ArticlePDF AvailableLiterature Review

Geographic Proximity between Auditor and Client: How Does It Impact Audit Quality?

January 2012
Auditing A Journal of Practice & Theory 31(2)

Authors:

City University of Hong Kong

Using a large sample of audit client firms, this paper investigates whether and how the geographic proximity between auditor and client affects audit quality proxied by accrual-based earnings quality. We define an auditor as a local auditor (1) if the auditor's practicing office is located in the same metropolitan statistical area (MSA) as the client's headquarters, and (2) if the geographic distance between the two cities where the auditor's practicing office and the client's headquarters are located is within 100 kilometers, or they are in the same MSA. As predicted, our empirical results are consistent with local auditors providing higher-quality audit services than non-local auditors. In addition, as predicted, this quality difference is weakened for diversified clients with more operating or geographic segments. The results are robust to a variety of sensitivity checks. Overall, our evidence suggests that informational advantages associated with local audits enable auditors to better constrain management's biased earnings reporting, with greater advantages for less diversified clients. JEL Classification: M42.

Content uploaded by Jeong-Bon Kim

Content may be subject to copyright.

Electronic copy available at: http://ssrn.com/abstract=1982959

Geographic Proximity between Auditor and Client:

How Does It Impact Audit Quality?

Jong-Hag Choi, Jeong-Bon Kim, Annie A. Qiu, and Yoonseok Zang

Forthcoming in

AJPT, May 2012

________________

*Jong-Hag Choi is an Associate Professor at Seoul National University (acchoi@snu.ac.kr).

Jeong-Bon Kim is a Chair Professor at City University of Hong Kong (jeongkim@cityu.edu.hk).

Annie Qiu is the Head of International Business Department at Citic Securities, Beijing

(qan@citics.com). Yoonseok Zang is an Associate Professor at Singapore Management University

(yszang@smu.edu.sg). We appreciate the helpful comments of Charles Chen, Sunhwa Choi, Shifei

Chung, Chris Hogan (editor), Lee-Seok Hwang, Akira Kanie, Bin Ke, Soo Young Kwon, Jay

Junghun Lee, Akihiro Noguchi, Jaime Schmidt, Haina Shi, Byron Song, Joon Yong Shin, Xijia Su,

Tracey Zhang, two anonymous reviewers, and seminar participants at City University of Hong

Kong, Nagoya University, Seoul National University, Korea Advanced Institute of Science and

Technology, Korea University, The Hong Kong Polytechnic University, and 2007 AAA Annual

Meeting. Choi (Kim) acknowledge financial support from Samil PricewaterhouseCoopers Korea

(the Social Sciences and Humanities Council of Canada via the Canada Research Chair program).

Zang thanks School of Accountancy Research Centre (SOAR) at SMU for financial assistance.

Correspondence: Yoonseok Zang, School of Accountancy, Singapore Management University

(SMU), 60 Stamford Road, Singapore 178900. Tel: +65-6828-0601. Fax: +65-6828-0600. E-mail:

yszang@smu.edu.sg

Electronic copy available at: http://ssrn.com/abstract=1982959

Geographic Proximity between Auditor and Client:

How Does It Impact Audit Quality?

SUMMARY: Using a large sample of audit client firms, this paper investigates whether

and how the geographic proximity between auditor and client affects audit quality proxied

by accrual-based earnings quality. We define an auditor as a local auditor (1) if the

auditor’s practicing office is located in the same metropolitan statistical area (MSA) as the

client's headquarters and (2) if the geographic distance between the two cities where the

auditor’s practicing office and the client’s headquarters are located is within 100

kilometers or they are in the same MSA. As predicted, our empirical results are consistent

with local auditors providing higher-quality audit services than non-local auditors. In

addition, as predicted, this quality difference is weakened for diversified clients with more

operating or geographic segments. The results are robust to a variety of sensitivity checks.

Overall, our evidence suggests that informational advantages associated with local audits

enable auditors to better constrain management’s biased earnings reporting, with greater

advantages for less diversified clients.

JEL classification: M42

Keywords: Auditor locality, geographic proximity, audit quality, diversification.

INTRODUCTION

Since the Enron debacle and the subsequent collapse of Arthur Andersen,

regulators, lawmakers, academic researchers, and the popular press have paid considerable

attention to engagement-specific factors determining the auditor–client relationship and

their impact on audit quality. The focus of this study is on a new engagement-specific

factor that may play an important role in the development of the auditor–client relationship:

geographic proximity between auditor and client, or auditor locality. Specifically, we

examine whether the geographic distance between auditor and client plays a role in

determining audit quality.

This study is motivated by a growing body of literature in accounting and finance

that documents the importance of geographic proximity between economic agents (e.g.,

DeFond et al. 2011; Kedia and Rajgopal 2011; Malloy 2005). This strand of research

suggests that geographic proximity lowers the information asymmetry between economic

agents by facilitating information flows and monitoring. Building upon the implication

from this recent literature, we expect that local auditors possess an informational

advantage over non-local auditors because proximity to clients can facilitate the

acquisition of more idiosyncratic client information, such as client-specific incentives,

means, and opportunities for substandard reporting. We posit that this informational

advantage attenuates managerial opportunism in financial reporting because greater client-

specific knowledge enables auditors to better identify and reign in aggressive reporting

practices (Krishnan 2003; Myers et al. 2003; Reichelt and Wang 2010). We expect,

however, that this effect of auditor locality is weakened for diversified clients with more

operating divisions or geographic segments, because the audit advantage associated with

geographic proximity to client headquarters is likely to be smaller for such clients.

Informational advantages arising from geographic proximity are well documented

in the contexts of portfolio decisions and investment performance (Baik et al. 2010;

Bodnaruk 2009; Ivkovich and Weisbenner 2005), analysts’ forecasting decisions (Malloy

2005), knowledge transfers (Audretsch and Feldman 1996; Audretsch and Stephan 1996),

and the monitoring and regulatory effectiveness of the U.S. Securities and Exchange

Commission (SEC) (DeFond et al. 2011; Kedia and Rajgopal 2011). If geographic

proximity facilitates information transfers and monitoring, then auditors located closer to

their clients should be better able to assess the clients’ incentives and abilities for

opportunistic earnings management. Such client-specific knowledge is vital for auditors to

plan audits effectively, to identify relevant audit risks, and to interpret audit evidence

properly, which in turn helps them rely less on management’s subjective estimates when

assessing accrual choices (Knechel et al. 2007).

As in many other studies (e.g., Chung and Kallapur 2003; Frankel et al. 2002), we

assert that biased earnings reporting can be used to draw inferences about audit quality,

and we use accrual-based earnings quality measures as proxies for audit quality. As our

main proxies for audit quality, we use two measures of discretionary accruals, which are

estimated by using (1) the Ball–Shivakumar (2006) model, in which the asymmetric

timeliness in recognition of gains versus losses is controlled for, and (2) the methodology

of Kothari et al. (2005), in which the performance of matching firms is adjusted. To

supplement these measures, we also use two additional measures of accrual quality:

(3) one developed by Dechow and Dichev (2002) and modified by McNichols (2002) and

(4) another developed by Francis, LaFond, Olsson, and Schipper (2005, FLOS hereafter),

which focuses on a discretionary component of accrual quality. We expect that local

auditors with more client-specific knowledge are better able to deter their clients from

engaging in aggressive accrual choices than non-local auditors. We therefore predict that

the clients of local auditors will exhibit lower levels of absolute discretionary accruals and

higher levels of accrual quality relative to the clients of non-local auditors.

Our empirical results using 12,439 firm–year observations collected over the years

2002–2005 reveal the following. First, after controlling for a comprehensive set of

variables known to affect the extent of opportunistic earnings management, we find that

the clients of local auditors report a lower level of absolute discretionary accruals and a

higher level of accrual quality compared to those of non-local auditors. Second, we find

that these associations are relatively weaker or disappear for diversified clients with more

operating and/or geographic segments. This evidence is consistent with the view that local

audit advantages are greater when information transfers between a firm and its

stakeholders, including auditors, occur mostly at the corporate headquarter level and

auditors perform their audit work primarily at headquarters. For more diversified firms, a

larger part of business operations are performed in multiple locations other than where the

firms are headquartered. As a result, an auditor’s informational advantage associated with

geographic proximity to headquarters is likely to diminish. Finally, our results are robust

to controlling for potential self-selection bias associated with local versus non-local

auditor choices and to restricting the sample only to the clients of audit offices engaging in

both local and non-local audits.

Our study contributes to the existing literature in the following ways. To our

knowledge, this is the first study that provides direct evidence that auditor–client

geographic proximity is an important engagement-specific factor influencing audit quality.

While the issue of geographic proximity between economic agents has been investigated

in various contexts, no previous study has examined the issue in the context of audit

quality, with the exception of DeFond et al. (2011). However, the focus of DeFond et al.

(2011) is not on auditor–client geographic proximity but, rather, on the geographic

proximity between local audit offices and SEC regional offices. Evidence provided in our

study fills this void and helps better understand the role of auditor–client geographic

proximity in the development of the auditor–client relationship.

Second, this study helps explain why large audit firms continue to expand their

practicing offices to many cities. The results of this study suggest that decentralized local

office structures reduce information asymmetries between clients and auditors, thereby

allowing office-level auditors to improve client-specific knowledge. Finally, findings in

this study will be of interest to regulators. Regulators have been concerned about whether

a close relationship or social bonding between auditor and client is detrimental to the

auditor’s objectivity. While some prior studies find that their cozy relationship impairs

audit quality (e.g., Davis et al. 2009; Menon and Williams 2004), the results of this study

suggest that proximity to clients, which likely increases social bonding, can be beneficial

to audit quality.

The next section reviews the extant literature and develops research hypotheses.

The third section discusses variable measurements and empirical models. The fourth

section describes our sample and presents descriptive statistics. The fifth section presents

our empirical results. The final section concludes the paper.

EXTANT LITERATURE AND HYPOTHESIS DEVELOPMENT

An audit firm typically provides audit services through a practicing office located

near its clients. As noted by Francis et al. (1999), it is the local engagement offices, not the

national headquarters of the audit firm, that ―contract for and oversee the delivery of

audits‖ and ―issue audit reports for the clients who are headquartered in the same

geographical locale.‖ In a related vein, former SEC commissioner Wallman (1996)

emphasizes that auditing research should pay more attention to city-level (or office-level)

analyses rather than national-level analyses because local practicing offices make the most

of audit decisions with respect to a particular client.

The main focus of prior office-level studies has been on the questions of

(1) whether auditor independence is impaired for the audits of large clients by individual

audit offices (e.g., Chung and Kallapur 2003; Craswell et al. 2002; Reynolds and Francis

2000), (2) whether auditor industry expertise is firm-wide or office specific (e.g., Ferguson

et al. 2003; Francis et al. 2005; Reichelt and Wang 2010), and (3) whether audit quality is

associated with the size of the audit engagement offices (e.g., Choi et al. 2010; Francis and

Yu 2009, 2011). However, prior literature has devoted little attention to the role of

auditor–client geographic proximity in determining audit quality.

Research in financial economics provides evidence suggesting that geographic

proximity between economic agents matters in explaining their decision-making behavior

and contractual relationships. A growing body of the ―home or local bias‖ literature in

finance finds that equity investors overweight domestic (or local) stocks in their portfolio

choices, primarily because they are more familiar with domestic (or local) stocks and have

advantages in obtaining information (Coval and Moskowitz 1999; Covrig et al. 2006;

Ivkovich and Weisbenner 2005). This informational advantage also enables local

individual and institutional investors to better monitor firms (Baik et al. 2010; Peterson

and Rajan 2002) and to earn superior returns than non-local investors (Bodnaruk 2009;

Coval and Moskowitz 2001; Ivkovich and Weisbenner 2005). Furthermore, Malloy (2005)

reports that geographically proximate analysts provide more accurate earnings forecasts

than other analysts, suggesting that the former have an informational advantage over the

latter.

