferguson & Stokes [2002]; Ferguson et al. [2003]). 3 Questions exist

concerning the extent to which the Big N firms are able to provide consistent

quality of audit services around the world, including industry expertise. The

SEC [2000, 6] states, “We are concerned that audit firms may not have

developed and maintained adequate internal quality control systems at a

global level.”4 Hence, whether clients in countries other than the U.S. (and,

perhaps, Australia) recognize and seek industry expertise from Big N auditors

remains an empirical issue.

We measure an auditor’s industry expertise based on the auditor’s

industry market share in the client’s (non-U.S.) home country. Recent studies

by Ferguson et al. [2003], Francis et al. [2005], and Francis et al. [2006]

investigate the role of audit firms’ individual office (city-level) industry

expertise. Ferguson et al. [2003] and Francis et al. [2005] document that

higher audit fees of industry leaders in the Australia and the U.S. audit market

3
De Beelde [1997] uses 1994 data to investigate whether Big N auditors exhibit industry
specialization. His international (non-U.S.) coverage consists of seven European countries,
plus Japan. He finds little evidence of industry specialization as he defines it, and does not
examine determinants of client auditor choice. Fan and Wong [2005] find that firms in eight
East Asian countries employ Big N auditors when the firms are subject to high agency costs.
That study does not investigate audit firm industry specialization.
4
In [2000: fn. 8] the SEC expands on this theme: “See, for example, 34-40945, AAER-1098
(PricewaterhouseCoopers) and letters from the SEC Chief Accountant to the AICPA SEC
Practice Section dated November 30 ,1998, and December 9, 1999 regarding the need for
global quality internal controls over independence matters, available on the SEC website at
<www.sec.gov>.” Radebaugh and Gray [1997, 651] state: “Because most international public
accounting firms are mixtures of different national public accounting firms, the quality of
work performed is bound to vary.”

3
are driven by auditors that are city-specific industry leaders. Francis et al.

[2006] show that earnings quality is higher when the auditor is a city-specific

industry leader, but not when an auditor is a national leader without also

being a city-specific industry leader. We employ national-level measures of

industry expertise, rather than city-level measures, because the database on

which we primarily rely, for information about who audits whom, does not

provide city-level information.5

This study employs 1993-2005 data from 29 countries, other than the

U.S., to investigate factors affecting clients’ choice of audits provided by

industry specialized Big N auditors. 6 We employ two sets of explanatory

variables proxying for international clients’ demand for industry specialist

audits. The first set consists of variables capturing characteristics of clients

and their industries that are associated with demand for high quality and/or

industry specialist audits (Francis & Wilson [1988]; DeFond [1992]; Craswell

et al. [1995]). These include proxies for client size, financial leverage, growth

opportunities, capital intensity, reliance on external capital, profitability,

industry regulation, and industry concentration ratio. Our second set of

explanatory variables consists of country-specific institutional factors

representing legal protection of outside investors, quality of financial

reporting, and national economic development.

5
Most countries in our sample are small enough for audit personnel to travel anywhere
within those countries, meet with client managers, and return to their offices, in a single day.
Smaller countries frequently contain only one or two cities in which most publicly-traded
clients are located. Thus, the distinction between city-level and national-level industry
expertise is not as important in our study as for studies of larger nations such as the U.S.
6
Use of dependent variables that capture whether clients of Big N auditors choose industry
leaders avoids confounding the demand for Big N audits with demand for audits provided by
industry specialists. It is worthwhile to note that all of our sample companies are audited by
the Big N. Thus, we investigate whether clients having certain characteristics (e.g., large size)
choose industry specialists from among the Big N.

4
 The results provide substantial support for our hypotheses.7 Of the six

client-specific explanatory variables, three have significant coefficients, with

expected signs, across all estimated models in which they appear.

Specifically, choice of industry specialist auditors by international Big N

clients is positively associated with client size (LSALE), client growth

opportunities (market-to-book equity ratio, MB), and client capital intensity

(CAPINT). One of the two industry-level variables, membership in a

regulated industry (REGIND), is positively associated with choice of industry

specialist Big N auditors in nine out of ten models in which it appears. At the

country level, the choice of industry specialists from among the Big N is

more prevalent in countries that offer greater legal protection to investors

(LAW_ENF, VOTING and LEGAL), in countries having better financial

accounting reporting quality (DISC, FIN_TAX and ACCTG), and in wealthier

countries with more developed stock markets (LGDP, SMDEV and ECON).

These results hold even among clients located in the less developed

economies.8 Weaker evidence suggests that clients having greater financial

leverage (LEV) tend to purchase industry-specialist Big N audits (significant

in five out of ten models). The remaining three explanatory variables:

issuance of equity (ISSUE), client losses (LOSS), and industry concentration

(HINDEX) either have coefficients that are almost all insignificant, or some

significant coefficients with signs opposite to expectations.

Our study contributes to the literature in several ways. This study is

one of the first to investigate whether industry- and country-level factors

7
The results summarized are those in Table 4.
8
That is, clients in relatively wealthier countries tend to purchase industry specialist audits, even
among the set of less-developed economies.

5
systematically affect choice of industry specialist auditors around the world.9

We find that client-specific factors and country-level factors are more

important than industry level factors. Second, our study builds on recent

advances in the finance literature on the role of legal protection for financial

market development, ownership structure, and private control benefits. We

document that the level of investor protection appears to affect the demand

for industry specialist Big N auditors. Third, our study also contributes to a

growing body of accounting research on the economic effects of differences

in financial reporting across countries. We provide empirical evidence that

demand for Big N industry specialist audits is positively related to the quality

of financial reporting environments world-wide. Fourth, we show that

national wealth and, to a lesser extent national stock market development,

have an important effect on auditor choice. Finally, this study investigates

choice among Big N audit firms in international audit markets. Numerous

studies have investigated the choice between Big N and non-Big N audit

firms, mostly using U. S. data, but few studies have investigated client choice

among the Big N.

The remainder of the paper is organized as follows. A second section

develops hypotheses and relates them to explanatory variables. Section three

presents our models and dependent variables. Section four describes the

sample and reports our primary test results. A fifth section provides

additional analyses, and a final section presents a summary and conclusions.

9
We note that our use of client-level factors to explain demand for industry expert auditors is
not unprecedented in the literature (see Godfrey & Hamilton [2005]). Our use of client-level
factors in an international setting, however, is new. We control for country-level factors
because of the international nature of our sample clients.

6
 2. Hypotheses and Explanatory Variables

In this study, we consider client-specific, industry-level and country-

level factors that enhance or hinder the demand for audits provided by Big N

accounting firms having expertise in various industries. We assume that an

audit firm’s industry expertise is costly to develop, but results in higher

quality audits.10 High quality audits arguably increase the economic value of

financial accounting information (Bushman & Smith [2001]). The audit

market is not fully integrated across countries due to the fact that some

knowledge is not easily transferable, audit techniques require modifications,

and differences in regulation exist around the globe. However, in spite of

these barriers, we expect to observe employment of industry specialists in

settings in which demand for high quality audits is derived from the potential

benefit of more reliable financial information. We turn now to a detailed

discussion of factors that should increase or decrease international demand for

audits by industry specialists.

2.1 Firm-Specific Factors

Prior international evidence indicates that clients seeking higher

quality audits choose Big N auditors (Fan & Wong [2005]). Industry

specialization also can be viewed as a dimension of international audit quality

(Craswell et al. [1995]). Therefore we expect that international client choice

of Big N auditors who also are industry specialist auditors, represents demand

10
Industry expertise is costly to develop not only because it requires gathering and analyzing
industry data, and modifying audit procedures for different industries, but also because a
‘critical mass’ of clients must be accumulated in a target industry, in the face of competition
from other audit firms. This could lead to heavy discounting of initial audit fees.

7
for audit quality higher than that provided by Big N auditors who are not

industry specialists. We expect the same variables that explain choice of a Big

N versus non-Big N auditor also will explain choice of an industry specialist

auditor from among the Big N. DeFond [1992] and Francis and Wilson [1988]

demonstrate that the demand for quality-differentiated (Big Six) audits is an

increasing function of proxies for firms’ agency costs. Agency costs are

problematic when information asymmetry exists between principals and

agents, and when opportunities exist for agents to transfer corporate wealth to

themselves. Client-specific demand for high quality audits should be

positively associated with proxies for those conditions: opportunity to transfer

wealth via accruals (capital intensity), proxies for information asymmetry

(growth opportunities), and opportunities for managers to expropriate capital

provided by owners and lenders (client size, financial leverage, need for

external capital). We employ client profitability (loss or no loss) as a control

variable, without specifying an expected sign.11 See Godfrey and Hamilton

[2005] and Francis et al. [1999] for additional discussion of these explanatory

variables.12 These two studies examine the demand for auditor specialization

in the U.S. and Australia. We complement these studies by assessing whether

the demand for auditor industry specialization applies around the world. Our

first hypothesis, stated in alternate form, is:

11
The expected coefficient sign for LOSS is unclear. Clients having poor financial
performance arguably are less likely to seek (or be accepted by) large, high quality auditors.
However, Choi and Wong [2007] find a positive association between LOSS and choice of
high quality auditor.
12
We do not include a variable that captures the operating cycle as the inclusion of that
variable reduces our sample by 50%.

8
 H1: Choice of an industry specialist among Big N auditors is positively

associated with client size, financial leverage, growth, capital intensity,

and the need for external financing.

Client size (LSALEj) is measured by log of sales. Financial leverage

(LEVj) is measured by a long-term debt to assets ratio. We use MBj (the

market-to-book equity ratio) to proxy for growth opportunities. CAPINTj is

the capital intensity measured by gross property plant and equipment divided

by sales. The need for external capital (ISSUEj) is an indicator variable that

equals ‘one’ if the change in external equity is greater than 15% and ‘zero’

otherwise. LOSSj is a control variable that equals ‘one’ if net income is

negative and ‘zero’ otherwise.