A few recent studies in accounting and auditing also examine issues related to

geographic proximity between economic agents. Kedia and Rajgopal (2011) find that

firms located closer to SEC regional offices are less likely to restate their prior years’

financial statements. The results of this study suggest that management’s assessment of ex

ante misreporting costs is higher for firms located nearer to SEC regional offices because

geographic proximity lowers information asymmetry and facilitates monitoring. DeFond

et al. (2011) document that non-Big 4 audit offices located farther from SEC regional

offices are less likely to issue going concern audit opinions, suggesting that non-Big 4

auditors’ incentives to be independent are influenced by their geographic proximity to

SEC regional offices. In sum, these studies indicate that geographic proximity mitigates

information asymmetries and enhances monitoring effectiveness.

In a similar vein, we argue that geographic proximity or auditor locality is

associated with audit quality, because informational advantages arising from the proximity

help auditors develop knowledge about client-specific characteristics, such as client

incentives, abilities, and opportunities for opportunistic earnings management, and about

client business risk that entails audit risks. Local auditors can develop such knowledge

through various ways. For example, they can more easily obtain valuable private

information about a client firm through informal talks with its executives, employees,

suppliers, customers, and competitors.1 Local auditors can more frequently visit client

firms and observe what goes on there directly at a lower cost. They are able to learn more

about client-specific news from local media such as newspapers, radios, and TV stations.

As local community constituents, they are more familiar with local regulations, business

practices, and market conditions. Local auditors have natural opportunities to establish

1 In terms of communicating with clients and others, we believe that geographic proximity will improve

communication and information quality because it facilitates more face-to-face communication. Prior studies

in psychology, communication, information systems, and organizational behavior suggest that face-to-face

communication is more effective through the support for a higher level of interaction than other electronic

forms of communication, such as e-mail and videoconferencing (e.g., Baltes et al. 2002; Doherty-Sneddon et

al. 1997; Hambley et al. 2007; Hinds and Morensen 2005). Local auditors have clear advantages in face-to-

face communications with their clients and other stakeholders, compared with non-local auditors.

personal ties and social interactions with the executives of client firms. These

opportunities are an important mechanism for information exchange (Hong et al. 2004),

helping auditors to better evaluate their clients’ characteristics and incentives.

Obtaining client-specific knowledge such as internal control structure and

opportunities for substandard reporting is vital for auditors to plan audits effectively, to

identify relevant audit risks, and to interpret audit evidence properly (Knechel et al.

2007).2 For example, Bedard and Johnstone (2004) show that auditors adjust their audit

planning when auditors perceive that a client has an increased risk of earnings

management. Johnson et al. (2002) argue that adequate client-specific knowledge

facilitates the detection of material errors and allows auditors to rely less on management

estimates. Consistent with this argument, survey evidence of Carcello et al. (1992)

suggests that knowledgeable audit teams, frequent communications between auditors and

management, and frequent visits by the audit engagement partner and senior manager to

an audit site are among the 10 highest rated attributes of audit quality. Moreover, prior

studies suggest that auditors who gain client-specific knowledge through their extended

tenure with specific clients (Myers et al. 2003) or industry specialization (Reichelt and

Wang 2010) are better able to mitigate clients’ aggressive accrual choices.

Based on the above arguments, we expect that auditor–client proximity helps

auditors develop better knowledge about client-specific incentives, abilities, and

opportunities for substandard reporting, leading to local auditors being more effective in

2 For instance, superior knowledge of a client’s suppliers, customers, and employees will help auditors to

verify management’s subjective attestation of accruals in accounts payable, accounts receivable, and payroll

liabilities. Similarly, a better understanding of a client’s future strategic plans and the managers’

personalities and compensation schemes will help evaluate the risk of earnings management.

monitoring client reporting behavior and constraining biased financial reporting. Thus our

first hypothesis (in alternative form) is as follows.3

H1: Audits performed by local auditors are of higher quality than audits

performed by non-local auditors, other things being equal.

We next posit that the positive association between auditor locality and audit

quality is weaker for clients with more diversified structure. We use the corporate

headquarters location as the client firm location. Corporate headquarters is likely to be the

center of information exchange between the firm and its suppliers, service providers, and

investors (Coval and Moskowitz 1999); thus auditors perform a substantial amount of

audit work there. However, if a client firm has many operating divisions or geographic

segments outside its headquarters, the local audit advantages discussed above will be

attenuated. It will be difficult for auditors to visit all the divisions or segment offices of

diversified firms scattered throughout the U.S. or other countries and to maintain close

face-to-face communications with executives and other employees. As a result, we expect

the relation between auditor locality and audit quality to be stronger for less diversified

client firms because corporate headquarters generate relatively richer information for

audits for such firms. To provide empirical evidence of this prediction, we test the

following hypothesis (in alternative form).

H2: The positive relation between auditor locality and audit quality is weaker for

diversified clients with more operating divisions or geographic segments, other

things being equal.

3 In contrast to this view, the works of Gul et al. (2006) and Wang et al. (2008) find that audit quality in

China is, in general, lower for local than for non-local audits, because local auditors are subject to greater

political influences of local governments, who are often the controlling shareholders of both local client

firms and auditors. However, we do not expect such results to be applicable to the U.S. setting, where a high

level of auditor independence is required and the ownership of local governments is minimal. Thus we do

not formally develop this view as a competing hypothesis.

MEASUREMENT OF VARIABLES AND MODEL SPECIFICATION

Definition of Local Auditors

We use the location of the audit engagement office as the auditor location because

it is an office-based engagement partner or audit team, and not the national headquarters,

that actually administers individual audit contracts and issues audit opinions (Ferguson et

al. 2003; Francis and Yu 2009, 2011). Following prior studies (Coval and Moskowitz

1999; Francis et al. 2005; Pirinsky and Wang 2006), we use the location of corporate

headquarters as the client firm location. We differentiate local auditors from non-local

auditors in the following two ways. We first define an auditor as a local auditor if the

audit engagement office is located in the same metropolitan statistical area (MSA) where

audit clients are headquartered (DMSA = 1), and as a non-local auditor otherwise (DMSA

= 0).

We adopt this MSA-based differentiation because an MSA can be viewed as a

reasonable geographic boundary in which most social and economic interactions between

community members take place. However, this MSA-level differentiation does not take

into account the actual geographic distance between auditor and client. Some MSAs are

located within a narrow geographic boundary, particularly in the eastern U.S., while

several MSAs in some western states are much larger than those in the eastern states. As a

result, some client headquarters can be more proximate to audit offices in adjacent MSAs

than those in the same MSA. For this reason, we also adopt an alternative approach in

which an auditor is defined as a local auditor if the audit engagement office is located

within 100 kilometers from the client’s headquarters or if both the audit office and client

headquarters are located in the same MSA (D100 = 1), and as a non-local auditor

otherwise (D100 = 0).4 We choose 100 kilometers as the cut-off value following Coval

and Moskovitz (2001), Kedia and Rajgopal (2011), and Malloy (2005). Similar to these

studies, we consider any distance within 100 kilometers to be a reasonable commute to

work.5

Measurement of Audit Quality

Prior studies often use either accruals or audit opinions to proxy for audit quality.

This study uses accruals rather than audit opinions to proxy for audit quality for the

following reasons: Audit opinion is an extreme measure of audit quality because modified

opinions comprise only a small proportion of audit opinions, and thus, unlike accrual-

based measures, audit opinions do not address audit quality differentiation for a broad

cross section of firms (Myers et al. 2003).6 As a result, there is little cross-sectional

variation in audit opinions, which can lead to a lack of statistical power in empirical tests.

4 We use the indicator variables (DMSA and D100) rather than a continuous variable to capture the distance

between the audit office and the client’s headquarters for two reasons: First, since we compute the distance

between the centers of the two cities where the auditor office and client headquarters are located, due to data

limitations (see footnote 14 for more details), analyses with a continuous variable will introduce more noise

into our tests, particularly when the auditor office and client headquarters are located in the same city.

Second, the effect of geographic proximity may not differ significantly within a distance short enough to

commute regularly. Consistent with our expectations, our untabulated results show that the continuous

variable of distance is significant only when the distance is over and beyond a certain threshold. In addition,

the models using the indicator variable demonstrate higher explanatory power in our empirical analyses than

those using the continuous variable of distance.

5 The validity of the D100 measure can be harmed by variations in population density, traffic conditions,

availability of public transportation, and its definition of combined factors. We performed several sensitivity

checks to mitigate the concern. First, we used 150, 200, and 250 kilometers to determine a combined

variable for MSA-based and distance-based measures. Second, we used pure distance-based indicator

variables. Third, we used state-based differentiation, as in Baik et al. (2010). The results of these sensitivity

analyses are qualitatively similar to the tabulated results with respect to the variable of interests in most

cases. Thus, the inferences remain unchanged. Footnote 23 explains exceptional cases when the results of

the alternative proxies are different from the tabulated results where appropriate. In general, we find that the

use of an MSA-based measure yields more significant results than the use of distance-based measures, which

implies that an MSA represents a more reasonable economic boundary that permits variations in locality

characteristics.

6 In addition, auditors are less likely to compromise their audit opinions due to their concerns over legal

liability, while they are more likely to tolerate biased reporting of earnings because failures to detect biased

accruals have less of an impact on an auditor’s legal liability compared with failures to issue appropriate

audit opinions (DeFond et al. 2002). This may be a reason why DeFond et al. (2011) fail to find any

significant relation between auditor–client distance and the auditor’s propensity to issue a going concern

audit opinion in their untabulated analyses. We tried to replicate this test and found the same result.

As in many other studies, we view absolute discretionary (abnormal) accruals (DA)

as an outcome of opportunistic earnings management. It is well known that the traditional

DA measure using the Jones (1991) model is noisy (Dechow et al. 1995). To alleviate this

concern, we use two alternative measures of DA: One is obtained from the augmented

Jones model of Ball and Shivakumar (2006), which takes into account the asymmetric

timeliness in gain versus loss recognition, and the other is estimated by the performance-

matched modified Jones model (Kothari et al. 2005). We denote these two measures as

DA1 and DA2, respectively.

The augmented Jones model of Ball and Shivakumar (2006) in Eq. (1) explains the

computation of our first measure, DA1:

jt jt-1 1 jt-1 2 jt jt-1 3 jt jt-1

4 jt jt-1 5 jt 6 jt-1 jt jt

ACCR / A = β[1 / A ] + β[ΔREV / A ]+ β[PPE / A ]

+β[CFO / A ] + βDCFO +β[(CFO / A )* DCFO ]+ε

(1)

where, for firm j and year t (or t - 1), ACCR denotes total accruals (income before

extraordinary items minus cash flow from operations); A, ΔREV, and PPE represent total

assets, changes in net sales, and gross property, plant, and equipment, respectively; CFO

represents cash flows from operations; DCFO is an indicator variable that equals one if

CFO is negative, and zero otherwise; and ε is the error term. We estimate Eq. (1) for each

two-digit Standard Industrial Classification (SIC) industry and year with at least 10

observations. Our first measure of abnormal accruals, DA1, is the difference between

actual total accruals deflated by lagged total assets and the fitted values of Eq. (1).

Our second measure of abnormal accruals, DA2, is computed as follows. For each

two-digit SIC industry and year with at least 10 observations, we estimate the cross-

sectional version of the modified Jones model as

jtjtjtjtjtjtjtjtjt APPEARECREVAAACCR



  ]/[]/)[(]/1[/ 1312111

(2)

where the residuals are DA before adjusting for firm performance. Following the

procedures proposed by Kothari et al. (2005), we match each firm–year observation with

another from the same two-digit SIC industry with the closest return on assets (ROA) in

each year. We then compute performance-matched abnormal accruals, DA2, by taking the

difference between the unadjusted DA and the ROA-matched firm’s DA.