2.2 Industry-Level Factors

2.2.1 Industry Concentration

Firms in industries characterized by small numbers of powerful,

differentiated producers (i.e. firms in concentrated industries) desire auditors

who do not also audit their competitors (Kwon [1996]). As the degree of

concentration in an industry increases, surviving clients will be reluctant to

use the same auditors, because they do not want to risk the transfer of

proprietary information to their competitors. In addition, the benefits to

investors of financial accounting information arguably are less in highly

concentrated industries (Bushman & Smith [2001]). Thus industry specialist

9
auditors are less likely to be employed in concentrated industries. 13 Our

second hypothesis, in alternate form, is:

H2: Choice of an industry specialist among Big N auditors is negatively

associated with a proxy for industry concentration.

The proxy for industry concentration is the Herfindahl index (HINDEXk),

∑
n 2
which is defined as j =1
s j where s j is market share of firm j based on sales

in industry k.14 Larger values correspond to more concentrated industries.

2.2.2 Regulated Industries

Banks, insurance companies, and several other businesses are subject to

incremental regulation in most countries. Industry regulation requires audit

firms to be familiar with the specialized reporting rules and filing

requirements set by government or private sector regulatory bodies. For

example in the U.S., the American Institute of Certified Public Accountants

(AICPA) publishes individual industry audit guidelines for oil and gas

companies, airlines, security dealers, finance companies, and other regulated

industries. Those international audit firms that invest in acquisition of

regulated industry expertise are likely to be viewed as providers of higher

13
It is worth noting that audit firms arguably have less incentive to develop industry expertise
in highly concentrated industries since, by definition, such industries tend to contain only a
few large clients. If Big N firms tend to have fairly equal shares among clients in a
concentrated industry, then designation of one firm as the industry leader creates a distinction
without a difference. Supply-side considerations therefore suggest that there might be no
association between industry concentration and employment of an industry specialist auditor.
14
Subscript “k” denotes industry membership which is determined by the SIC code following
the classification schemes used by Frankel et al. (2002): agriculture (0100–0999), mining &
construction (1000–1999, excluding 1300–1399), food (2000–2111), textiles &
printing/publishing (2200–2799), chemicals (2800–2824, 2840–2899), pharmaceuticals
(2830–2836), extractive (2900–2999, 1300–1399), financial institutions (6000–6999), durable
manufacturers (3000–3999, excluding 3570–3579 and 3670–3679), transportation (4000–
4899), utilities (4900–4999), retail (5000–5999), services (7000–8999, excluding 7370–7379),
computers (3570–3579, 3670–3679, 7370–7379).

10
quality audits, for clients in those regulated industries. 15 In essence, client

choice of an industry specialist auditor conveys a message of greater

information quality in regulated industries. Our third hypothesis, in alternate

form, is:

H3: Choice of an industry specialist among Big N auditors is positively

associated with membership in a regulated industry.

Similar to Francis et al. [1999], and following Eichenseher and Danos

[1981], we define variable REGINDk to represent regulated industries.

REGINDk is coded ‘one’ if the industry is railroad (SICs 4011 and 4100),

trucking (4210 and 4213), airlines (4512, 4513, 4522, and 4581), telephone

communications (4812 and 4813), electric companies (4911), gas companies

(4922, 4923, 4924), personal credit (6141), and insurance (6311), and ‘zero’

otherwise.

2.3 Country-Level Factors

Our expectations, regarding the effects of country-level factors on

demand for industry specialist audits, are based on the idea that the value of

audits is derived from the value of high quality financial information. In turn,

high quality information is more valuable in nations that protect investor

wealth, in nations that are more economically developed, and in which

national regulators are committed to facilitating or requiring high quality

financial information.

15
The existence of special reporting rules, filing guidelines, and audit standards arguably
serve as a barrier to entry for audit firms desiring to serve regulated clients. The largest
international audit firms are likely to have the resources needed to hurdle this barrier to entry.
It is also possible that leading national (rather than international) audit firms will acquire
knowledge of the industry-specific regulations that are particular to their home countries.

11
2.3.1 Protection of Outside Investors

Research in finance suggests that protection of outside investors is a

key determinant of financial market development, capital and ownership

structures, dividend policies, and firms’ equity valuations (e.g., La Porta et al.

[1997, 1998, 2000]). Bushman and Smith [2001] argue that the benefits of

financial accounting information are greater in countries that protect investors

against expropriation of wealth by governments. We assert that the benefits

of higher quality audits (by Big N industry specialists) likewise are greater in

such countries. Using categories of investor protection drawn from La Porta

et al. ([1997, 2000]), we investigate whether such protection has a systematic

influence on client auditor choice among the Big N. Our fourth hypothesis, in

alternate form, is:

H4: Choice of an industry specialist among Big N auditors is positively

associated with proxies for national levels of legal protection for

investors.

We employ three proxies for investor protection. The first proxy is a

law enforcement index (LAW_ENFl) which is the mean score of three legal

enforcement variables reported in La Porta et al. [1998], and used in Leuz et

al. [2003].16 The law enforcement index values range from zero to ten, with

higher scores for greater law enforcement. The second proxy for shareholder

legal protection is a measure of shareholder voting rights. ‘Antidirector

rights’, reported in La Porta et al. [1998], indicates how easy it is for

16
Subscript “l” denotes country “l”. The three variables are (1) the mean for 1980-1983 of a
variable provided by Business International Corp., capturing the efficiency and integrity of
the judicial system; (2) the mean for 1982-1995 of a rule of law variable obtained from
International Country Risk; and (3) the mean for 1982-1995 of a corruption variable that
assesses the corruption in government, obtained from International Country Risk.

12
shareholders to exercise their voting rights.17 This index ranges from zero to

five, with higher scores indicating greater protection of shareholders. For

convenience, we name this variable VOTINGl. The third proxy, LEGALl, is

the principal component for the two legal variables, derived from a factor

analysis. We expect all three variables to be positively associated with choice

of Big N industry specialist auditors.

2.3.2 Financial Reporting Environment

National financial reporting environment is an important institutional

factor that affects a company’s accounting information quality. Previous

studies suggest that higher reporting quality, through expanded disclosure, is

associated with greater market liquidity and lower cost of equity capital (see

Healy & Palepu [2001], for a detailed review). In an international setting,

Young and Guenther [2003] show that countries whose financial accounting

environments lead to greater disclosure of value-relevant accounting, are

associated with higher international capital mobility. Hence, we expect a

greater demand for high-quality (industry-specialized) Big N auditors in a

strong financial reporting environment, to assure outside investors of the

quality and credibility of accounting information in decision making (Fan &

Wong [2005]). Our fifth hypothesis, stated in alternate form, is:

17
The index aggregates the following components of shareholder rights: (1) the ability to vote
by mail; (2) the ability to gain control of shares during the shareholders’ meeting; (3) the
possibility of cumulative voting for directors; (4) the ease of calling an extraordinary
shareholders meeting; and (5) the availability of a mechanism allowing minority shareholders
to make legal claims against the directors.

13
 H5: Choice of an industry specialist among Big N auditors is positively

associated with proxies for quality of national financial reporting

environment.

We use three proxies to measure the extent of accounting information

provided in each country. The first proxy is the disclosure index (DISCl)

developed by the Center for International Financial Analysis and Research

(CIFAR [1995]), and used by Saudagaran and Diga [1997] among others. CIFAR

creates a country-specific index by rating the annual reports of at least three firms

in every country for inclusion or omission of 90 specific items.18 Each country is

given a score ranging from zero to 90, with higher scores indicating more

disclosure. The second proxy for financial reporting environment is the extent of

financial and tax accounting conformity reported by Hung [2000]. 19 In some

countries, financial reports effectively reflect tax laws, which in turn are

influenced by political, economic, and social objectives. Because the primary

objective of tax rules is not to satisfy the information needs of capital market

participants, the value relevance of financial reports in countries with high tax-

book conformity is reduced (Ali & Hwang [2000]). The value of financial

information (and the derived value of higher quality audits) potentially is greater

when there is weak link between tax and financial accounting rules. Using Hung’s

ratings, we create variable FIN_TAXl to capture the extent of tax and financial

accounting alignment (tax-to-book). FIN_TAXl is an indicator variable that equals

18
The 90 items include specific disclosures in the following seven categories: general information
(8 items), income statement (11 items), balance sheet (14 items), funds flow statement (5 items),
accounting policy disclosure (20 items), shareholders’ information (20 items), and other
supplementary information (12 items).
19
Hung [2000] classifies a country’s alignment between financial and tax accounting rules based
on the following criteria: existence of deferred taxes, dominance of legal form over substance,
allowance of additional accelerated depreciation, and dependency of amortization on tax laws.

14
‘one’ if tax accounting and financial accounting methods diverge, and ‘zero’ if

they are similar.20 The last proxy, ACCTGl, is the principal component for the

financial reporting variables, derived from a factor analysis. If higher quality

audits are more valuable in nations requiring better disclosure and financial

reporting, we expect all three variables to be positively associated with choice of

Big N industry specialist auditors.

2.3.3 National Economic Development

Claessens and Laeven [2003] and Doidge et al. [2007] posit that a

high level of national economic development is associated with higher quality

institutions that facilitate private contracting. This in turn should increase

demand for reputable information intermediaries such as high-quality,

industry-specialist auditors, to assure and certify information that is used in

the contracting. Our sixth hypothesis, stated in alternate form, is:

H6: Choice of an industry specialist among Big N auditors is positively

associated with proxies for national economic development.

We employ three metrics that reflect national economic development. The

first is the level of annual GDP per capita in each country. We expect greater

economic wealth to have a positive impact on auditor choice because many

companies in less wealthy countries may not be able to afford hiring high

quality auditors (Choi & Wong [2007]). Gross Domestic Product per capita

(GDP), based on 2000 constant prices, is collected from the World Bank’s

20
The various country-level variables (LAW_ENF, VOTING, DISC, and FIN_TAX) each are
measured at a specific point in time (in the 1980s or 1990s) rather than being measured each
year. We assume that countries’ institutional environments remain stable over lengthy periods
of time. To the extent this assumption is incorrect, it should bias against finding significant
coefficients for these variables.