To supplement two discretionary accrual measures, we also employ the accrual

quality measure developed by Dechow and Dichev (2002) and modified by McNichols

(2002). Dechow and Dichev (2002) view that accruals are of high quality if they map into

past, current, and future cash flows effectively. McNichols (2002) suggests that including

the original Jones model variables (i.e., ΔREV and PPE) in the Dechow–Dichev model

improves the performance of estimation. Thus, we obtain this accrual quality measure by

calculating the time series standard deviation of residuals from a cross-sectional regression

estimated from the following equation

1 2 1 3 3 1 4 5

jt jt jt jt jt jt jt

WCA CFO CFO CFO REV PPE

      



        

(3)

where, for firm j and year t (or t - 1), ∆WCA represents change in working capital.7 All

the variables above are deflated by average total assets. Equation (3) is estimated for each

two-digit SIC code industry with at least 10 observations in a given year. Then we

calculate the standard deviation of residuals υjt, termed AQ1 for each firm over the years t -

4 to t. The larger the standard deviation of residuals, the greater the noise in earnings and

the lower the quality of earnings. Hence AQ1 is considered an inverse measure of audit

quality.

7 Here ∆WCA is defined as -(∆AR + ∆INV + ∆AP + ∆TAX + ∆OTH), where ∆AR is change in receivables,

∆INV is change in inventory, ∆AP is change in payables, ∆TAX is change in tax payable, and ∆OTH is

change in other assets.

It is suggested by FLOS that poor accrual quality can be due to innate features of a

firm’s business model, such as its operating environment and complexity of transactions,

or it could be due to discretionary factors such as managerial accounting choices and

accrual manipulation. Since the innate factors of AQ1 are mainly driven by a client firm’s

operating environment, audit quality could be better proxied by the discretionary factors of

AQ1. For this reason, we decompose AQ1 into innate and discretionary components based

on the approach outlined in FLOS8 and use the discretionary component, termed AQ2, as

an alternative proxy for audit quality.

Empirical Model for Testing the Effect of Auditor Locality on Audit Quality

To test our hypotheses, we estimate the following regression model that links the

magnitude of absolute abnormal accruals with our variable of interest, that is, auditor

locality, and other control variables known to affect the extent of earnings management:

jt 0 1 jt 2 jt 3 jt jt 4 jt

5 jt 6 jt 7 jt 8 jt 9 jt

10 jt 11 jt 12 jt 13 jt 14 jt 15 jt

16 jt 17 jt

| DA| = α+αDLOCAL +αLNBGS +αDLOCAL * LNBGS +αLNTA

+αBIG4 +αTENURE +αNAS +αINDSPEC +αCHGSALE

+αBTM +αLOSS +αLEV +αZMIJ +αISSUE +αCFO

+αLAGACCR +αCONCENT +(Indu jt

stry&Year Indicators)+ε

(4)

where, for firm j in year t, all variables are as defined in Table 1. Absolute abnormal

accruals, denoted by |DA|, are our proxy for the extent of opportunistic earnings

management. Here DLOCAL proxies for auditor locality or auditor–client geographic

proximity, and this variable is either DMSA or D100. The variable LNBGS represents the

natural log of the sum of the number of business and geographic segments minus one. If

business or geographic segment data for a given observation are missing from Compustat,

we assign it a value of one, following Francis and Yu (2009). Thus, the minimum LNBGS

is zero (i.e., log(2 - 1) = 0). Since the α1 coefficient measures only the partial effect of

8 Specifically, we regress the firm-specific standard deviation of residuals υjt from Eq. (3) over the years t - 4

to t (AQ1) on firm size (market capitalization), cash flow volatility, sales volatility, operating cycle, and loss

proportion following Eq. (8) of FLOS. The predicted values from this estimation capture the innate

component while unexplained portions (the residuals) capture the discretionary component.

DLOCAL on |DA| when LNBGS is equal to zero, we also estimate Eq. (4) after excluding

DLOCAL*LNBGS from the model. A negative coefficient on DLOCAL in such estimation

indicates that clients of local auditors, on average, report a lower level of absolute

discretionary accruals than those of non-local auditors, consistent with our first hypothesis.

For further evidence, we also test the coefficient of DLOCAL at each quartile value of

LNBGS. Our second hypothesis predicts a positive coefficient on the interaction between

DLOCAL and LNBGS in Eq. (4), consistent with the local audit advantages being weaker

for more operationally or geographically diversified clients.

[INSERT TABLE 1 ABOUT HERE!]

Equation (4) includes many control variables that are known to affect the

magnitude of earnings management. We include LNTA to control for the client size effect

(e.g., Dechow and Dichev 2002). Several studies show that Big 4 auditors and industry

specialists are more effective than non-Big 4 auditors and non-specialists, respectively, in

constraining opportunistic earnings management (Becker et al. 1998; Krishnan 2003).9 We

include BIG4 and INDSPEC to control for the effect of auditor brand name at the national

level and industry expertise at the MSA level, respectively. We include TENURE because

Johnson et al. (2002) and Myers et al. (2003) provide evidence that clients of longer-

tenure auditors report lower abnormal accruals. We include NAS to control for the effect of

non-audit fees on audit quality (Ashbaugh et al. 2003; Chung and Kallapur 2003; Frankel

et al. 2002). The variables BTM and CHGSALE are included to control for firm growth,

while LOSS is included to control for potential differences in earnings management

9 In contrast, Louis (2005) suggests that non-Big 4 auditors may provide better-quality audit service to small

clients. We also add an indicator variable for the middle four second-tier auditors in both Eqs. (4) and (5).

Prior studies by Hogan and Martin (2009) and Boone et al. (2010) suggest that second-tier auditors behave

differently from top-tier auditors and other small auditors. Because we obtain qualitatively similar results

after including this indicator variable, we do not tabulate them for simplicity. Based on p ≤ 10% (two tailed)

as a cut-off for the significance level, this paper defines ―qualitatively similar‖ to mean that in Eqs. (4) and

(5), (1) the coefficient on DLOCAL is negative and significant and (2) the coefficient of DLOCAL*LNBGS is

positive and significant.

between loss and profit firms. We also include ZMIJ and ISSUE to control for the effects

of financial distress (Reynolds and Francis 2000) and financing transactions (Teoh et al.

1998) on earnings management, respectively. The variable LEV is included because highly

levered firms may have greater incentives for earnings management due to their concerns

of debt covenant violations (Becker et al. 1998; DeFond and Jiambalvo 1994). We include

CFO to control for the potential correlation between accruals and cash flows (Kothari et

al. 2005). As in Ashbaugh et al. (2003) and Kim et al. (2003), we include LAGACCR to

control for the reversal of accruals over time. The variable CONCENT is included to

control for the effect of auditor concentration on our results (Kedia and Rajgopal 2011;

Kallapur et al. 2010).10 Finally, we include industry and year indicators to control for

possible variations in accounting standards and regulations across industries and over

years. Each industry is defined based on two-digit SIC codes.

Next, when AQ is used as the proxy for audit quality, we estimate the following

regression model:

jt 0 1 jt 2 jt 3 jt jt 4 jt

5 jt 6 jt 7 jt 8 jt 9 jt

10 jt 11 jt 12 jt jt

AQ = β+βDLOCAL + βLNBGS + βDLOCAL * LNBGS + βLNTA

+βBIG4 + βOPCYCLE + βSTD_CFO + βSTD_SALE + βNAS

+βINDSPEC + βZMIJ + βCFO +(Industry&Year Indicators)+ε

(5)

where, for firm j in year t, all variables are as defined in Table 1. We add OPCYCLE,

STD_CFO, and STD_SALE to Eq. (5) because Dechow and Dichev (2002) argue that a

longer operating cycle and greater operating volatility are associated with higher accrual

estimation errors. We also add in Eq. (5) several other control variables (NAS, INDSPEC,

ZMIJ, and CFO) included in Eq. (4) that are deemed to be associated with accrual

10 When we measure CONCENT using audit fees rather than the number of clients, we obtain qualitatively

similar results.

quality.11 Our variables of interest, DLOCAL and DLOCAL*LNBGS, are the same as in

Eq. (4).

SAMPLE AND DESCRIPTIVE STATISTICS

Sample

The initial list of our sample consists of all firms included in the Audit Analytics

database for the four-year period 2002–2005. We extract data on the state/city locations of

auditors’ practicing offices and client firms’ headquarters from the Audit Analytics

database.12 Since an MSA consists of one or more counties, we match their city-level

locations to the county codes of Federal Information Processing Standards using the U.S.

Census Bureau’s 2000 Places file and identify whether or not the practicing office of the

auditor and the client’s headquarters are located in the same MSA. Following Francis et al.

(2005), we delete observations if auditors or clients are not located in one of the 280

MSAs defined in the U.S. census of 2000.13 Next, we obtain the latitude and longitude

data for the cities of the practicing offices of auditors and the headquarters of client firms

using the U.S. Census Bureau’s Gazetteer 2001 City–State file. With these data, we

compute the actual geographic distances between the (centers of) two cities where auditor

offices and client headquarters are located.14

11 Alternatively, we repeat the tests after including in Eq. (5) all the control variables used in Eq. (4).

Untabulated results for the variables of interest are qualitatively identical to those tabulated in this study and

all statistical inferences remain unchanged.

12 While Compustat reports only the current state/county codes of firms’ headquarters, Audit Analytics

updates the information on the state/city locations of client headquarters on an annual basis. Thus, we use the

Audit Analytics database to identify client firm locations.

13 In our initial sample, we find 485 observations for which auditor or client headquarters are not located in

any of the MSAs. Applying this selection criterion, we remove all firms that are located in a remote area in

which local auditors hardly exist. As a sensitivity check, we repeat our analyses after including these

observations and treating them as (1) DMSA = 1, (2) DMSA = 0, or (3) DMSA =1 or 0, following the

definition of D100. We find, however, that the inclusion of these observations in our sample does not alter

our statistical inferences for any of the three cases.

14 When the auditor’s practicing office and the client’s headquarters are in the same city, the distance is

calculated as zero. It is possible that the effect of auditor locality is stronger when the two offices are located

We also obtain non-audit fees data from the Audit Analytics database. We retrieve

all other financial data from the Compustat Industrial annual file. We exclude financial

institutions and utility firms with SIC codes in the ranges 6000–6999 and 4900–4999,

respectively. After applying the above selection procedures and data requirements, we

obtain 12,439 firm–year observations located in 192 MSAs. These observations are

audited by auditors from 767 unique audit practice offices located in 110 MSAs. The

sample size for the tests using Eq. (5) decreases by almost a half, to 6,640, due to further

data requirements.

The Appendix provides information on the locations of clients and auditors in our

full sample. Column (1) of Panel A of the Appendix reports the number of clients (firm–

year observations) in each MSA where more than 100 clients are located. Column (2)

reports the number of local audits performed by audit offices located in the same MSA as

their clients. Column (3) shows the average percentage of local audits in each MSA. The

percentage varies across MSAs, suggesting that the choice of non-local auditors is not

concentrated in certain MSAs. We find that about 80% of audits are carried out by

auditors located in the same MSA, and about 83% are by auditors located within 100

kilometers. Accordingly, only about 3% of firms are audited by auditors located within

100 kilometers but not in the same MSA. There are not many cases (only 62 observations,

reported in the bottom row of Panel A of the Appendix) where auditors are located in the

same MSA as their clients but farther than 100 kilometers away from the clients’

headquarters. Panel B reports the number of observations by the distance between the

audit office and client headquarters when they are not located in the same MSA. It is clear

within a shorter distance within the city, but data unavailability limits such analyses (i.e., no street-level

addresses for auditors’ office are available in the Audit Analytics database). For more details on distance

computation, see Coval and Moskowitz (1999).

from Panel B that some auditors are located nearby even though they are not in the same

MSA as their client firms.

Panel C of the Appendix provides more detailed auditor location information for

1,691 clients located in the New York–Northern New Jersey–Long Island MSA.15 It

shows that about 15% of the clients in the MSA hire auditors from other MSAs, and about

13% of the clients hire auditors located farther than 100 kilometers from their headquarters.

It is noteworthy that some clients in the New York–Northern New Jersey–Long Island

MSA hire auditors located far away, such as from the San Francisco–Oakland–San Jose

MSA (4,120 kilometers away on average), the Los Angeles–Riverside–Orange County

MSA (3,949 kilometers away), and the Denver–Boulder–Greeley MSA (2,644 kilometers

away).16 In contrast, some auditors come from nearby MSAs, such as the Hartford MSA

(69 kilometers away) and the Philadelphia-Wilmington-Atlantic City MSA (73 kilometers

away). These auditors are likely to share the characteristics of the local auditors to the

client firms even though they are not located in the same MSA. This is why we define

D100 as a combined variable of the distance-based and MSA-based measures and use it

(in addition to DMSA, which is an MSA-based measure) to proxy for local audits.