15
World Development Indicators database. We log-transform the GDP data

(denoted by LGDP) since GDP is highly skewed. Our second proxy for

national economic development is the extent of development of national stock

markets. Ali & Hwang [2000] argue that investors in market-oriented

financial systems (where equity financing is more prevalent) demand more

information than those in bank-oriented financial systems (where debt

financing is more important). They also should demand higher quality

information. Fan & Wong [2005] provide evidence that firms employ high-

quality auditors when seeking external capital in the stock markets. The strong

information oversight provided by a high quality auditor is likely to reduce

information asymmetry between a company’s managers and investors, thus

lowering the cost of capital. Hence, we expect demand for industry specialist

Big N auditors to be greater where equity markets are more developed. Stock

market development (SMDEV) is measured by stock market capitalization

divided by gross (not per capita) GDP. The data for SMDEV are obtained from

Beck et al. [2000]. The final proxy is ECONl, which is the principal

component for the national economic development variables, derived from a

factor analysis. We expect all three variables to be positively associated with

choice of Big N industry specialist auditors.

3. Research Design

3.1 Measure of Auditor Industry Expertise

Auditor industry expertise is typically measured by an auditor’s

industry market share (e.g., Danos & Eichenseher [1982]; Balsam et al.

16
[2003]; Krishnan [2003]).21 We calculate a Big N auditor’s market share in a

given industry, and year, as follows:

J ik

∑ SALES
j =1
ijk

ADTR _ MSik = I k J ik
(1)
∑∑ SALES
i =1 j =1
ijk

We suppress the subscript denoting a specific year. SALES denotes client sales

revenue. The numerator is the sum of the sales of all Jik clients of audit firm i

in industry k, where the industry is as defined in Frankel et al. [2002]. The

denominator in equation (1) is the sales of all Jik clients in industry k, summed

over all Ik audit firms providing audits to that industry.22

Our measure of auditor i’s industry k expertise is based upon that

auditor’s market share among clients in industry k that are headquartered in

each individual country. The assumption is that audit firm i is more likely to

be perceived as offering expertise in industry k and country l if audit firm i has

a high market share in industry k within country l. Our dependent variable,

SPECjk, is defined as ‘one’ if client j, headquartered in nation l, purchases

audits from the auditor having a ‘large’ value of ADTR_MS in industry k in

nation l. SPECjk is defined as ‘zero’ otherwise. We consider an auditor to have

a large market share in the home industry k, if its value of ADTR_MS for k is

21
Conceptually, industry market share would be measured as audit fees earned by an auditor
in an industry, as a proportion of the total audit fees earned by all auditors that serve that
particular industry. Based on data availability, prior studies (such as Krishnan [2003]) use
client sales rather than auditor fees to compute proxies for auditor market shares in industries.
Following these studies, we use client sales to estimate industry market shares, since data
about audit fees are either costly to collect or not available in many of the sample countries
under investigation.
22
All Ik audit firms include both Big N and other audit firms.

17
at least 20% prior to 1998, 24% for the 1998-2001 period, and 30% for the

2002-2005 period.23

3.2 Model Specification

We estimate the following logistic regression model of auditor choice

using client-specific, industry-level, and country-level explanatory variables:

SPEC = αο + β1LSALEj + β2LEVj + β3MBj + β4CAPINTj + β5ISSUEj

+ β6LOSSj + β7HINDEXk + β8REGINDk + β9COUNTRYl + ejk (2)

where

SPEC = a dichotomous variable coded ‘one’ if client j purchases

audits from the auditor having at least 20%/24%/30%

market share (for the period prior to 1998, 1998-2001,

and after 2001 respectively) in industry k in the national

market, and ‘zero’ otherwise;

LSALEj = the log of client j’s sales;

LEVj = the long-term debt-to-asset ratio of client j;

MBj = client j’s market-to-book equity ratio;

CAPINTj = client j’s capital intensity measured by gross property,

plant & equipment divided by sales;

ISSUEj = a dichotomous variable coded ‘one’ if client j’s book

value of equity increases by more than 15% from the

prior year; ‘zero’ otherwise;

23
Following Neal & Riley [2004], we employ a cut off for ‘large’ market shares of (1/N)*1.2,
where N is the number of big international audit firms. The largest firms are the Big 6, during
the period 1993-1997, the Big 5 after the merger between Coopers and Lybrand and Price
Waterhouse in 1998, and Big 4 after the demise of Arthur Andersen in 2002.

18
LOSSj = a dichotomous variable coded ‘one’ if client j’s net

income is negative; ‘zero’ otherwise;

HINDEXk = Herfindahl concentration ratio of client j’s industry k;

REGINDk = a dichotomous variable coded ‘one’ if client j’s industry

k is regulated; ‘zero’ otherwise. See the text for

specification of regulated industries;

COUNTRYl = Country-level variables: LAW_ENFl, VOTINGl, LEGALl,

DISCl, FIN_TAXl, ACCTGl, LGDPl , SMDEVl and

ECON;

COUNTRY-level variables include:

LAW_ENFl = a law enforcement index for country l (see definition in

text); higher scores indicate better law enforcement;

VOTINGl = an investor voting rights index for country l (see

definition in text); higher scores indicate better voting

rights;

LEGALl = principal component extracted from LAW_ENF and VOTING;

DISCl = an information disclosure index for country l (see

definition in text); higher scores indicate more extensive

disclosure;

FIN_TAXl = a dichotomous variable coded ‘one’ if country l’s tax

accounting and financial reporting methods diverge

substantially (Hung [2000]); ‘zero’ otherwise;

ACCTGl = principal component extracted from DISC and FIN_TAX;

19
LGDPl = natural logarithm of GDP per capita for country l;

SMDEVl = stock market capitalization divided by gross GDP for

country l;

ECONl = principal component extracted from LGDP and SMDEV.

4. Empirical Results

4.1 Sample

Data for the sample period 1993-2005 are obtained from the Global

Vantage database. Our sample countries are determined as follows. First, we

begin with the 49 countries (with the necessary institutional data) that are

listed in La Porta et al. [1998]. To meaningfully compute and compare

auditors’ market shares, and clients’ industry concentration ratios across

industries and countries, we require at least 10 observations in each industry,

for a particular year, to be included in the sample. A total of 35 countries meet

the criterion (that is, have one or more industries containing 10 companies).

We remove Japan, Korea, and Pakistan from the sample, because the identity

of the auditors is not indicated in the database for these countries.24 We next

remove three countries (Ireland, New Zealand, and Turkey) since most of the

data for control variables are not available for these countries in the Global

Vantage database. Our final sample consists of the following 29 countries:

Australia, Austria, Belgium, Brazil, Canada, Chile, Denmark, Finland, France,

Germany, Greece, Hong Kong, India, Indonesia, Israel, Italy, Malaysia,

24
More than 95% of the auditor names are designated as ‘others’ in Global Vantage for these
three countries. For example, the local name for KPMG in Japan is Kainan Audit Corporation,
and it is recorded by Global Vantage as ‘other’. Hence, we are not able to obtain the identity
of the auditors in these countries.

20
Mexico, Netherlands, Norway, Philippines, Singapore, South Africa, Spain,

Sweden, Switzerland, Taiwan, Thailand, and the United Kingdom.

We employ 14 broadly-defined industries based on Frankel et al.

[2002]’s classification, providing a total of 39,053 firm-years, usable for the

study.25 Panels A, B, and C of Table 1 report the distribution of observations

by country, year, and industry, respectively. In Panel A, there is significant

variation in the number of firm-year observations across countries due to

differences in capital market development, country size, and the availability

of complete financial accounting data. Except for the United Kingdom, no

country contributes as much as eleven percent of the total firm-years. In the

subsequent sensitivity analysis, we find that our results are robust after

excluding clients in the United Kingdom, and after excluding clients located

in developed countries in general. The percent of companies audited by Big N

firms exceeds fifty percent for a majority of the sample countries. Countries

included in our tests are all drawn from Big N clients.

In Panel B, the number of firm-years grows over time because of the

increasing coverage of international companies in the Global Vantage

database. The decrease in firm-years for 2005 reflects incomplete database

coverage of that year. Panel C reports the distribution of observations by

industry. The largest number of firm-years is found among durable

manufacturers.

[Insert Table 1 here]

25
We winsorize each of the continuous control variables (LEV, CAPINT, LSALE, MB) used in the
regression at the top and bottom one percent to remove extreme values.

21
4.2 Descriptive Statistics for Variables

Table 2, Panel A, shows the means of the dependent variable, SPEC,

the means for client-specific variables (SALE, LEV, MB, CAPINT, ISSUE,

LOSS), and means for industry-level variables (HINDEX and REGIND), by

country. Table 2, Panel B presents the country-level explanatory variables. 26

The values for VOTING, LAW_ENF, and DISC are constant over time for an

individual country, e.g., the value for LAW_ENF for Australia is 9.51 for each

year. The values for SMDEV and GDP are mean values, computed over 1993-

2005. We note that the amounts of our country-level explanatory variables

differ substantially across countries having highly developed economies.

Consider, for example, investors’ voting rights (VOTING). The level of this

variable is highest in the U.K. and its former colonies such as Australia and

Canada. It is substantially lower in some countries in continental Europe:

Austria, Belgium, Denmark, and Germany. Yet all the countries named have

developed economies. The same is true to some extent for the other country-

level variables.

[Insert Table 2 here]

4.3 Variable Correlations

Table 3 presents the Pearson correlations for the dependent and

explanatory variables. We apply a log transformation to country GDP

(denoted as LGDP) due to the highly skewed distribution of the variable.

Consistent with our expectation, SPEC is positively and significantly

26
Eleven countries have a low conformity of accounting and tax reporting, while eight
countries have high conformity of accounting and tax rules. Ten countries are not coded as
these countries are not included in Hung [2000]’s sample and hence not reported in Hung’s
study.

22
correlated with most of the explanatory variables (LSALE, LEV, CAPINT,

REGIND, LAW_ENF, VOTING, LEGAL, FIN_TAX, ACCTG, LGDP).

However, inconsistent with our expectation, SPEC is positively associated

with HINDEX, and negatively associated with SMDEV and ECON. We

emphasize that these are univariate correlations, and we rely on the regression

analyses to draw inferences. The correlations among the country level

variables are high, so we employ them in two ways. First, we estimate models

in which we enter only one country-level variable at a time (or one country-

level principal component factor) into our regressions. Second, we estimate

one model in which all principal components extracted from the legal,

financial reporting and national economic development variables are entered

simultaneously. 27 The former regressions allow us to determine which

country-level variables (including principal component factors) are

individually significant, and whether any client-level or industry-level

variables are sensitive to inclusion of specific country-level variables. The

latter regression indicates which of the country-level principal component

factors have explanatory power when all are included.

[Insert Table 3 here]

4.4 Regression Results

The (logistic) regression model results of estimating equation (2) are

presented in Table 4. The model is estimated using a maximum of 39,053

firm-year observations. In five results columns the number of observations is

27
This procedure reduces a potential multi-collinearity problem. Factor analysis indicates that
LAW_ENF and VOTING are measuring one construct. The same is true for LGDP and
SMDEV, and for FIN_TAX and DISC. Hence, we use the principal components, LEGAL,
ACCTG, and ECON, derived from the factor analysis, in the combined model.

23
less due to missing data for variables DISC and FIN_TAX. The regressions in

Table 4 employ multiple observations per client over time. Such observations

are unlikely to be fully independent, and thus regression residuals may be

serially or cross-sectionally correlated. Hence, we run logistic regressions

with clustered robust errors to account for both serial and cross-sectional

correlations (Rogers [1993]; Williams [2000]; Petersen [2006]). Following

Choi & Wong [2007], for all tests reported below, the Wald-statistics are

based on clustered standard errors (clustered by country and year).28

The dependent variable captures whether or not the Big N auditor

chosen is an industry specialist. The first nine regression models include only

one country-level variable each, while the last regression model includes all

principal components derived from the factor analysis (LEGAL, ACCTG, and

ECON). Consistent with our expectation, SPEC is positively and significantly

associated with LSALE, MB, and CAPINT at the 1% level, providing support

for hypothesis H1 (i.e. three out of five proxies are significant with expected

signs). LEV is positively associated with SPEC in five models while ISSUE is

either not associated with SPEC, or is associated with a coefficient sign

opposite to expectation. LOSS is positively associated with SPEC in several

models. Among the industry-level variables, SPEC is consistently and

positively associated with REGIND, in agreement with H3. However,

inconsistent with H2, we do not find a significant and consistent association

between HINDEX and SPEC. This finding agrees with a recent study by the

General Accounting Office [2003]. The GAO reports that when large public

companies were asked whether they would choose an accounting firm as their

28
We also run the regression clustered by country, industry, and year. The results are similar.

24
auditor when that firm also audits one of their competitors, 92% of the

respondents said ‘yes’.29

Table 4 also explores the explanatory power of the country-level

variables. Because of the substantial correlations between some country-level

variables, we present some models that include only one of the country-level

variables in each regression. This enables us to determine whether the

significance levels of individual client-level and industry-level variables are

contingent on the inclusion of particular country-level variables. Consider

first the proxies employed to test H4: LAW_ENF, VOTING, and LEGAL. In

models 1-3, the coefficients of of these variables are positive as expected and

highly significant. Together, the coefficient results offer strong support for

H4: choice of an industry specialist auditor is positively associated with

proxies for national levels of legal protection. The significance of other

variables is unaffected by which of these country-level variables is included

in the model.

Next consider the results for variables used to test H5: DISC,

FIN_TAX, and ACCTG. In models 4-6, the coefficients for these variables are

positive as expected and highly significant. Together, the coefficient results

offer strong support for H5: choice of an industry specialist auditor is

positively associated with proxies for national quality of financial reporting

and disclosures. Inclusion of DISC has no effect on the significance of other

variables in the model. However, inclusion of either FIN_TAX or ACCTG has

29
Another possible explanation is that the Herfindahl index is a poor proxy for industry
concentration. We compute it using Global Vantage database, so only publicly traded
companies’ data are employed in constructing this index. Ideally the index should be
computed using data for both public and private firms (Ali et al. [2008]). We do not employ
private companies’ data because it is unavailable for most clients in an international setting.

25
the effect of reducing the explanatory power of REGIND, and increasing the

explanatory power of ISSUE and LOSS. Finally, consider the results for

variables used to test H6: LGDP, SMDEV and ECON. In models 7-9, the

coefficients for these variables are positive and significant, offering support

for H6. Inclusion of each of these variables does not affect the significance of

other variables in the models. Choice of an industry specialist auditor is

positively associated with proxies for national economic development.

We include all three country-level variables extracted from the factor

analysis in model 10. The coefficients of ACCTG and ECON are positive and

highly significant, while the coefficient of LEGAL is positive and marginally

significant. Our results suggest that legal environment, financial reporting

quality and national economic development have incremental power in

explaining choice of industry specialists among Big N auditors. Overall, our

results in Table 4 support the importance of several client-specific factors in

the decision to choose Big N industry-specialist auditors: size, growth

prospects, and capital intensity. The results also indicate the importance of

client membership in a regulated industry, but not of industry concentration.30

Finally, the results suggest that three country-level institutional factors affect

choice of industry specialists from among Big N auditors.

[Insert Table 4 here]

4.5 Additional Analyses

4.5.1 Alternative Dependent Variable

30
We note that a client-level factor, capital intensity, is significant. High capital intensity
serves as a barrier to entry, and capital intensity varies across industries. Thus it is likely that
the capital intensity variable pre-empts some of the explanatory power of industry
concentration.

26
 We perform several sensitivity tests to check the robustness of our

results. First, as an alternative to dependent variable SPEC, we define SPEC1jk

as equal to ‘one’ if client j, headquartered in nation l, purchases audits from

the auditor having the largest value of ADTR_MS in industry k in nation l.

SPEC1jk is defined as ‘zero’ otherwise. In essence, this dependent variable

defines only one auditor per industry-country pair as an industry specialist.

Results for the alternative dependent variable SPEC1 are reported in

Table 5. The results are consistent with those reported in Table 4, with the

following major exceptions. First, HINDEX is positively and significantly

associated with SPEC1, while in our main analysis of Table 4 we do not find

any association between HINDEX and SPEC. Second, the coefficients of

REGIND are less significant than in Table 4. Although the Table 4 and Table

5 results are largely in agreement, in those respects where they differ we

prefer to rely on Table 4. SPEC1 arguably is overly restrictive in limiting

industry-specialist status to a single Big N firm per industry-country pair.

[Insert Table 5 here]

4.5.2 Stability of Audit Specialists over Time

A maintained hypothesis of this study is that identities of Big N

industry specialists are fairly stable over the entire sample period. 31 We

attempt to measure the stability of audit specialists over three sub-periods:

before 1998 (Big N = Big 6), between 1998 and 2001 (Big N = Big 5), and

2002-2005 (Big N = Big 4). We compute a stability metric for each industry-

country pair over each sub-period. For example, consider the 2002-2005 sub-

31
If this assumption is not valid it should bias against the significance of our explanatory variables.

27
period during which Big N = Big 4. Suppose for a particular country (say

Singapore), and a particular industry (say Computers), the market leader for

the entire four years is KPMG. Then the stability metric is “1.” If KPMG is

the market leader for three of the four years, then the stability metric is ‘0.75.’

Hence the stability statistic captures the proportion of years that the same

auditor is the market leader in the industry over a specific period. The larger

the statistic is, the greater the stability of audit specialists over time. The mean

statistic for all countries and all industries for the period 2002-2005 is 0.87.

This number is quite high, indicating stable audit specialists over the years

2002-2005. We repeat the analyses for the period 1998-2001 and for 1993-

1998. The mean statistics are 0.87 and 0.90 respectively. 32 Overall, the

stability statistic is high, and does not vary much across sub-periods,

suggesting that the identities of audit specialists are stable over the entire

sample period 1993-2005.

4.5.3 Removing Firms cross-listed in the U.S. Market

We remove 2,119 firm-years relating to those firms that cross-list

their shares in the U.S. market, since the firms’ auditor choice may be related

to their overseas equity issues (Fan & Wong [2005]). The results, as reported

in Table 6, provide similar results as those in Table 4 except for the following

major differences. First, the coefficients of REGIND are not significant in

models 5, 6 and 10. Second, the coefficient of SMDEV is not significant in

32
We compute metrics for three sub-periods (rather than for the entire sample period) because the
identities of Big N industry specialists are non-comparable across periods due to Big N mergers.

28
model 8. Third, the coefficients of HINDEX are negative and significant in

models 5, 6, 7 and 10.

[Insert Table 6 here]

4.5.4 Alternative proxies for financial reporting

To supplement results for variables DISC and FIN_TAX, we use two

alternative measures to capture the quality and quantity of accounting

disclosure: (1) the disclosure index constructed by Young and Guenther

[2003]; and (2) the accrual index constructed by Hung [2000]. Our

(untabulated) results indicate that the disclosure index and accrual index are

positively and significantly associated with SPEC at the 1% level. We also

construct a principal component based on these two alternative proxies for

financial reporting quality. Our untabulated results indicate that, along with

LEGAL and ECON, this alternative principal component is positively and

significantly associated with SPEC at the 1% level.

4.5.5 Legal Origins

A concurrent study (Francis & Wang [2008]) suggests that demand

for Big N auditors is dependent on a country’s legal origin. We therefore test

whether legal origin (COMMON, coded ‘one’ if the country has a common

law origin, and ‘zero’ if the country has a code law origin) also is associated

with industry specialist auditor choice. The (untabulated) results indicate that

COMMON is positively and significantly associated with SPEC (Wald

statistic 5.98, p=0.01). This finding suggests that legal origin (similar to

LAW_ENF and VOTING) captures the legal environment in each country and

29
is positively associated with demand for industry specialist auditors. We also

estimate a model that includes COMMON, along with ACCTG and ECON.

Our untabulated results indicate that ACCTG and ECON continue to be

significant at the 1% level, but COMMON is not statistically significant.33

4.5.6 Clients’ Demand for Audit Specialists in Less Developed Countries

In our main analysis, we show that demand for audit specialists is

systematically associated with client and industry characteristics. Some of

these results have been documented previously in developed economies (e.g.,

Australia) and it is worthwhile to explore whether such relations still hold in

developing economies. We address this issue in two ways to ensure that our

findings are not being driven by the large number of observations provided by

the developed economies. First, given that firms in the United Kingdom

constitute 25% of the total sample, we examine whether the exclusion of these

firms affects the main results. The results are similar to those reported in

Table 4 except for the following. First, the coefficients of LEV are positive

and significant for only three (out of ten) models. Second, DISC and SMDEV

are not significant in models that contain each of them as the only country-

level variable. Third, the coefficient of LEGAL is no longer significant when

ACCTG and ECON are also added to the model.