Descriptive Statistics and Univariate Tests

Panel A of Table 2 presents the descriptive statistics for our two discretionary

accrual measures, |DA1| and |DA2|, and two accrual quality measures, AQ1 and AQ2,

15 We choose the New York–Northern New Jersey–Long Island MSA, for example, because the number of

client firms located in this MSA is the largest of our sample.

16 To check if there exist any special reasons to hire long-distance, non-local auditors, we choose the state of

California (where the state-by-state sample size is the largest) and identify clients that hire auditors located

at least 300 miles away from client headquarters. We find that 50 client firms (101 observations) belong to

this category. For these firms, we search for 10-Ks from the EDGAR database to see if these firms have

special connections with the states in which their auditors are located. We find that, out of 50 client firms, 16

have other major offices or plants (except for headquarters) in the states where their auditors are located and

four moved their headquarters to California from different states but continued to hire their previous auditors.

However, we have been unable to find any compelling evidence that the remaining 30 clients have any

special connections to the states in which their auditors are located.

separately, for the local auditor sample (DMSA = 1 or D100 = 1) and the non-local auditor

sample (DMSA = 0 or D100 = 0), along with univariate test results for differences in the

mean and median between the two samples. As shown in Panel A of Table 2, both |DA1|

and |DA2| are significantly lower for the clients of local auditors than for those of non-

local auditors. For example, the mean (median) value of |DA1| is 0.0846 (0.0473) for the

clients of local auditors located in the same MSA, and 0.1046 (0.0570) for those of non-

local auditors located in a different MSA. The differences are significant at the 1% level (t

= -7.90). We also find similar differences for the accrual quality measures.

[ INSERT TABLE 2 HERE! ]

Panel B of Table 2 reports the descriptive statistics for all other variables included

in our main regressions. The sample firms have 3.5 business or geographic segments on

average (LNBGS = 0.9170). The average client size (LNTA) is 12.2340, which is

equivalent to about $200 million of total assets. About 77% of clients are audited by one

of the Big 4 auditors (BIG4), and the average logged auditor tenure (TENURE) is 1.8410,

which is interpreted as about six years of auditor tenure. On average, non-audit service

fees (NAS) are about 67% of total fees, and about 44% of clients hire industry specialists

(INDSPEC). The distributions of the other variables are also shown in Panel B of Table 2.

When the descriptive statistics in Panel B are compared with other studies on audit office-

level analyses, we find that ours are quite comparable to those of Choi et al. (2010) and

Francis and Yu (2009).

Panel A of Table 3 presents the Pearson correlation matrix for all the variables

included in Eq. (4). The two abnormal accrual measures, |DA1| and |DA2|, are highly

correlated with the correlation coefficient of 0.445 (p < 0.01). The auditor locality

indicator, DMSA and D100, is negatively correlated with |DA1| and |DA2| (p < 0.01 for

both). Both |DA1| and |DA2| are significantly correlated with many control variables,

supporting their inclusion as control variables. Finally, we note that the correlations

between the control variables are mostly not very high except for those between BIG4 and

LNTA (0.547) and between NAS and LNTA (0.436). This finding suggests that

multicollinearity is unlikely to be a serious problem.

[ INSERT TABLE 3 HERE! ]

Panel B of Table 3 presents the Pearson correlations among the variables included

in Eq. (5), which can be summarized as follows: First, as expected, the two accrual quality

metrics, AQ1 and AQ2, are highly correlated with each other. Second, both accrual quality

metrics are negatively correlated with our measures of auditor locality, DMSA and D100,

which is consistent with the results of univariate tests (as reported in Panel A of Table 2).

Third, control variables are highly correlated with our accrual quality measures. Finally,

consistent with the findings in Panel A of Table 3, the correlations between our control

variables are mostly not very high except for those between BIG4 and LNTA (0.520) and

between NAS and LNTA (0.436).

EMPIRICAL RESULTS

Main Results Using Discretionary Accrual Measures

Table 4 reports the results of the regression in Eq. (4): In section A (B), |DA1|

(|DA2|) is used as the dependent variable. All reported t-statistics are on an adjusted basis,

using standard errors corrected for clustering at the firm level and heteroskedasticity. As

shown in columns (1a) and (3a), we first estimate Eq. (4) after excluding

DLOCAL*LNBGS from the model. The coefficients on DMSA and D100 are both negative

and significant (-0.0046 with t = -1.77 and -0.0050 with t = -1.73) at the 10% level in two-

tailed tests.17 These results are consistent with H1, that local auditors are, on average,

more effective than non-local auditors in deterring opportunistic earnings management or

biased financial reporting. These results should be interpreted cautiously, however,

because the omission of DLOCAL* LNBGS can create a potential problem of correlated

omitted variables. In what follows, we test our H1 and H2 using the results of estimating

the full model in Eq. (4).

When we include DLOCAL*LNBGS in the model, the coefficients on both DMSA

and D100 remain all significantly negative, as shown in columns (2a), (4a), (1b), and (2b).

Furthermore, the interaction term between DLOCAL and LNBGS is positive and

significant across all columns in both sections A and B of Table 4; for example, in column

(2a), the coefficient on the interaction term is positive and significant at the 5% level

(0.0086 with t = 2.45). These results support H2, implying that the effect of local audits on

audit quality is weaker for more diversified client firms.18

[ INSERT TABLE 4 HERE! ]

To further examine the validity of H1, we investigate whether the clients of local

auditors report a lower level of discretionary accruals at different levels of firm

diversification. When we set the value of LNBGS at the 25th percentile (LNBGS = 0 from

Panel B of Table 2), the coefficient on DMSA and DMSA*LNBGS in column (2a) of

17 When |DA2| is used as the dependent variable in the model excluding DLOCAL*LNBGS, the coefficients

on DMSA and D100 are -0.0037 (t = -1.49) and -0.0028 (t = -1.15). While the coefficient on DMSA is

significant with p = 0.0685 in a one-tailed test, the coefficient on D100 is insignificant.

18 To examine whether local audits constrain either or both income-increasing and income-decreasing

accruals, we split the full sample into two subsamples, with income-increasing (positive) and income-

decreasing (negative) accruals (i.e., DA1 > 0 and DA1 < 0), and then estimate Eq. (4) separately.

Untabulated results suggest that the coefficients on DLOCAL and DLOCAL*LNBGS are significant with

predicted signs when we use the subsample of income-increasing accruals. However, when we use the

subsample of income-decreasing accruals, both coefficients are insignificant, although they have the

expected signs. These results suggest that informational advantages associated with local audits are related to

more accurate accruals in general, but the effect is stronger for reducing income-increasing accruals. This

finding is not very surprising, because auditors tend to be more concerned about their clients’ income-

increasing accruals (Kim et al. 2003).

Table 4 is translated into -0.0119 (-0.0119 + [0.0086 * 0] = -0.0119). This implies that

discretionary accruals of clients of local auditors are lower than those of clients of non-

local auditors by 0.0119, when the value of LNBGS is equal to its 25th percentile value.

However, when we set the value of LNBGS at the 50th (LNBGS = 1.0986), 75th (LNBGS =

1.6094), or 99th (LNBGS = 2.4849) percentile value, we fail to find significant differences

in the level of discretionary accruals between local and non-local auditors. As reported in

the bottom three rows of Table 4, the F-test result for the sum of the coefficients on

DLOCAL and DLOCAL*LNBGS times the 50th, 75th, or 99th percentile value of LNBGS

is not significant at the 10% level. The results using the estimated coefficients in other

columns are qualitatively identical. These results suggest that local auditors perform

higher-quality audit services than non-local auditors, but only for relatively less diversified

clients.19

To examine the economic significance of our results, we translate the estimated

coefficients of the variables of interest into the magnitude of absolute discretionary

accruals as a percentage of lagged total assets and calculate the percentage difference

between local and non-local audits, using the estimated coefficients reported in column

(2a) of Table 4. The estimated coefficient on DMSA, -0.0119, and that on DMSA*LNBGS,

0.0086, mean that, on average, the clients of local auditors exhibit an approximately 4%

lower level of absolute discretionary accruals than those of non-local auditors when we set

19 To further explore this issue, we form four subsamples based on the number of segments (i.e., firms with

two segments (one business segment and one geographic segment), three segments, four or five segments,

and six or more segments). We then estimate Eq. (4), after excluding LNBGS and DLOCAL*LNBGS, for

each of four subsamples. When DMSA is used as DLOCAL and the dependent variable is |DA1|, the

coefficient on DMSA is equal to -0.0123 (t = -2.56) for the subsample of firms with two segments (4,272

observations). However, the coefficient is not significant for all other subsamples. For example, the

coefficient is -0.0004 (t = -0.10) for the subsample of firms with six or more segments (3,321 observations).

When we perform additional analyses for firms with seven or more segments, eight or more segments, and

nine or more segments, the results show that the coefficient on DMSA is insignificant, with a negative sign

for each of these subsamples.

all the other variables to their respective mean values.20 For relatively more diversified

firms (LNBGS = 1.6094, which is the 75th percentile value of the variable), local audits

yield a lower level of absolute discretionary accruals by about 3%, whereas for less

diversified firms (LNBGS = 0, which is the 25th percentile value), local audits yield a

lower absolute discretionary accruals by about 12% compared with those of non-local

audits. The 9% difference between the 25th and 75th percentiles suggests that the effect of

local audits varies substantially with the level of firm diversification.

The coefficients on the control variables are, overall, in line with the evidence

reported in prior earnings management research. The coefficient on LNBGS is negative

and significant, suggesting that diversified firms with relatively stable operations have

higher earnings quality. The coefficient on LNTA is highly significant, with a negative

sign across all four columns, suggesting that large client firms tend to engage in earnings

management to a lesser extent than small client firms (Dechow and Dichev 2002). The

coefficient on BIG4 is significantly negative in all specifications, suggesting that Big 4

auditors are more effective than non-Big 4 auditors in constraining aggressive earnings

management. The coefficient on BTM (CHGSALE) is significantly negative (positive),

which suggests that high-growth firms manage earnings more aggressively. The

coefficients on LOSS and ZMIJ are significant with an expected positive sign in most

cases, suggesting that client firms in financial distress are more likely to engage in

earnings management. Consistent with evidence reported in previous research (e.g.,

Ashbaugh et al. 2003; Kim et al. 2003), the coefficients on CFO and LAGACCR have a

negative sign. Finally, consistent with the findings in Kallapur et al. (2010), the auditor

20 The average magnitude of absolute discretionary accruals as a percentage of lagged total assets estimated

from the coefficients reported in column (2a) is 0.1039 for the clients of non-local auditors and 0.0999 for

the clients of local auditors when we set all the other variables at their respective mean values.

concentration measure is negatively associated with the magnitude of absolute

discretionary accruals.

As robustness checks, we re-estimate Eq. (4), using the performance-unadjusted

abnormal accruals as the dependent variable—that is, using the modified Jones model in

Eq. (2), as specified in Dechow et al. (1995). We also run the median regression and year-

by-year regressions after excluding year indicators. We also repeat the tests using various

subsamples. Though not tabulated, the results from these robustness checks are, overall,

qualitatively similar to those reported in Table 4.21

Main Results Using Accrual Quality Measures

We further examine whether local audits are related to the accrual quality of client

firms. As explained earlier, we use both the accrual quality measure (AQ1) developed by

Dechow and Dichev (2002) and modified by McNichols (2002) and its discretionary

component (AQ2) estimated based on the approach outlined in FLOS as proxies for audit

quality. We then estimate the regression model in Eq. (5). Table 5 reports the regression

results using the reduced sample of 6,640 firm–year observations.