Second, we document results after deleting all developed countries

from the sample. To identify developed and developing countries, we use the

33
One possible explanation for this non-result might be that COMMON does not capture the
variation in legal enforcement that exists even within the common- and code- law countries.
For example, Thailand has a common law origin, but it has low enforcement
(LAW_ENF=0.48). On the other hand, Sweden has a code-law origin but it has high
enforcement (LAW_ENF=10).

30
DEV index as reported in Table 1 of Hail and Leuz [2006]. Specifically, a

country is considered as developing if its equity market is not included in the

Morgan Stanley Capital International database. Based on this definition, 12

countries are considered as developing, namely: Brazil, Chile, Greece, India,

Indonesia, Israel, Malaysia, Mexico, Philippines, South Africa, Taiwan, and

Thailand. We re-estimate our models using observations drawn only from

these developing economies. The results are reported in Table 7.

The first two models report the regression results for the full sample,

using dependent variables SPEC and SPEC1, while the last two models report

the regression results for the same dependent and explanatory variables, using

a sample that includes only clients in developing countries. Although sample

size decreases dramatically in models 3 and 4 (7,511 observations versus

39,053), the results for firms in developing countries are remarkably similar to

those in models 1 and 2. This shows that the characteristics of clients that

choose industry specialists from among the Big N auditors do not vary

systematically between developed and developing countries. In untabulated

results, we include the nine country-level variables in the models that are

estimated using only clients from developing countries. Our results indicate

that the coefficient estimates for LSALE, LEV, MB, and CAPINT continue to

be significant. For the country-level variables, only VOTING, SMDEV, and

LEGAL are positive and significant.

[Insert Table 7 here]

5. Conclusion

31
 In this study, we investigate client choice of industry specialist

auditors, from among the Big N, in an international (non-U.S.) setting. We

investigate client-, industry-, and country-level factors hypothesized to

enhance or decrease client demand for audits by auditors having industry

expertise. We measure auditors’ industry expertise based on industry market

shares in clients’ home countries. Using data for 29 countries and 14

industries from 1993-2005, we find strong evidence that international client

choice of Big N industry specialist auditors is positively associated with

client size, client growth opportunities, and client capital intensity. At the

country level, the choice of industry specialists among Big N auditors is

strongly associated with legal environment, quality of financial reporting, and

with national economic development. Our results also suggest, less strongly,

that choice of industry specialists from among the Big N auditors is higher in

regulated industries. Our results are generally robust to several sensitivity

tests.

32
 REFERENCES

Ali, A., S. Klasa, and P. E. Yeung. 2008. “The Limitations of Industry

Concentration Measures Constructed with Compustat Data: Implications
for Finance Research.” Review of Financial Studies (forthcoming).

Ali, A., and L. Hwang. 2000. “Country-specific Factors Related to Financial

Reporting and the Value Relevance of Accounting Data.” Journal of
Accounting Research 38 (Spring): 1-21.

Balsam, B., Krishnan, J., and J. Yang. 2003. “Auditor Industry Specialization
and Earnings Quality.” Auditing: A Journal of Practice & Theory 22
(September): 71-97.

Beck, T. A. Demirguc-Kunt, and R. Levine. 2000. “A New Database on the

Structure and Development of the Financial Sector.” World Bank
Economic Review 14 (September): 597-605.

Bedard, J., and S. Biggs. 1991. “The Effect of Domain-Specific Experience

on Evaluation of Management Representation in Analytical Procedures.”
Auditing: A Journal of Practice & Theory 10 (Supplement): 77-95.

Bushman, R., and A. Smith. 2001. “Financial Accounting Information and

Corporate Governance.” Journal of Accounting and Economics 32
(December): 237-333.

Carcello, J., and A. Nagy. 2004. “Client Size, Auditor Specialization and
Fraudulent Financial Reporting.” Managerial Auditing Journal 19 (5):
651-668.

Center for International Financial Analysis and Research (CIFAR). 1995.

International Accounting and Auditing Trends. Princeton, NJ: CIFAR
Publications, Inc.

Choi J., and T. Wong. 2007. “Auditors’ Governance Functions and Legal
Environments: An International Investigation.” Contemporary
Accounting Research 24 (Spring): 1-36.

Claessens, S., and L. Laeven. 2003. “Financial Development, Property Rights,

and Growth.” Journal of Finance 58 (December): 2401-2436.

Craswell, A., Francis, J., and S. Taylor. 1995. “Auditor Brand Name
Reputations and Industry Specializations.” Journal of Accounting and
Economics 20 (December): 297-322.

Danos, P., and J. Eichenseher. 1982. “Audit Industry Dynamics: Factors

Affecting Changes in Client-Industry Market Share.” Journal of
Accounting Research 20 (Autumn): 604-616.

33
De Beelde, I. 1997. “An Exploratory Investigation of Industry Specialization
of Large Audit Firms.” The International Journal of Accounting 32 (3):
337-355.

DeFond, M. 1992. “The Association Between Changes in Client Firm

Agency Costs and Auditor Switching.” Auditing: A Journal of Practice &
Theory 11 (Spring): 16-31.

Doidge, C., Karolyi, G., and R. Stulz. 2007. “Why Do Countries Matter So
Much for Corporate Governance?” Journal of Financial Economics 86
(October): 1-39.

Dunn, K., B. Mayhew. 2004. “Auditor Firm Industry Specialization and

Client Disclosure Quality.” Review of Accounting Studies 9 (March): 35-
58.

Eichenseher, J., and P. Danos. 1981. “The Analysis of Industry-Specific

Auditor Concentration: Toward an Explanatory Model.” The Accounting
Review 56 (July): 479-492.

Fan, J. and T. Wong. 2005. “Do External Auditors Perform a Corporate

Governance Role in Emerging Markets? Evidence from East Asia.”
Journal of Accounting Research 43 (March): 35-72.

Ferguson, A., J. Francis, and D. Stokes. 2003. “The Effects of Firm-Wide and
Office-Level Industry Expertise on Audit Pricing.” The Accounting
Review 78 (April): 429-48.

Ferguson, A., and D. Stokes. 2002. “Brand Name Audit Pricing, Industry
Specialization, and Leadership Premiums Post-Big 6 and Big 8 Mergers.”
Contemporary Accounting Research 19 (Spring): 77-110.

Francis, J., and E. Wilson. 1988. “Auditor changes: A Joint Test of Theories
Relating to Agency Costs and Auditor Differentiation.” The Accounting
Review 63 (October): 663-682.

Francis, J., Maydew, E., and H. Sparks. 1999. “The Role of Big 6 Auditors in
the Credible Reporting of Accruals.” Auditing: A Journal of Practice &
Theory 18 (Fall): 17-34.

Francis, J.R., Reichelt, K. and D. Wang. 2005. “The Pricing of National and
City-Specific Reputations for Industry Expertise in the U.S. Audit
Market.” The Accounting Review 80 (January): 113-136.

Francis, J., Reichelt, K. and D. Wang. 2006. National Versus Office-Specific

Measures of Auditor Industry Expertise and Effects on Client Earnings
Quality. Working paper, University of Missouri-Columbia.

34
Francis, J., and D. Wang. 2008. “The Joint Effect of Investor Protection and Big 4
Audits on Earnings Quality Around the World.” Contemporary Accounting
Research 25 (Spring): 157-191

Frankel, R. M., M. F. Johnson, and K. K. Nelson. 2002. “The Relation Between

Auditors’ Fees for Non-Audit Services and Earnings Management.” The
Accounting Review 77 (Supplement): 71–106.

General Accounting Office. 2003. Accounting Firm Consolidation. Selected

Large Public Company Views on Audit Fees, Quality, Independence, and
Choice. (September).

Godfrey, J. and J. Hamilton. 2005. “The Impact of R&D Intensity on Demand for
Specialist Auditor Services.” Contemporary Accounting Research 22 (Spring):
55-93.

Hail, L., and C. Leuz. 2006. “International Differences in the Cost of Equity
Capital: Do Legal Institutions and Securities Regulation Matter?” Journal of
Accounting Research 44 (June): 485-531.

Healy, P. and Palepu, K. 2001. “Information Asymmetry, Corporate

Disclosure, and the Capital Markets: A Review of the Empirical
Disclosure Literature. Journal of Accounting and Economics 31
(September): 405-440.

Hogan, C., and D. Jeter. 1999. “Industry Specialization by Auditors.”

Auditing: A Journal of Practice & Theory 18 (Spring): 1-17.

Hung, M. 2000. “Accounting Standards and Value Relevance of Financial

Statements: An International Analysis.” Journal of Accounting and
Economics 30 (December): 401-420.

Krishnan G. 2003. “Does Big 6 Auditor Industry Expertise Constrain

Earnings Management?” Accounting Horizons 17 (Supplement): 1-16.

Kwon, S. 1996. “The Impact of Competition Within the Client’s Industry on

the Auditor Selection Decision.” Auditing: A Journal of Practice &
Theory 15 (Spring): 53-70.

La Porta, R., Lopez-de-Silanes, F., Shleifer, A., and R. Vishny. 1997. “Legal
Determinants of External Finance.” Journal of Finance 52 (July): 1131-
1150.

La Porta, R., Lopez-de-Silanes, F., Shleifer, A., and R. Vishny. 1998. “Law
and Finance.” Journal of Political Economy 106 (December): 1113-1155.

La Porta, R., Lopez-de-Silanes, F., Shleifer, A., and R. Vishny. 2000.

“Investor Protection and Corporate Governance.” Journal of Financial
Economics 58 (October): 3-27.

35
Leuz, C., Nanda, D., and P. Wysocki. 2003. “Earnings Management and
Investor Protection: An International Comparison.” Journal of Financial
Economics 69 (September): 505-527.

Mayhew B., and M. Wilkins. 2003. “Audit Firm Industry Specialization as a

Differentiation Strategy: Evidence From Fees Charged to Firms Going
Public.” Auditing: A Journal of Practice & Theory 22 (September): 33-
52.

Neal, T., and R. Riley Jr. 2004. “Auditor Industry Specialist Research
Design.” Auditing: A Journal of Practice & Theory 23 (September): 169-
177.

O’Keefe, T., Kin, R., and K. Gaver. 1994. “Audit Fees, Industry
Specialization, and Compliance with GAAS Reporting Standards.”
Auditing: A Journal of Practice & Theory 13 (Fall): 41-55.