21 For example, for the regressions using the performance-unadjusted abnormal accruals and the median

regression, the coefficients on DMSA for the regression in column (2a) of Table 4 are -0.0188 (t = -3.65)

and -0.0091 (t = -4.80), respectively, and those on DMSA*LNBGS are 0.0127 (t = 3.31) and 0.0072 (t = 4.41),

respectively. In year-by-year regressions, all four yearly regressions yield negative (positive) coefficients on

DMSA (DMSA*LNBGS) and two (three) of them are significant at least at the 10% level. The average of the

yearly coefficients on DMSA (DMSA*LNBGS) is -0.0123 (0.0087) and the corresponding Fama–MacBeth-

style t-statistic is 2.79 (2.46). We also perform several additional tests using various subsamples. First, we

restrict our sample to the largest 100 MSAs in terms of population. Second, we use the median size of MSAs

to classify large and small MSAs and repeat the analyses with both subsamples. Third, we divide the full

sample into observations from the six largest MSAs and from the other MSAs and repeat the analyses with

both subsamples. Fourth, we delete observations with auditors coming from cities located farther than 500

(1,000) kilometers away and repeat the analyses. In sum, we find evidence consistent with the predictions of

both H1 and H2 in all of these analyses. The coefficients on both DLOCAL and DLOCAL*LNBGS remain

significant across almost all cases; the only exception is that we fail to find a significant coefficient on

DLOCAL when we use observations from the six largest MSAs and DLOCAL is proxied by DMSA.

In Table 5, the coefficient on DLOCAL is negative and significant across all four

columns.22 It also shows that the coefficients on DLOCAL*LNBGS are positive and

significant in all columns. These results suggest that while auditor locality is associated

with less noise in client accruals (i.e., greater accrual quality) in general, the association is

weaker for clients with a more diversified business structure. These findings lend further

support to the results reported in Table 4. The estimated coefficients of the control

variables in Table 5 are generally consistent with those reported in Table 4, with the

notable exception that the coefficient on CFO is positively (negatively) significant in

Table 5 (Table 4).

In addition, we use the absolute values of residuals from Eq. (5) to proxy for

accrual quality, as suggested by Dechow and Dichev (2002) and used by Srinidhi and Gul

(2007), in untabulated analyses. Similar to the tabulated results, we find that the variables

of interest are all significant. Overall, the results using our accrual quality measures, AQ1

and AQ2, also support our hypotheses H1 and H2. Although not separately tabulated, we

conduct a variety of sensitivity analyses and find that our results are robust to different

regression methods and alternative measures of our test variables.23

[ INSERT TABLE 5 HERE! ]

22 When AQ1 [AQ2] is used as the dependent variable in the model excluding DLOCAL*LNBGS, the

coefficients on DMSA and D100 are -0.0011 (t = -1.37) and -0.0013 (t = -1.44) [-0.0016 (t = -1.98) and -

0.0016 (t = -1.73)]. These coefficients are all significant at least at the 10% level in one-tailed tests.

23 For example, the coefficient on DMSA for the regression in column (1a) of Table 5 is -0.0028 (t = -1.82)

and that on DMSA*LNBGS is 0.0021 (t = 1.69) for the median regression. In year-by-year regressions, all

four yearly regressions yield negative (positive) coefficients on DMSA (DMSA*LNBGS) and two of them are

significant at least at the 10% level. The average of the yearly coefficient on DMSA (DMSA*LNBGS) is -

0.0026 (0.0014) and the corresponding Fama–MacBeth-style t-statistics is 1.65 (1.90). We also perform

several additional tests using various subsamples similar to those described in footnote 21. Most of these

analyses reveal evidence consistent with the predictions of both H1 and H2. The only exceptions are that

(1) the coefficients on DMSA and DMSA*LNBGS become insignificant when we use observations from the

six largest MSAs, and (2) the coefficient on DLOCAL*LNBGS becomes insignificant when we use 150, 200,

or 250 kilometers instead of 100 kilometers as cut-off values for the distance-based measure or the combined

measure of both distance-based and MSA-based measures.

Further Analyses

Our results so far are based on the full sample of 12,439 or 6,640 observations,

depending on the dependent variables used. While this full sample consists of pooled

client firms audited by all the audit offices in our dataset, it includes one group of auditors

having both local and non-local clients and the other group having only local clients. One

cannot therefore rule out the possibility that a systematic difference in audit quality exists

between these two groups of auditors. In this case, our reported results may suffer from

unknown confounding effects. To alleviate this concern, we obtain a more homogeneous

sample of audit clients by restricting our sample to the clients of audit offices that engage

in both local and non-local audits. We then re-estimate our regression models to see if the

same audit office (which audits both local and non-local client firms) exhibits differential

audit quality depending on its clients’ geographic location.

We find that the empirical results for this subsample analysis are qualitatively

similar to the tabulated results. For example, when we repeat the test equivalent to that

reported in the column (2a) of Table 4 (n = 9,367 from 359 distinct audit offices of 95

audit firms), the coefficients on DMSA and DMSA*LNBGS are -0.0078 (t = -1.66) and

0.0066 (t = 1.88), respectively. When we repeat the test reported in column (1a) of Table 5

(n = 5,045 from 319 distinct audit offices of 75 audit firms), the coefficients on DMSA and

DMSA*LNBGS are -0.0020 (t = -1.47) and 0.0013 (t = 1.37), respectively. These results

are significant at least at the 10% level in one-tailed tests. In sum, our results from the

reduced sample reveal that, when auditors have both local and non-local clients, audit

quality tends to be higher for local clients than for non-local clients, and this tendency is

significantly weakened for more diversified clients.

Controlling for Potential Self-Selection Bias

To address potential self-selection bias associated with local versus non-local

auditor choices, we re-estimate our main regression models in Eqs. (4) and (5) by applying

the Heckman-type (1979) two-stage treatment effect approach. Given that previous

research has paid little attention to the issue of auditor locality, little is known about the

factors that determine a client firm’s decision to appoint a local or non-local auditor. In an

attempt to identify an initial list of potential determinants, we searched proxy statements,

held discussions with auditors, and conducted an extensive review of the extant auditing

literature. As a result, we first identified 13 variables (LNTA, LNBGS, LEV, LOSS, GCM,

CAIN, ISSUE, INST, ANA, BIG4, NAS, NBIG, and WAGE) that are likely to influence local

versus non-local auditor choices, for the first-stage probit auditor choice model.

We expect a positive coefficient on LNTA, because large firms can easily hire local

auditors, since auditors are less likely to turn away large clients in the same locale. We

expect a negative coefficient on LNBGS because diversified firms are more likely to hire

auditors from another MSA, since their plants, offices, or branches are more spread out to

other regions. It would be more difficult for financially unhealthy clients to hire local

auditors because local auditors, who are familiar with the financial problems of potential

local clients, may be reluctant to audit them. In such a case, one would observe negative

coefficients on leverage (LEV), the loss indicator (LOSS), and the going concern opinion

indicator (GCM). If a client firm holds substantial long-term operating assets, more audit

work would be required in the region where its factory/plant/warehouse is located rather

than at its headquarters. Such a client is therefore less likely to appoint auditors near its

headquarters. We therefore expect a negative coefficient on our capital intensity measure

(CAIN). Firms that issue a significant amount of new equity or debt (ISSUE) have stronger

incentives to manage earnings (Teoh et al. 1998). Such firms may be more likely to hire

non-local auditors for more aggressive earnings management. Since institutional investors

demand higher-quality audits to enhance corporate monitoring (Han et al. 2008), client

firms with higher institutional ownership (INST) may be more likely to appoint local

auditors. Similarly, since analysts’ earnings forecast performance is related to audit quality

(Behn et al. 2008), client firms with more analysts following (ANA) may feel more

pressure to appoint local auditors. Therefore, we expect the coefficient on ISSUE to be

negative and the coefficients on INST and ANA to be positive.

Because Big 4 auditors have offices in many different MSAs, it is possible that the

clients of Big 4 auditors will hire a Big 4 office in their own MSA rather than the same

audit firm’s office in a different MSA. Thus, we expect a positive sign for BIG4. Because

local auditors may be less likely to turn away clients that have a greater chance of

procuring non-audit services, we expect a positive coefficient on NAS. If a client firm

prefers a particular Big 4 audit firm, the likelihood of appointing a local Big 4 auditor will

be greater when more Big 4 auditors are available in the MSA. We therefore expect a

positive coefficient on NBIG (the number of Big 4 offices in the MSA). We include the

mean hourly rate of auditor wages in each MSA (WAGE) to control for the effect of wage

levels in an MSA in the likelihood of appointing local auditors.24

With the determinants mentioned above, we estimate the following probit auditor

choice model in Eq. (6) using two different dependent variables, namely, DMSA and D100.

We find, however, that the estimated results using DMSA are qualitatively identical with

24 On the one hand, if the audit fee level is high in an MSA, cost-conscious clients are less likely to appoint

local auditors (Jensen and Payne 2005). On the other hand, quality-conscious clients are still likely to

appoint local auditors to the extent that the higher fee level is accompanied by higher-quality audit services.

We therefore do not predict a sign for the coefficient on WAGE.

those using D100. For brevity, we therefore report only the results of our probit estimation

with DMSA as the dependent variable25:

DMSAjt* = -2.3394 + 0.0040 LNTAjt + 0.022 LNBGSjt – 0.0015 LEVjt

(13.53***) (0.32) (0.76) (-0.05)

– 0.0924 LOSSjt – 0.1921 GCMjt – 0.1823 CAINjt – 0.0773 ISSUEjt

(-2.87***) (-3.33***) (-2.60***) (-2.64***)

+ 0.3807 INSTjt + 0.0471 ANAjt + 0.2867 BIG4jt (6)

(4.50***) (2.01**) (7.31***)

+ 0.1630 NASjt + 0.5906 NBIGmt + 0.0244 WAGEmt + Industry Indicators

(2.85***) (30.59***) (5.07***)

where the subscripts, j m, and t denote an individual client firm, an MSA, and time,

respectively, and the numbers in parentheses denote z-statistics. The superscript *** (**)

denotes a p-value of less than 1% (5%) in two-tailed tests, respectively. The variable

DMSA* represents the likelihood that a client chooses a local auditor, and is ex post set to

equal one if the local audit office is located in the same MSA where the client firm is

headquartered, and zero otherwise. The exact definitions of all variables are provided in

Table 1.

The results of the above probit estimation show that all the coefficients on

significant variables have expected signs. The explanatory power of the model, measured

by pseudo-R2, is 17.46%, suggesting that the model explains the choice between local

versus non-local auditors reasonably well. In the second stage, we estimate Eqs. (4) and

(5) after adding the inverse Mills ratio (LAMBDA), obtained from the first-stage probit

estimation as an additional control variable. Untabulated results from the second-stage

regressions show that the coefficients of the variables of interest remain qualitatively

25 Because of the concern over the efficacy of the first-stage selection model (Francis and Lennox 2008), we

include various variables in the first-stage model that are not in the second stage. Furthermore, when we

check the variance inflation factor (VIF) in the second-stage regressions, as Francis and Lennox (2008)

recommend, we find that the VIF values are mostly very small. We also considered a number of additional

variables that are not included in our final model, but we find that the coefficients of these variables are

insignificant and that adding these variables does not improve the model’s explanatory power.

identical with those reported in Tables 4 and 5. All statistical inferences remain unchanged

with this two-stage treatment effect approach.26 In short, the above results suggest that our

reported results are robust to the potential self-selection bias associated with local versus

non-local auditor choices.

CONCLUSION

While many studies have already examined the effect of geographic proximity in

the contexts of domestic and international portfolio decisions, analysts’ forecast accuracy,

and other areas of finance and accounting research, little attention has been paid to the

issue in the context of the auditor–client relationship. Our results show that auditor–client

geographic proximity or auditor locality has a positive impact on audit quality by

constraining opportunistic earnings management or improving accrual quality. We also

find that this impact is relatively weaker or absent for diversified firms with more

operating or geographic segments.

These results help us better understand why local audits are so prevalent and Big 4

audit firms have continuously expanded their practicing offices to cities in which their

clients are headquartered. Furthermore, this study will be of interest to regulators, since it

demonstrates that the proximity of auditors to clients can be beneficial to audit quality.

As in many other studies, we measure audit quality using accrual-based proxies.

We admit, however, that the accrual-based proxies are not the only valid empirical

measures of audit quality and that these measures are often criticized for inherent

26 For example, when we perform a two-stage analysis for the regression in column (2a) of Table 4, the

coefficient on DMSA is -0.0112 (t = -2.44) and that on DMSA*LNBGS is 0.0086 (t = 2.46). When we

perform a two-stage analysis for the regression in column (1a) of Table 5, the coefficient on DMSA is -

0.0024 (t = -1.65) and that on DMSA*LNBGS is 0.0014 (t = 1.71). While the F-tests using the 50th, 75th, or

99th percentile value of LNBGS (similar to those reported in the bottom three rows of Tables 4 and 5) yield

all insignificant results, the F-tests using the 25th percentile value of LNBGS yield significant results.

measurement errors, although we use several advanced approaches to alleviate this

concern. We therefore recommend continued research using alternative measures of audit

quality to further validate our findings and better understand the role of auditor locality in

shaping the auditor–client relationship.

[INSERT APPENDIX HERE!]

REFERENCES

Ashbaugh, H., R. LaFond, and B. Mayhew. 2003. Do non-audit services compromise auditor

independence? Further evidence. The Accounting Review 78 (3): 611–639.

Audretsch, D., and M. Feldman. 1996. R&D spillovers and the geography of innovation and

production. American Economic Review 86 (3): 630–652.

Audretsch, D., and P. Stephan. 1996. Company–scientist locational links: The case of

biotechnology. American Economic Review 86 (3): 641–652.

Baik, B., J.-K. Kang, and J.-M. Kim 2010. Local institutional investors, information asymmetries,

and equity returns. Journal of Financial Economics 97: 81–106.

Ball, R., and L. Shivakumar. 2006. The role of accruals in asymmetrically timely gain and loss

recognition. Journal of Accounting Research 44: 207–241.

Baltes, B. B., M. W. Dickson, M. P. Sherman, C. C. Bauer, and J. S. LaGanke. 2002. Computer-

mediated communication and group decision making: A meta-analysis. Organizational

Behavior and Human Decision Processes 87 (1): 156–179.

Becker, C., M. DeFond, J. Jiambalvo, and K. Subramanyam, 1998. The effect of audit quality on

earnings management. Contemporary Accounting Research 15 (1): 1–24.

Bedard, J. C., and K. Johnstone. 2004. Earnings management risk, corporate governance risk, and

auditors’ planning and pricing decisions. The Accounting Review 79 (2): 277–304.

Behn, B. K., J.-H. Choi, and T. Kang. 2008. Audit quality and properties of analyst earnings

forecasts. The Accounting Review 83 (2): 327–349.

Bodnaruk, A. 2009. Proximity always matters: Local bias when the set of local companies changes.

Review of Finance 13 (4): 629–656.

Boone, J. P., I. K. Khurana, and K. K. Raman. 2010. Do the Big 4 and the second-tier firms

provide audits of similar quality? Journal of Accounting and Public Policy 29 (4): 330–352.

Carcello, J. V., R. H. Hermanson, and N. T. McGrath. 1992. Audit quality attributes: The

perceptions of audit partners, preparers, and financial statement users. Auditing: A Journal of

Practice and Theory 11 (1): 1–15.

Choi, J.-H., F. Kim, J.-B. Kim, and Y. Zang. 2010. Audit office size, audit quality and audit

pricing. Auditing: A Journal of Practice and Theory 29 (1): 73–97.

Chung, H., and S. Kallapur. 2003. Client importance, non-audit services, and abnormal accruals.

The Accounting Review 78 (4): 931–955.

Coval, J., and T. Moskovitz, 1999. Home bias at home: Local equity preference in domestic

portfolios. Journal of Finance 54 (6): 2045–2073.

Coval, J., and T. Moskovitz, 2001. The geography of investment: Informed trading and asset

prices. Journal of Political Economy 109 (4): 811–841.

Covrig, V. M., M. DeFond, and M. Hung. 2006. Home bias, foreign mutual fund holdings, and the

voluntary adoption of international accounting standards. Journal of Accounting Research 45

(1): 41–70.

Craswell, A., D. J. Stokes, and J. Laughton. 2002. Auditor independence and fee dependence.

Journal of Accounting and Economics 33 (2): 253–275.

Davis, L. R., B. S. Soo, and G. M. Trompeter. 2009. Auditor tenure and the ability to meet or beat

earnings forecasts. Contemporary Accounting Research 26 (2): 517-548.

Dechow, P. M., and I. D. Dichev. 2002. The quality accruals and earnings: The role of accrual

estimation errors. The Accounting Review 77 (Supplement): 35–59.

Dechow, P. M., R. G. Sloan, and A. P. Sweeney. 1995. Detecting earnings management. The

Accounting Review 70 (2): 193–225.

DeFond, M., J. R. Francis, and X. Hu. 2011. The geography of SEC enforcement and auditor

reporting for financially distressed clients. Working paper, University of South California.

DeFond, M., and J. Jiambalvo, 1994. Debt covenant violations and manipulations of accruals.

Journal of Accounting and Economics 17 (1–2): 145–176.

DeFond, M., K. Raghunandan, and K. Subramanyam. 2002. Do non-audit services fees impair

auditor independence? Evidence from going concern audit opinions. Journal of Accounting

Research 40 (4): 1247–1274.

Doherty-Sneddon, G., A. Anderson, C. O’Malley, S. Langton, S. Garrod, and V. Bruce. 1997.

Face-to-face and video-mediated communication: A comparison of dialogue structure and task

performance. Journal of Experimental Psychology: Applied 3 (2): 105–125.

Ferguson, A., J. R. Francis, and D. J. Stokes. 2003. The effects of firm-wide and office-level

industry expertise on audit pricing. The Accounting Review 78 (2): 429–449.

Francis, J., R. LaFond, P. Olsson, and K. Schipper. 2005. The market pricing of accruals quality.

Journal of Accounting and Economics 39 (2): 295–327.

Francis, J. R., and C. Lennox. 2008. Selection models in accounting research. Working paper,

University of Missouri-Columbia.

Francis, J. R., K. Reichelt, and D. Wang. 2005. The pricing of national and city-specific reputation

for industry expertise in the U.S. audit market. The Accounting Review 80 (1):113–136.

Francis, J. R., D. J. Stokes, and D. Anderson. 1999. City markets as a unit of analysis in audit

research and the re-examination of Big 6 market shares. Abacus 35 (2): 185–206.

Francis, J. R., and M. Yu. 2009. The effect of Big four offices on audit quality. The Accounting

Review 84 (5): 1521–1552.

Francis, J. R., and M. Yu. 2011. Office size of Big four auditors and client restatements. Working

paper, University of Columbia-Missouri.

Frankel, R., M. Johnson, and K. Nelson, 2002. The relation between auditors' fees for non-audit

services and earnings quality. The Accounting Review 77 (Supplement): 71–105.

Gul, F. A., J.-B. Kim, and A. Qiu. 2006. Does the political economy affect auditor incentives?

Some Chinese evidence on non-Big 4 audit quality. Working paper, Hong Kong Polytechnic

University.

Hambley, L. A., T. A. O’Neill, and T. J. B. Kline. 2007. Virtual team leadership: The effects of

leadership style and communication medium on team interaction styles and outcomes.

Organizational Behavior and Human Decision Processes 103 (1): 1–20.

Han, S., T. Kang, and L. Rees. 2008. The association between institutional ownership and audit

characteristics. Working paper, Singapore Management University.

Heckman, J. 1979. Sample selection bias as specification error. Econometrica 47: 153–161.

Hinds, P. J., and M. Mortensen. 2005. Understanding conflict in geographically distributed teams:

The moderating effects of shared identity. Shared context, and spontaneous communication.

Organization Science 16 (3): 290–307.

Hogan, C., and R. D. Martin. 2009. Risk shifts in the market for audits: An examination of changes

in risk for ―second tier‖ audit firms. Auditing: A Journal of Practice and Theory 28 (2): 93–

118.

Hong, H., J. Kubik, and J. Stein. 2004. Social interaction and stock market participation. Journal

of Finance 59 (1): 137–163.

Ivkovich Z., and S. Weisbenner, 2005. Local does as local is: Information content of the

geography of individual investors' common stock investments. Journal of Finance 60 (1):

267–306.

Jensen, K., and J. Payne. 2005. The introduction of price competition in a municipal audit market.

Auditing: A Journal of Practice and Theory 24 (2): 137–152.

Johnson, V. E., I. K. Khurana, and J. K. Reynolds. 2002. Audit-firm tenure and the quality of

financial reports. Contemporary Accounting Research 19 (4): 637–660.

Jones, J. 1991. Earnings management during import relief investigations. Journal of Accounting

Research 29 (1): 193–228.

Kallapur, S., S. Sankaraguruswamy, and Y. Zang. 2010. Audit market competition and audit

quality. Working paper, Indian School of Business.

Kedia, S., and S. Rajgopal. 2011. Do the SEC’s enforcement preferences affect corporate

misconduct? Journal of Accounting and Economics 51 (3): 259-278.

Kim, J.-B., R. Chung, and M. Firth. 2003. Auditor conservatism, asymmetric monitoring and

earnings management. Contemporary Accounting Research 20 (2): 323–359.

Knechel, W., S. Salterio, and B. Ballou. 2007. Auditing: Assurance and Risk, 3rd edition. South-

Western College Publishing.

Kothari, S. P., A. J. Leone, and C. E. Wasley. 2005. Performance matched abnormal accrual

measures. Journal of Accounting and Economics 39 (1): 163–197.

Krishnan, G. 2003. Does Big 6 auditor industry expertise constrain earnings management?

Accounting Horizons 17 (Supplement): 1–16.

Louis, H. 2005. Acquirers’ abnormal returns and the non-Big 4 auditor clientele effect. Journal of

Accounting and Economics 40 (1–3): 75–99.

Malloy, C. 2005. The geography of equity analysts. Journal of Finance 60 (2): 719–755.

McNichols, M. F. 2002. Discussion of the quality of accruals and earnings: The role of accrual

estimation errors. The Accounting Review 77: 61–69.

Menon, K., and D. Williams. 2004. Former audit partners and abnormal accruals. The Accounting

Review 79 (4): 1095–1118.

Myers, J., L. Myers, and T. Omer. 2003. Exploring the term of the auditor–client relationship and

the quality of earnings: A case for mandatory auditor rotation? The Accounting Review 78 (3):

779–799.

Petersen, M., and R. Rajan. 2002. Does distance still matter? The information revolution in small

business lending. Journal of Finance 57 (6): 2533–2570.

Pirinsky, C., and Q. Wang. 2006. Does corporate headquarters location matter for stock returns?

Journal of Finance 61 (4): 1991–2015.

Reichelt, K., and D. Wang. 2010. National and office-specific measures of auditor industry

expertise and effects on audit quality. Journal of Accounting Research 48 (3): 647–686.

Reynolds, J. K., and J. Francis. 2000. Does size matter? The influence of large clients on office-

level auditor reporting decisions. Journal of Accounting and Economics 30: 375–400.

Srinidhi, B. N., and F. A. Gul. 2007. The differential effects of auditors’ nonaudit and audit fees on

accrual quality. Contemporary Accounting Research 24 (2): 595–629.

Teoh, S. T., I. Welch, and T. J. Wong. 1998. Earnings management and the post-issue performance

of seasoned equity offerings. Journal of Financial Economics 50: 63–99.

Wallman, S. 1996. The future of accounting, Part III: Reliability and auditor independence.

Accounting Horizon 10 (4): 76–97.

Wang, Q., T. J. Wong, and L. Xia. 2008. State ownership, institutional environment and auditor

choice: Evidence from China. Journal of Accounting and Economics. 46 (1): 112–134.

Zmijewski, M. E. 1984. Methodological issues related to the estimation of financial distress

prediction models. Journal of Accounting Research 22 (Supplement): 59–82.

APPENDIX

In Panel A, column (1) reports the number of clients (firm–year observations) in each MSA where

more than 100 clients are located. Column (2) reports the number of local audits performed by the

audit offices located in each MSA for the clients located in the same MSA. Column (3) reports the

percentage of local audits out of the total number of clients in each MSA. Panel B reports the

number of observations by the distance between auditor and client when they are not located in the

same MSA. Panel C provides a more detailed breakdown of clients located in the New York–

Northern New Jersey–Long Island MSA. Column (1) shows the number of clients that hire

auditors from different MSAs. Column (2) shows the average distance of the clients and their

respective auditors.