Owhoso, V., Messier, W., and J. Lynch. 2002. “Error Detection by Industry-
Specialized Teams During Sequential Audit Review.” Journal of
Accounting Research 40 (June): 883-900.

Petersen, M. 2006. “Estimating Standard Errors in Finance Panel Data Sets:

Comparing Approaches. Working paper, Northwestern University and
NBER.

Radebaugh, L., and S. Gray. 1997. International Accounting and

Multinational Enterprises. Fourth Edition. New York, NY: John Wiley &
Sons, Inc.

Rogers, W. 1993. “Regression Standard Errors in Clustered Samples.” STATA

Technical Bulletin 13: 19-23.

Saudagaran, S., and J. Diga. 1997. “Financial Reporting in Emerging Capital

Markets: Characteristics and Policy Issues.” Accounting Horizons 11
(June): 41-64.

Securities and Exchange Commission. 2000. Concept Release on International

Accounting Standards. Release No. 33-7801 (February 16). Washington, D.C.:
SEC.

Simunic, D., and M. Stein. 1987. Product differentiation in auditing: Auditor

choice in the market for unseasoned new issues. Vancouver BC: Canadian
Certified General Accountants’ Research Foundation.

Solomon, I., Shields, M., and R. Whittington. 1999. “What Do Industry-

Specialist Auditors Know?” Journal of Accounting Research 37 (Spring):
191-208.

36
Williams, R. 2000. “A Note on Robust Variance Estimation for Clustered-
Correlated Data.” Biometrics 56 (June): 645-646.
Wright, S., and A. Wright. 1997. “The Effect of Industry Specialization on
Hypothesis Generation and Audit Planning Decisions. Behavioral
Research in Accounting 9: 273-294.

Young, D., and D. Guenther. 2003. “Financial Reporting Environments and

International Capital Mobility.” Journal of Accounting Research 41
(June): 553-57.

37
 TABLE 1
Distribution of Sample Firms by Country, Year, and Industry

Panel A: Sample distribution by country

No. of Percent of Percent of firms
firm-years Firms represented audited by Big N
Country in the sample
Australia 2,097 5.37 78.61
Austria 149 0.38 47.70
Belgium 232 0.59 62.50
Brazil 641 1.64 82.98
Canada 3,970 10.17 91.41
Chile 253 0.65 87.68
Denmark 773 1.98 83.30
Finland 529 1.35 72.55
France 2,059 5.27 42.38
Germany 2,667 6.83 46.54
Greece 70 0.18 36.30
Hong Kong 593 1.52 87.71
India 136 0.35 8.96
Indonesia 182 0.47 45.65
Israel 81 0.21 56.09
Italy 979 2.51 91.81
Malaysia 3,341 8.56 65.27
Mexico 228 0.58 75.25
Netherland 1,296 3.32 91.21
Norway 659 1.69 92.24
Philippines 112 0.29 23.78
Singapore 2,405 6.16 86.02
South Africa 388 0.99 87.62
Spain 586 1.50 90.94
Sweden 1,682 4.31 80.63
Switzerland 1,107 2.83 77.30
Taiwan 1,234 3.16 77.30
Thailand 845 2.16 39.25
United kingdom 9,759 24.99 80.30
Total 39,053 100.00

38
 TABLE 1 (continued)

Panel B: Sample distribution by year

No. of Percent of firm-years
Year Firm-years represented in sample
1993 1,340 3.43
1994 1,565 4.01
1995 2,045 5.24
1996 2,446 6.26
1997 3,227 8.26
1998 3,820 9.78
1999 4,041 10.35
2000 3,943 10.10
2001 3,782 9.68
2002 3,694 9.46
2003 3,597 9.21
2004 3,620 9.27
2005 1,933 4.95
39,053 100.00

Panel C: Sample distribution by industry

No. of Percent of firm-years
Industry Firm-years represented in sample
Agriculture 321 0.82
Chemicals 929 2.38
Computers 3,894 9.97
Durable manufacturers 11,230 28.76
Extractives 1,170 3.00
Financial institutions 594 1.52
Food 2,102 5.38
Mining & construction 2,756 7.06
Pharmaceuticals 540 1.38
Retail 4,658 11.93
Services 3,513 9.00
Textile & printing/publishing 3,734 9.56
Transportation 2,718 6.96
Utilities 894 2.29
Total 39,053 100.00

The sample consists of 39,053 client firm-years for 29 countries over the period 1993-
2005. The sample only includes clients audited by the Big N. Panels A and B show the
distribution of observations by country and year respectively. Panel C shows the
distribution of observations by industry. Following Frankel et al. (2002), industry
membership is determined by the SIC code as follows: agriculture (0100–0999), mining
& construction (1000–1999, excluding 1300–1399), food (2000–2111), textiles &
printing/publishing (2200–2799), chemicals (2800–2824, 2840–2899), pharmaceuticals
(2830–2836), extractive (2900–2999, 1300–1399), financial institutions (6000–6999),
durable manufacturers (3000–3999, excluding 3570–3579 and 3670–3679),
transportation (4000–4899), utilities (4900–4999), retail (5000–5999), services (7000–
8999, excluding 7370–7379), computers (3570–3579, 3670–3679, 7370–7379).

39
 TABLE 2
Means for Firm- and Industry Level Explanatory Variables and Country-Level Index Values
Panel A: Mean firm- and industry level variables
Country SPEC SALE LEV MB CAPINT ISSUE LOSS HINDEX REGIND
Australia 0.40 1,233 0.15 2.27 1.63 0.31 0.22 0.20 0.02
Austria 0.53 2,796 0.10 1.08 1.74 0.25 0.13 0.15 0.00
Belgium 0.32 12,582 0.13 5.41 0.73 0.33 0.25 0.26 0.00
Brazil 0.67 4,421 0.16 4.19 1.93 0.31 0.25 0.10 0.15
Canada 0.45 1,383 0.17 2.33 2.81 0.41 0.27 0.15 0.07
Chile 0.62 77,143 0.13 9.20 1.85 0.19 0.15 0.24 0.23
Denmark 0.59 4,609 0.16 2.12 0.79 0.23 0.20 0.18 0.02
Finland 0.54 4,173 0.15 2.12 0.55 0.29 0.20 0.23 0.00
France 0.26 9,520 0.14 2.44 0.65 0.28 0.21 0.14 0.03
Germany 0.46 5,421 0.09 2.18 0.87 0.25 0.28 0.15 0.04
Greece 0.44 41,559 0.14 3.06 0.80 0.43 0.06 0.15 0.00
Hong Kong 0.46 4,408 0.09 1.50 1.43 0.25 0.26 0.22 0.14
India 0.01 15,064 0.13 4.78 0.60 0.39 0.03 0.10 0.04
Indonesia 0.36 95,713 0.07 9.32 1.54 0.37 0.41 0.24 0.08
Israel 0.16 308 0.06 3.86 0.53 0.42 0.42 0.35 0.00
Italy 0.41 13,993 0.13 8.73 1.16 0.22 0.23 0.24 0.09
Malaysia 0.30 572 0.08 1.64 1.59 0.24 0.26 0.09 0.02
Mexico 0.45 20,370 0.20 4.27 1.40 0.41 0.16 0.21 0.14
Netherland 0.45 3,135 0.13 3.75 0.58 0.30 0.17 0.25 0.03
Norway 0.49 4,913 0.19 2.70 1.68 0.36 0.29 0.33 0.04
Philippines 0.46 6,302 0.13 1.45 8.13 0.23 0.41 0.28 0.01
Singapore 0.41 405 0.09 1.70 1.20 0.26 0.26 0.14 0.03
South Africa 0.54 6,488 0.07 2.14 0.85 0.45 0.12 0.15 0.00
Spain 0.66 28,100 0.12 2.91 1.37 0.34 0.07 0.19 0.06
Sweden 0.41 13,001 0.12 2.29 0.98 0.36 0.32 0.19 0.03
Switzerland 0.42 2,604 0.16 2.27 1.12 0.25 0.14 0.21 0.11
Taiwan 0.36 25,887 0.12 1.77 1.06 0.33 0.20 0.08 0.01
Thailand 0.31 6,466 0.11 1.67 1.66 0.29 0.20 0.15 0.02
United Kingdom 0.45 1,477 0.12 3.20 1.06 0.29 0.21 0.10 0.04

40
 TABLE 2 (continued)

Panel B: Country level variables

Country LAW_ENF VOTING LEGAL DISC FIN_TAX ACCTG GDP SMDEV ECON
Australia 9.51 4 1.03 80 1 0.05 20,222 1.04 0.66
Austria 9.36 2 0.00 62 - - 21,073 0.14 -0.26
Belgium 9.44 0 -0.94 68 0 -0.24 21,479 0.86 0.51
Brazil 6.13 3 -0.60 56 1 2.19 3,389 0.29 -1.15
Canada 9.75 5 1.60 75 1 0.50 18,473 0.51 0.06
Chile 6.52 5 0.50 78 - - 4,822 0.92 -0.28
Denmark 10 2 0.22 75 1 0.50 28,272 0.56 0.35
Finland 10 3 0.71 83 0 -1.58 18,650 0.32 -0.15
France 8.68 3 0.26 78 0 -1.13 20,148 0.45 0.04
Germany 9.05 1 -0.59 67 0 -0.15 21,310 0.36 -0.03
Greece 6.82 2 -0.86 61 - - 12,420 0.58 -0.11
Hong Kong 8.91 5 1.31 73 1 0.68 22,107 2.48 2.23
India 5.58 5 0.18 61 1 1.75 395 0.33 -2.34
Indonesia 2.9 2 -2.18 - - 736 0.28 -2.03
Israel 7.72 3 -0.07 74 - - 18,311 0.40 -0.07
Italy 7.07 1 -1.26 66 0 -0.06 16,303 0.13 -0.42
Malaysia 7.72 4 0.42 79 - - 3,393 2.82 1.53
Mexico 5.37 1 -1.83 71 - - 5,531 0.26 -0.91
Netherland 10 2 0.22 74 1 0.59 21,154 1.13 0.78
Norway 10 4 1.19 75 1 0.50 31,325 0.19 0.02
Philippines 3.47 3 -1.50 64 - - 952 0.66 -1.48
Singapore 8.93 4 0.83 79 1 0.14 20,842 1.35 1.01
South Africa 6.45 5 0.48 79 1 0.14 2,876 1.51 0.04
Spain 7.14 4 0.23 72 0 -0.59 14,511 0.84 0.26
Sweden 10 3 0.71 83 0 -1.58 20,837 0.47 0.08
Switzerland 10 2 0.22 80 0 -1.31 31,168 1.39 1.28
Taiwan 7.4 3 -0.17 58 - - 12,662 1.03 0.39
Thailand 4.89 2 -1.51 66 - - 2,134 0.67 -1.01
United Kingdom 9.22 5 1.42 85 1 -0.40 21,609 1.33 1.01