Panel A: Number of observations by the location of clients

MSA

(1)

Number of

clients

(2)

Number of

local audits

(3)

Percentage

[(2)/(1)]

New York–Northern New Jersey–Long Island

1,691

1,434

San Francisco–Oakland–San Jose

1,282

1,203

Los Angeles–Riverside–Orange County

786

705

Boston–Worcester–Lawrence

751

698

Chicago–Gary–Kenosha

491

459

Houston–Galveston–Brazoria

499

426

Dallas–Fort Worth

443

385

Washington–Baltimore

410

355

Minneapolis–St. Paul

402

379

Philadelphia–Wilmington–Atlantic City

376

288

Atlanta

331

275

San Diego

322

223

Denver–Boulder–Greeley

297

266

Miami–Fort Lauderdale

191

144

Seattle–Tacoma–Bremerton

191

172

Cleveland–Akron

168

156

St. Louis

165

142

Phoenix–Mesa

159

128

Detroit–Ann Arbor–Flint

147

121

Portland–Salem

132

119

Pittsburgh

131

104

Milwaukee–Racine

122

Others (all combined)

2,952

1,715

Number of clients audited by auditors in the same MSA

9,996 (80%)

Number of clients audited by auditors from different MSAs

2,443 (20%)

Number of clients audited by auditors within 100 kilometers

10,366 (83%)

Number of clients audited by auditors farther than 100 kilometers

2,073 (17%)

Number of clients audited by auditors within 100 kilometers but

not in the same MSA

432 (3%)

Number of clients audited by auditors not within 100 kilometers

but in the same MSA

62 (0.5%)

Appendix (continued)

Panel B: Distance between auditor and client when they are not located in the same MSA

Distance

0–

100 km

100–

250 km

250–

500 km

500–

1,000 km

1,000–

2,000 km

or more

Total

Number of

Observations

432

662

324

275

380

370

2,443

Panel C: Locations of auditors for clients in the New York–Northern New Jersey–Long

Island MSA

MSA where auditors are located

(1)

Number of

observations

(2)

Average

Distance

(kilometers)

New York–Northern New Jersey–Long Island

1,434

Hartford

Boston–Worcester–Lawrence

275

Philadelphia–Wilmington–Atlantic City

San Francisco–Oakland–San Jose

4,120

Los Angeles–Riverside–Orange County

3,949

Minneapolis–St. Paul

1,640

Miami–Fort Lauderdale

1,708

Indianapolis

1,055

Washington–Baltimore

287

Denver–Boulder–Greeley

2,644

Oklahoma City

2,131

Providence–Fall River–Warwick

137

Albany–Schenectady–Troy

179

Birmingham

1,379

Charlotte–Gastonia–Rock Hill

840

Cleveland–Akron

653

San Diego

3,917

West Palm Beach–Boca Raton

1,662

Other MSAs (16 MSAs combined)

Number of clients audited by auditors in the same MSA

1,434 (85%)

Number of clients audited by auditors from different MSAs

257 (15%)

Number of clients audited by auditors within 100 kilometers

1,479 (87%)

Number of clients audited by auditors farther than 100 kilometers

212 (13%)

Table 1 Variable Definition and Measurement

(continued on next page)

|DA|

Absolute value of discretionary (abnormal) accruals. The current study uses two

proxies: DA1 and DA2, where DA1 is abnormal accruals as measured by Ball and

Shivakumar’s (2006) method; DA2 is performance-matched abnormal accruals

as measured by Kothari et al.’s (2005) method.

Accrual quality, where AQ1 is the accrual quality as measured by Dechow and

Dichev’s (2002) method; AQ2 comprises discretionary components of accrual

quality as measured by Francis et al.’s (2005) method.

DLOCAL

Indicator variable for auditor location. We use two proxies, DMSA and D100,

where DMSA is one if the audit office and client firm’s headquarters are located

in the same MSA, and zero otherwise; D100 is one if the distance between the

audit office and the client firm’s headquarters is less than 100 kilometers from

the client firm’s headquarter or if the audit office and the client firms’

headquarters are located in the same MSA, and zero otherwise.

LNBGS

Natural log of the sum of the number of business and geographic segments

minus one. If business or geographic segment data are missing for a given

observation from Compustat, we assign it a value of one. Thus, the minimum

value of LNBGS is zero (i.e., log(2 - 1) = 0).

LNTA

Natural log of total assets in thousands of dollars.

BIG4

One if the auditor is one of the Big 4 firms, and zero otherwise.

TENURE

Auditor tenure, measured as the natural log of the number of years the incumbent

auditor has served the client firm.

NAS

Relative importance of non-audit services, measured as the ratio of the natural

log of non-audit fees over the natural log of total fees.

INDSPEC

An indicator variable for auditor industry expertise that equals one if the audit

firm is the industry leader for the audit year in the audit market of the MSA

where the auditor is located, and zero otherwise. We calculate each audit firm’s

industry market share of audit fees for an MSA as the proportion of audit fees

earned by each firm in the total audit fees earned by all audit firms in the MSA

that serve the same industry. Each industry is defined based on two-digit SIC

codes.

CHGSALE

Change in sales deflated by lagged total assets.

BTM

Book-to-market ratio, winsorized at zero and four.

LOSS

One if the firm reports a loss for the year, and zero otherwise.

LEV

Leverage, measured as total liabilities divided by total assets.

ZMIJ

Zmijewski’s (1984) financial distress score, winsorized at five and minus five.

ISSUE

One if the sum of debt or equity issued during the past three years is more than

5% of the total assets, and zero otherwise.

CFO

Operating cash flows, taken from the cash flow statement, deflated by lagged

total assets.

LAGACCR

One-year lagged total accruals. Accruals are defined as income before

extraordinary items minus operating cash flows from the statement of cash flow

deflated by lagged total assets.

CONCENT

A measure of auditor concentration by each MSA, measured by the Herfindahl

index of the number of clients for each audit office.

OPCYCLE

Length of the operating cycle, the log of the sum of a firm’s days accounts

receivable and days inventory.

STD_CFO

Cash flow variability, the standard deviation of a firm’s rolling five-year cash

flows from operations, scaled by averaged total assets.

STD_SALES

Sales variability, the standard deviation of a firm’s rolling five-year sales

revenues, scaled by averaged total assets.

LNSALES

Natural log of sales.

NEGEAR

Inventory and receivables divided by total assets.

GCM

One if the firm receives a going concern audit opinion in the current year, and

zero otherwise.

CAIN

Capital intensity measured by long-term assets divided by total assets.

NBIG

Number of Big 4 audit offices in the MSA.

WAGE

Mean hourly auditor wage in the client firm’s MSA, from Occupational

Employment Statistics, Bureau of Labor Statistics, U.S. Department of Labor.

ANA

Natural log of one plus the number of analysts following the firm.

INST

Natural log of one plus the percentage of institutional shareholdings.

Table 2 Descriptive Statistics

Panel A: Descriptive Statistics and Results of Univariate Tests

DMSA = 1a

DMSA = 0 b

Test for

Equality

(t-value)

Variable

Mean

Median

Std. Dev.

Mean

Median

Std. Dev.

│DA 1│

0.0846

0.0473

0.1062

0.1046

0.0570

0.1333

-7.90***

│DA 2│

0.1105

0.0737

0.1077

0.1262

0.0847

0.1209

-6.29***

AQ1

0.0446

0.0351

0.0330

0.0493

0.0385

0.0356

-4.42***

AQ2

-0.0004

-0.0039

0.0257

0.0016

-0.0029

0.0278

-2.46**

D100 = 1c

D100 = 0 d

Test for

Equality

(t-value)

Variable

Mean

Median

Std. Dev.

Mean

Median

Std. Dev.

│DA 1│

0.0848

0.0476

0.1068

0.1078

0.0590

0.1360

-8.44***

│DA 2│

0.1108

0.0739

0.1080

0.1277

0.0855

0.1225

-6.26***

AQ1

0.0448

0.0351

0.0334

0.0495

0.0395

0.0344

-4.03***

AQ2

-0.0003

-0.0040

0.0261

0.0018

-0.0026

0.0262

-2.28**

Panel B: Descriptive Statistics for Variables e

Variable

Mean

Std. Dev.

25%

Median

75%

99%

│DA 1│

0.0885

0.1123

0.0001

0.0198

0.0490

0.1122

0.5444

│DA 2│

0.1136

0.1106

0.0010

0.0328

0.0761

0.1588

0.4748

AQ1

0.0454

0.0335

0.0066

0.0224

0.0357

0.0587

0.1745

AQ2

0.0261

-0.0498

-0.0149

-0.0037

0.0099

0.0968

LNBGS

0.9170

0.7692

1.0986

1.6094

2.4849

LNTA

12.2340

2.1814

7.4972

10.7087

12.2428

13.6899

17.3259

BIG4

0.7686

0.4217

TENURE

1.8410

0.9115

1.3863

1.9459

2.4849

3.4965

NAS

0.6698

0.2649

0.5895

0.7543

0.8523

0.9355

INDSPEC

0.4405

0.4965

CHGSALE

0.1035

0.4827

-0.8534

-0.0176

0.0613

0.1937

1.4570

BTM

0.5736

0.6060

0.2245

0.4249

0.7220

3.6553

LOSS

0.4044

0.4908

LEV

0.5273

0.4565

0.0481

0.2704

0.4612

0.6519

2.2089

ZMIJ

-1.2567

2.0870

-4.5345

-2.7215

-1.5831

-0.3322

5.0000

ISSUE

0.4948

0.4943

CFO

0.0300

0.2427

-0.8780

-0.0140

0.0722

0.1399

0.4369

LAGACCR

-0.0976

0.2977

-0.8618

-0.1315

-0.0698

-0.0248

0.3348

CONCENT

0.1738

0.0964

0.0702

0.1138

0.1563

0.2240

0.5000

OPCYCLE

4.7492

0.6902

2.3195

4.4296

4.8192

5.1642

6.1768

STD_CFO

0.0734

0.0692

0.0080

0.0323

0.0554

0.0899

0.3412

STD_SALE

0.2015

0.1985

0.0166

0.0853

0.1485

0.2550

0.9261

Here ***, **, and * denote p-values of less than 1%, 5%, and 10%, with two-tailed tests, respectively.

a The sample size is 9,996 for │DA 1│ and │DA 2│and 5,408 firm–year observations for AQ1 and AQ2.

b The sample size is 2,443 for │DA 1│ and │DA 2│and 1,232 firm–year observations for AQ1 and AQ2.

c The sample size is 10,428 for │DA 1│ and │DA 2│and 5,687 firm–year observations for AQ1 and AQ2.

d The sample size is 2,011 for │DA 1│ and │DA 2│and 953 firm–year observations for AQ1 and AQ2.

e The sample size is 12,439 for │DA 1│, │DA 2│, and 15 variables from LNBGS to CONCENT, and 6,640 firm–

year observations for AQ1, AQ2, and the other three variables from OPCYCLE to STD_SALE.