41
 TABLE 2 (continued)
Definitions of variables:
SPEC = 1 if client j purchases audits from the auditor having at least 20%/24%/30% market share (for the period prior to 1998, 1998-2001, and after
2001 respectively) in industry k in the national market, and 0 otherwise;
SALE = sales in millions US$;
LEV = long-term debt to assets ratio;
MB = Market-to-book ratio;
CAPINT = gross property plant and equipment divided by sales.
ISSUE = 1 if the annual change in equity is greater than 15% and 0 otherwise;
LOSS = 1 if net income is negative and 0 otherwise;
HINDEX =
∑
n 2
j =1
s j where Sj is market share of firm j based on sales in industry k;
REGIND = 1 if client j is in operating in regulated industry. Following Francis et al. (1999), regulated industries are defined as the following SIC codes:
railroad (4011 and 4100), trucking (4210 and 4213), airlines (4512, 4513, 4522, and 4581), telephone communications (4812 and 4813),
electric companies (4911), gas companies (492, 4923, and 4924), personal credit (6141), and insurance (6311);
LAW_ENF = mean score of three legal enforcement variables reported in La Porta et al. (1998), and used in Leuz et al. (2003). The three variables are (1)
the mean for 1980-1983 of a variable provided by Business International Corp., capturing the efficiency and integrity of the judicial system;
(2) the mean for 1982-1995 of a rule of law variable obtained from International Country Risk; and (3) the mean for 1982-1995 of a
corruption variable that assesses the corruption in government, obtained from International Country Risk. The law enforcement index values
range from zero to ten, with higher scores for greater law enforcement;
VOTING = Voting rights index which indicates how easy it is for shareholders to exercise their voting rights. This index ranges from 0 to 5, and is
constructed by La Porta et al. (1998). It aggregates the following components of shareholder rights: (1) the ability to vote by mail, (2) the
ability to gain control of shares during the shareholders’ meeting, (3) the possibility of cumulative voting for directors, (4) the ease of
calling an extraordinary shareholders meeting, and (5) the availability of mechanism allowing minority shareholders to make legal claims
against the directors. This index ranges from zero to five, with higher scores indicating greater protection of shareholders;
LEGAL = principal component extracted from LAW_ENF, and VOTING via factor analysis;
DISC = disclosure level from Saudagaran and Diga (1997, Table 2). The original source is the Center for International Financial Analysis and
Research (CIFAR 1995). The higher the number, the higher is the quality of disclosure;
FIN_TAX = an indicator variable. It equals 1 if tax accounting and financial reporting diverge and 0 otherwise. This index is constructed by and Hung
(2000);
ACCTG = principal component extracted from DISC, and FIN_TAX via factor analysis;
GDP = mean real GDP Per Capita (in US$ of year 2000 buying power), over the period 1993-2005. LGDP is the natural logarithm of GDP;
SMDEV = mean values of stock market development measured by stock market capitalization divided by GDP, over the period 1993-2005.
ECON = principal component extracted from LGDP, and SMDEV via factor analysis.

42
 TABLE 3
Correlation Matrices

SPEC LSALE LEV MB CAPINT ISSUE LOSS HINDEX REGIND LAW_ENF VOTING LEGAL DISC FIN_TAX ACCTG LGDP SMDEV ECON

SPEC 1.00
LSALE 0.16* 1.00
LEV 0.07* 0.29* 1.00
MB 0.00 -0.04* -0.07* 1.00
CAPINT 0.01* -0.03* 0.26* -0.16* 1.00
ISSUE -0.01 0.00 -0.02* 0.18* -0.10* 1.00
LOSS -0.04* -0.25* -0.05* -0.10* 0.05* -0.16* 1.00
HINDEX 0.03* 0.13* 0.08* 0.01 0.12* 0.01 0.01 1.00
REGIND 0.05* 0.12* 0.12* 0.01 0.17* -0.01 -0.02* 0.17* 1.00
LAW_ENF 0.05* -0.12* 0.13* 0.09* -0.11* 0.04* 0.01 0.15* -0.01 1.00
VOTING 0.02* -0.27* 0.00 0.09* 0.01 0.04* 0.00 -0.32* 0.01 0.16* 1.00
LEGAL 0.03* -0.32* 0.05* 0.10* -0.02* 0.05* 0.02* -0.22* 0.00 0.47* 0.92* 1.00
DISC 0.00 -0.32* -0.02* 0.08* -0.15* -0.01 -0.01 -0.29* -0.04* 0.34* 0.62* 0.63 1.00
FIN_TAX 0.04* -0.25* 0.01 0.06* 0.05* 0.03* -0.01 -0.21* -0.01 0.16* 0.73* 0.68 0.36* 1.00
ACCTG 0.04* -0.01 0.02* -0.01 0.17* 0.03* 0.02* 0.19* 0.04* 0.04* 0.04* 0.07* 0.58* 0.54* 1.00
LGDP 0.03* -0.12* 0.06* 0.06* -0.19* 0.00 0.05* 0.10* 0.01 0.65* 0.09* 0.29 0.39* -0.06* -0.30 1.00
SMDEV -0.02* -0.26* -0.08* 0.03* -0.07* 0.00 0.00 -0.25* -0.01 0.05* 0.45* 0.37 0.60* 0.36* -0.24 0.23* 1.00
ECON -0.03* -0.30* -0.09* 0.00 -0.01 -0.04* -0.01 -0.29* -0.02* 0.05* 0.43* 0.35* 0.59* 0.48* -0.15* 0.13* 0.81* 1.00

Notes:

See Table 2 for variable definitions. LSALE is the natural logarithm of SALE. LGDP is the natural logarithm of GDP. The full sample for the correlation
coefficients is 39,053 client firm-years for 29 countries over the period 1993-2005.

* Significance level of 0.01 (two-tailed).

43
 TABLE 4
Logistic Regression Results Explaining Choice of Big N Industry Audit Specialists:
Dependent variable: SPEC
Exp. Model Model Model Model Model Model Model Model Model Model
Sign 1 2 3 4 5 6 7 8 9 10
-2.666 -1.766 -1.655 -2.785 -1.713 -1.230 -3.444 -1.434 -1.447 -1.524
Intercept
(117.17)*** (151.72)*** (283.46)*** (45.23)*** (203.47)*** (162.62)*** (80.05)*** (150.48)*** (209.09)*** (242.84)***
LSALE + 0.171 0.164 0.181 0.171 0.195 0.171 0.160 0.151 0.158 0.193
(153.38)*** (172.69)*** (214.72)*** (230.68)*** (243.40)*** (157.87)*** (138.19)*** (129.63)*** (144.11)*** (251.58)***
LEV + 0.156 0.274 0.132 0.271 -0.015 0.104 0.213 0.391 0.385 0.036
(1.61) (5.35)** (1.17) (6.08)*** (0.01) (0.57) (3.39)* (11.36)*** (11.24)*** (0.06)
MB + 0.005 0.004 0.005 0.005 0.003 0.003 0.005 0.004 0.005 0.003
(27.64)*** (18.12)*** (26.36)*** (26.32)*** (10.39)*** (10.39)*** (22.23)*** (16.97)*** (22.20)*** (12.94)***
CAPINT + 0.012 0.009 0.011 0.011 0.013 0.011 0.012 0.010 0.010 0.012
(17.17)*** (11.23)*** (14.38)*** (17.50)*** (15.82)*** (13.75)*** (16.44)*** (12.38)*** (14.03)*** (16.38)***
ISSUE + -0.038 -0.044 -0.049 -0.031 -0.068 -0.065 -0.035 -0.034 -0.025 -0.062
(1.71) (2.41) (2.97)* (1.11) (5.22)** (4.59)** (1.42) (1.32) (0.72) (4.47)**
LOSS ? 0.032 0.029 0.046 0.047 0.123 0.089 0.020 0.015 0.026 0.120
(0.70) (0.61) (1.61) (1.75) (10.20)*** (4.94)** (0.28) (0.16) (0.46) (9.68)***
HINDEX - -0.091 0.221 0.197 0.112 -0.163 -0.345 -0.122 0.054 0.101 -0.225
(0.25) (1.37) (1.16) (0.37) (0.77) (3.51)* (0.43) (0.08) (0.29) (1.48)
REGIND + 0.218 0.179 0.178 0.201 0.116 0.118 0.213 0.194 0.194 0.091
(9.99)*** (6.86)*** (6.55)*** (8.46)*** (2.57)* (2.81)* (9.62)*** (7.96)*** (8.11)*** (1.64)
LAW_ENF + 0.142
(35.37)***
VOTING + 0.095
(13.56)***
LEGAL + 0.230 0.079
(39.60)*** (2.79)*
DISC + 0.017
(11.29)***
FIN_TAX + 0.404
(22.24)***
ACCTG + 0.175 0.204
(11.38)*** (15.50)***
LGDP + 0.216
(31.71)***
SMDEV + 0.091
(2.67)*
ECON + 0.113 0.164
(5.44)** (6.32)***
χ2 stat 355.86*** 355.73*** 444.40*** 436.33*** 496.26*** 338.91*** 330.73*** 285.28*** 581.16*** 542.55***
N 39,053 39,053 39,053 38,871 32,558 32,558 39,053 39,053 32,558 32,558

44
 TABLE 4 (continued)
Notes:

SPECjk equals ‘one’ if client j, headquartered in nation l purchases audits from the auditor having at least 20%/24%/30% market share (for the period prior to
1998, 1998-2001, and after 2001 respectively) in industry k in the national market. SPECjk is defined as ‘zero’ otherwise. See Table 2 for other variable
definitions. LSALE is the natural logarithm of SALE. LGDP is the natural logarithm of GDP.

We report the Wald statistic in the parenthesis. */**/*** Significance level of 0.1, 0.05 or 0.01 (two-tailed) respectively.

45
 TABLE 5
Logistic Regression Results Explaining Choice of Big N Industry Audit Specialists:
Dependent variable: SPEC1
Exp. Model Model Model Model Model Model Model Model Model Model
Sign 1 2 3 4 5 6 7 8 9 10
-3.011 -2.230 -2.167 -2.681 -2.244 -1.838 -3.853 -2.062 -2.059 -2.086
Intercept
(141.90)*** (225.30)*** (370.46)*** (59.25)*** (267.90)*** (282.85)*** (80.79)*** (362.71)*** (401.59)*** (346.50)***
LSALE + 0.146 0.138 0.152 0.143 0.159 0.139 0.138 0.132 0.139 0.157
(134.79)*** (152.37)*** (173.70)*** (157.18)*** (157.64)*** (105.51)*** (137.21)*** (146.57)*** (161.74)*** (163.85)***
LEV + 0.203 0.311 0.194 0.293 0.049 0.149 0.239 0.401 0.392 0.097
(2.36) (6.08)*** (2.24) (5.88)*** (0.12) (1.05) (3.82)** (10.11)*** (9.89)*** (0.43)
MB + 0.004 0.003 0.004 0.004 0.003 0.001 0.004 0.003 0.004 0.003
(25.22)*** (15.58)*** (22.77)*** (20.53)*** (9.74)*** (2.85)* (21.25)*** (16.30)*** (20.90)*** (8.49)***
CAPINT + 0.010 0.008 0.009 0.009 0.009 0.008 0.010 0.008 0.009 0.009
(14.95)*** (9.30)*** (11.77)*** (12.68)*** (11.16)*** (8.62)*** (14.91)*** (10.85)*** (12.19)*** (10.59)***
ISSUE + -0.017 -0.022 -0.025 -0.009 -0.024 -0.023 -0.015 -0.015 -0.005 -0.019
(0.39) (0.66) (0.85) (0.10) (0.65) (0.55) (0.28) (0.30) (0.03) (0.39)
LOSS ? 0.057 0.052 0.066 0.069 0.139 0.112 0.049 0.046 0.056 0.450
(2.15) (1.75) (2.99)* (3.36)* (12.74)*** (7.57)*** (1.49) (1.37) (2.06) (5.72)**
HINDEX - 0.429 0.654 0.642 0.573 0.504 0.369 0.396 0.554 0.598 0.450369
(5.93)** (13.49)*** (13.57)*** (10.95)*** (7.27)*** (3.86)** (4.95)** (10.31)*** (11.85)*** (5.72)**
REGIND + 0.142 0.114 0.111 0.134 0.073 0.073 0.140 0.117 0.119 0.048
(4.91)** (3.23)* (2.99)* (4.49)** (1.13) (1.12) (4.75)** (3.55)* (3.56)* (0.51)
LAW_ENF + 0.116
(23.72)***
VOTING + 0.069
(8.06)***
LEGAL + 0.178 0.062
(22.99)*** (2.94)*
DISC + 0.009
(4.32)**
FIN_TAX + 0.339
(15.54)***
ACCTG + 0.150 0.179
(10.06)*** (12.59)***
LGDP + 0.197
(21.43)***
SMDEV + 0.111
(5.14)**
ECON + 0.124 0.145
(8.08)*** (6.97)***
χ2 stat 236.47*** 242.92*** 266.91*** 237.30*** 291.74*** 234.98*** 237.17*** 242.52*** 268.75*** 329.61***
N 39,053 39,053 39,053 38,871 32,558 32,558 39,053 39,053 39,053 32,558

46
Notes:

SPEC1jk equals ‘one’ if client j, headquartered in nation l, purchases audits from the auditor having the largest value of ADTR_MS in industry k in nation l.
SPEC1jk is defined as ‘zero’ otherwise. See Table 2 for definitions of other variables. LSALE is the natural logarithm of SALE. LGDP is the natural
logarithm of GDP.

We report the Wald statistic in the parenthesis. */**/*** Significance level of 0.1, 0.05 or 0.01 (two-tailed) respectively.

47
 TABLE 6
Logistic Regression Results Explaining Choice of Big N Industry Audit Specialists:
Dependent variable: SPEC (Removing clients cross-listed in the U.S.)
Exp. Model Model Model Model Model Model Model Model Model Model
Sign 1 2 3 4 5 6 7 8 9 10
-2.538 -1.651 -1.543 -2.611 -1.596 -1.079 -3.301 -1.297 -1.307 -1.376
Intercept
(95.91)*** (122.74)*** (216.55)*** (39.42)*** (166.99)*** (119.28)*** (70.47)*** (116.03)*** (161.52)*** (184.27)***
LSALE + 0.156 0.147 0.166 0.156 0.178 0.152 0.145 0.134 0.140 0.173
(115.34)*** (125.43)*** (160.35)*** (170.08)*** (181.08)*** (113.64)*** (100.49)*** (93.18)*** (103.99)*** (180.56)***
LEV + 0.149 0.272 0.121 0.278 -0.003 0.116 0.212 0.394 0.389 0.057
(1.39) (4.94)** (0.91) (6.13)*** (0.01) (0.67) (3.22)* (11.10)*** (11.06)*** (0.15)
MB + 0.005 0.004 0.005 0.005 0.003 0.001 0.005 0.004 0.004 0.003
(19.71)*** (13.00)*** (19.23)*** (19.25)*** (6.55)*** (1.59) (15.90)*** (11.86)*** (15.50)*** (7.30)***
CAPINT + 0.011 0.008 0.009 0.010 0.011 0.010 0.011 0.008 0.009 0.011
(14.36)*** (8.65)*** (11.71)*** (14.33)*** (13.03)*** (10.51)*** (13.64)*** (9.57)*** (10.83)*** (13.13)***
ISSUE + -0.045 -0.051 -0.056 -0.037 -0.071 -0.068 -0.040 -0.039 -0.030 -0.065
(2.24) (2.98)* (3.79)** (1.47) (5.34)** (4.70)** (1.83) (1.62) (1.05) (4.53)**
LOSS ? 0.010 0.008 0.025 0.026 0.104 0.067 -0.001 -0.007 0.002 0.099
(0.07) (0.04) (0.49) (0.53) (7.51)*** (2.82)* (0.01) (0.04) (0.00) (6.73)***
HINDEX - -0.273 0.044 0.025 -0.066 -0.359 -0.575 -0.308 -0.137 -0.088 -0.439
(2.20) (0.05) (0.02) (0.12) (3.90)** (10.12)*** (2.73)* (0.51) (0.22) (5.94)**
REGIND + 0.165 0.129 0.129 0.155 0.065 0.058 0.165 0.146 0.146 0.038
(5.19)** (3.28)* (3.09)* (4.60)** (0.72) (0.61) (5.17)** (4.13)** (4.20)** (0.26)
LAW_ENF + 0.141
(32.31)***
VOTING + 0.097
(13.61)***
LEGAL + 0.234 0.073
(38.16)*** (3.36)*
DISC + 0.017
(10.49)***
FIN_TAX + 0.418
(23.44)***
ACCTG + 0.196 0.222
(14.13)*** (18.04)***
LGDP + 0.214
(30.45)***
SMDEV + 0.083
(2.20)
ECON + 0.100 0.167
(4.19)** (6.45)***
χ2 stat 258.29*** 249.07*** 313.99*** 309.45*** 362.25*** 255.89*** 227.52*** 199.77*** 227.94*** 385.42***
N 36,934 36,934 36,934 36,752 30,604 30,604 36,934 36,934 36,934 30,604

48
Notes:

We delete 2,119 client firm-years, relating to those clients that cross-list their shares in the U.S. market, since the firms’ auditor choice may be related to their
overseas equity issues. See Table 2 for definitions of the variables. LSALE is the natural logarithm of SALE. LGDP is the natural logarithm of GDP.

We report the Wald statistic in the parenthesis. */**/*** Significance level of 0.1, 0.05 or 0.01 (two-tailed) respectively.

49
 TABLE 7
Logistic Regression Results Explaining Choice of Big N Industry Audit Specialists:
Full Sample versus Developing Countries

Full Sample Developing Countries

Exp. Sign Model Model Model Model
1 2 3 4
DV=SPEC DV=SPEC1 DV=SPEC DV=SPEC1
-1.289 -1.886 -1.876 -2.339
Intercept
(149.18)*** (303.87)*** (12.05)*** (69.20)***
LSALE + 0.146 0.126 0.155 0.155
(109.15)*** (107.67)*** (36.21)*** (35.41)***
LEV + 0.380 0.388 0.630 0.896
(10.80)*** (9.41)*** (6.07)*** (9.36)***
MB + 0.004 0.003 0.007 0.004
(14.27)*** (12.74)*** (45.35)*** (20.65)***

CAPINT + 0.009 0.008 0.013 0.017

(11.61)*** (9.60)*** (3.75)** (6.15)***
ISSUE + -0.034 -0.015 0.066 -0.031
(1.33) (0.30) (1.50) (0.29)
LOSS ? 0.008 0.037 -0.207 -0.237
(0.04) (0.85) (5.88)** (5.11)**
HINDEX - 0.020 0.519 0.872 0.389
(0.01) (8.53)*** (1.65) (0.50)
REGIND + 0.209 0.136 0.556 0.494
(9.58)*** (4.61)** (6.56)*** (8.54)***

256.41*** 236.46*** 163.58*** 113.53***

χ2 stat
N 39,053 39,053 7,511 7,511

Notes:

See Tables 2 and 5 for variable definitions. In models 3 and 4 the sample only includes firms in the developing countries. Countries are considered as
developing if the country’s equity market is not included in the Morgan Stanley Capital International database (Hail and Leuz [2006] Table 1). Based on this
definition, 12 countries are included in models 3 and 4, namely, Brazil, Chile, Greece, India, Indonesia, Israel, Malaysia, Mexico, Philippines, South Africa,
Taiwan, and Thailand. Models 1 and 2 provide results using the full sample.

We report the Wald statistic in the parenthesis. */**/*** Significance level of 0.1, 0.05 or 0.01 (two-tailed) respectively.

50