Table 3 Correlation Matrix

Panel A: Pearson Correlations between Discretionary Accruals, Auditor Locality, and Control Variables

│DA1│

│DA2│

DMSA

D100

LN-

BGS

LNTA

BIG4

TEN-

URE

NAS

IND-

SPEC

CHG-

SALE

BTM

LOSS

LEV

ZMIJ

ISSUE

CFO

LAG-

ACCR

│DA2│

0.445

(<0.01)

DMSA

-0.071

(<0.01)

-0.056

(<0.01)

D100

-0.076

(<0.01)

-0.056

(<0.01)

0.888

(<0.01)

LN-

BGS

-0.141

(<0.01)

-0.131

(<0.01)

0.061

(<0.01)

0.068

(<0.01)

LNTA

-0.319

(<0.01)

-0.305

(<0.01)

0.127

(<0.01)

0.122

(<0.01)

0.362

(<0.01)

BIG4

-0.212

(<0.01)

-0.195

(<0.01)

0.163

(<0.01)

0.173

(<0.01)

0.197

(<0.01)

0.547

(<0.01)

TEN-

URE

-0.098

(<0.01)

-0.077

(<0.01)

0.045

(<0.01)

0.042

(<0.01)

0.116

(<0.01)

0.253

(<0.01)

0.287

(<0.01)

NAS

-0.144

(<0.01)

-0.148

(<0.01)

0.097

(<0.01)

0.099

(<0.01)

0.222

(<0.01)

0.436

(<0.01)

0.357

(<0.01)

0.224

(<0.01)

INDS-

PEC

-0.118

(<0.01)

-0.106

(<0.01)

-0.020

(0.02)

-0.002

(0.798)

0.092

(<0.01)

0.318

(<0.01)

0.304

(<0.01)

0.132

(<0.01)

0.157

(<0.01)

CHG-

SALE

0.028

(<0.01)

0.002

(0.870)

0.017

(0.06)

0.013

(0.162)

0.001

(0.992)

0.064

(<0.01)

0.012

(0.19)

0.010

(0.27)

0.026

(<0.01)

0.018

(0.05)

BTM

-0.083

(<0.01)

0.138

(<0.01)

-0.035

(<0.01)

-0.018

(0.042)

-0.009

(0.295)

-0.065

(<0.01)

-0.042

(<0.01)

-0.034

(<0.01)

-0.021

(0.02)

-0.012

(0.19)

-0.124

(<0.01)

LOSS

0.256

(<0.01)

0.198

(<0.01)

-0.063

(<0.01)

-0.069

(<0.01)

-0.135

(<0.01)

-0.360

(<0.01)

-0.171

(<0.01)

-0.127

(<0.01)

-0.168

(<0.01)

-0.134

(<0.01)

-0.178

(<0.01)

0.086

(<0.01)

LEV

0.151

(<0.01)

0.079

(<0.01)

-0.071

(<0.01)

-0.083

(<0.01)

-0.024

(<0.01)

-0.068

(<0.01)

-0.131

(<0.01)

-0.038

(<0.01)

-0.061

(<0.01)

0.012

(0.19)

-0.085

(<0.01)

-0.187

(<0.01)

0.153

(<0.01)

(continued on next page)

ZMIJ

0.238

<(0.01)

-0.138

(<0.01)

-0.090

(<0.01)

-0.104

(<0.01)

-0.076

(<0.01)

-0.142

(<0.01)

-0.147

(<0.01)

-0.078

(<0.01)

-0.098

(<0.01)

-0.013

(0.16)

-0.128

(<0.01)

-0.181

(<0.01)

0.433

(<0.01)

0.761

(<0.01)

ISSUE

0.081

(<0.01)

0.053

(<0.01)

-0.049

(<0.01)

-0.053

(<0.01)

-0.056

(<0.01)

-0.015

(0.09)

-0.033

(<0.01)

-0.030

(<0.01)

-0.031

(<0.01)

0.003

(0.73)

0.093

(<0.01)

-0.134

(<0.01)

0.099

(<0.01)

0.151

(<0.01)

0.244

(<0.01)

CFO

-0.273

(<0.01)

-0.072

(<0.01)

0.061

(<0.01)

0.068

(<0.01)

0.163

(<0.01)

0.369

(<0.01)

0.179

(<0.01)

0.083

(<0.01)

0.191

(<0.01)

0.113

(<0.01)

0.124

(<0.01)

0.043

(<0.01)

-0.502

(<0.01)

-0.119

(<0.01)

-0.421

(<0.01)

-0.189

(<0.01)

LAG-

ACCR

-0.122

(<0.01)

-0.216

(<0.01)

0.034

(<0.01)

0.035

(<0.01)

0.053

(<0.01)

0.101

(<0.01)

0.070

(<0.01)

0.043

(<0.01)

0.062

(<0.01)

0.041

(<0.01)

0.035

(<0.01)

0.051

(<0.01)

-0.139

(<0.01)

-0.163

(<0.01)

-0.078

(<0.01)

-0.078

(<0.01)

0.175

(<0.01)

CON-

SERV

-0.054

(<0.01)

-0.082

(<0.01)

-0.031

(<0.01)

-0.014

(0.117)

0.026

(<0.01)

0.046

(<0.01)

0.078

(<0.01)

0.040

(<0.01)

0.085

(<0.01)

0.243

(<0.01)

-0.012

(0.170)

0.041

(<0.01)

-0.049

(<0.01)

-0.041

(<0.01)

-0.049

(<0.01)

-0.021

(0.020)

0.029

(0.001)

0.015

(0.087)

Panel B: Pearson Correlations between Accrual Quality, Auditor Locality, and Control Variables

AQ1

AQ2

DMSA

D100

LNBGS

LNTA

BIG4

OPCYCLE

STD_CFO

STD_SALE

NAS

INDSPEC

ZMIJ

AQ2

0.779

(<0.01)

DMSA

-0.054

(<0.01)

-0.030

(0.014)

D100

-0.049

(<0.01)

-0.028

(0.023)

0.858

(<0.01)

LNBGS

-0.131

(<0.01)

0.020

(0.099)

0.033

(<0.01)

0.046

(<0.01)

LNTA

-0.452

(<0.01)

-0.020

(0.104)

0.099

(<0.01)

0.079

(<0.01)

0.347

(<0.01)

BIG4

-0.240

(<0.01)

-0.035

(<0.01)

0.118

(<0.01)

0.132

(<0.01)

0.210

(<0.01)

0.520

(<0.01)

OPCYCLE

0.156

(<0.01)

-0.000

(1.000)

-0.002

(0.861)

0.011

(0.359)

0.161

(<0.01)

-0.140

(<0.01)

-0.062

(<0.01)

(continued on next page)

STD_CFO

0.542

(<0.01)

0.000

(1.000)

-0.026

(0.033)

-0.020

(0.108)

-0.186

(<0.01)

-0.411

(<0.01)

0.184

(<0.01)

0.101

(<0.01)

STD_

SALE

0.317

(<0.01)

0.000

(1.000)

-0.024

(0.054)

-0.037

(<0.01)

-0.112

(<0.01)

-0.225

(<0.01)

-0.134

(<0.01)

-0.128

(<0.01)

0.344

(<0.01)

NAS

-0.169

(<0.01)

0.022

(0.070)

0.048

(<0.01)

0.044

(<0.01)

0.189

(<0.01)

0.436

(<0.01)

0.341

(<0.01)

0.001

(0.934)

-0.166

(<0.01)

-0.089

(<0.01)

INDSPEC

-0.202

(<0.01)

-0.047

(<0.01)

-0.041

(<0.01)

-0.017

(0.159)

0.112

(<0.01)

0.334

(<0.01)

0.308

(<0.01)

-0.105

(<0.01)

-0.149

(<0.01)

-0.081

(<0.01)

0.190

(<0.01)

ZMIJ

0.114

(<0.01)

0.077

(<0.01)

-0.064

(<0.01)

-0.081

(<0.01)

0.009

(0.486)

0.034

(<0.01)

-0.031

(<0.01)

-0.130

(<0.01)

0.087

(<0.01)

0.074

(<0.01)

-0.002

(0.856)

0.046

(<0.01)

CFO

-0.251

(<0.01)

0.052

(<0.01)

0.040

(<0.01)

0.047

(<0.01)

0.078

(<0.01)

0.312

(<0.01)

0.168

(<0.01)

-0.192

(<0.01)

-0.359

(<0.01)

-0.063

(<0.01)

0.181

(<0.01)

0.124

(<0.01)

-0.411

(<0.01)

Two-tailed p-values are presented in parentheses.

Table 4 Association between Auditor Locality and Clients’ Discretionary Accruals

Section A

Using |DA1| as the Dependent Variable

Section B

Using |DA2| as the Dependent

Variable

Expected

Sign

(1a)

DLOCAL =

DMSA

(2a)

DLOCAL =

DMSA

(3a)

DLOCAL =

D100

(4a)

DLOCAL =

D100

(1b)

DLOCAL =

DMSA

(2b)

DLOCAL =

D100

DLOCAL

-0.0046

(-1.77*)

-0.0119

(-2.75***)

-0.0050

(-1.73*)

-0.0121

(-2.58***)

-0.0090

(-2.26**)

-0.0083

(-1.97**)

LNBGS

-0.0053

(-3.68***)

-0.0121

(-3.72***)

-0.0053

(-3.67***)

-0.0124

(-3.42***)

-0.0096

(-2.98***)

-0.0102

(-2.97***)

DLOCAL*LN

BGS

0.0086

(2.45**)

0.0086

(2.24**)

0.0062

(1.78*)

0.0066

(1.82*)

LNTA

-0.0106

(-14.44***)

-0.0106

(-14.48***)

-0.0106

(-14.46***)

-0.0106

(-14.49***)

-0.0107

(-15.08***)

-0.0107

(-15.11***)

BIG4

-0.0131

(-3.96***)

-0.0129

(-3.90***)

-0.0130

(-3.94***)

-0.0128

(-3.88***)

-0.0120

(-3.78***)

-0.0121

(-3.80***)

TENURE

0.0003

(0.29)

0.0002

(0.16)

0.0003

(0.27)

0.0002

(0.14)

0.0015

(1.26)

0.0015

(1.25)

NAS

0.0064

(1.27)

0.0069

(1.38)

0.0064

(1.27)

0.0068

(1.35)

-0.0017

(-0.36)

-0.0018

(-0.39)

INDSPEC

0.0008

(0.38)

0.0008

(0.36)

0.0009

(0.41)

0.0008

(0.36)

0.0027

(1.24)

0.0028

(1.27)

CHGSALE

0.0200

(4.11***)

0.0201

(4.13***)

0.0200

(4.10***)

0.0200

(4.10***)

0.0085

(2.60***)

0.0085

(2.58***)

BTM

-0.0112

(-5.32***)

-0.0112

(-5.33***)

-0.0111

(-5.30***)

-0.0111

(-5.31***)

-0.0117

(-5.90***)

-0.0117

(-5.87***)

LOSS

0.0155

(5.59***)

0.0156

(5.62***)

0.0155

(5.58***)

0.0155

(5.61***)

0.0121

(4.70***)

0.0120

(4.68***)

LEV

0.0068

(0.87)

0.0067

(0.86)

0.0068

(0.87)

0.0067

(0.86)

0.0063

(1.31)

0.0063

(1.31)

ZMIJ

0.0050

(3.04***)

0.0051

(3.05***)

0.0050

(3.04***)

0.0051

(3.05***)

0.0007

(0.59)

0.0007

(0.60)

ISSUE

0.0002

(0.08)

0.0001

(0.06)

0.0001

(0.02)

0.0001

(0.07)

0.0008

(0.39)

0.0008

(0.41)

CFO

-0.0420

(-3.41***)

-0.0417

(-3.39***)

-0.0419

(-3.41***)

-0.0417

(-3.39***)

-0.0302

(-3.64***)

-0.0301

(-3.64***)

LAGACCR

-0.0162

(-2.53**)

-0.0162

(-2.53**)

-0.0162

(-2.53**)

-0.0162

(-2.53**)

-0.0077

(-1.87*)

-0.0078

(-1.88*)

CONCENT

-0.0247

(-2.55**)

-0.0242

(-2.52**)

-0.0245

(-2.53**)

-0.0238

(-2.47**)

-0.0195

(-2.02**)

-0.0190

(-1.98**)

Intercept

0.2482

(21.61***)

0.2540

(20.90***)

0.2488

(21.55***)

0.2545

(20.72***)

0.2745

(26.69***)

0.2745

(26.39***)

Ind. & Year

Indicators

Included

12,439

0.1886

0.1891

0.1886

0.1891

0.1371

Test for

[

α1

+ (

α3

* 1.0986)] = 0

F = 0.88

F = 0.80

F = 0.66

F = 0.13

Test for

[

α1

+ (

α3

* 1.6094)] = 0

F = 0.34

F = 0.23

F = 0.08

F = 0.42

Test for

[

α1

+ (

α3

* 2.4849)] = 0

F = 2.77

F = 2.25

F = 1.18

F = 1.77

This table reports the results of the regression in Eq. (4). All variables are as defined in Table 1. All reported t-statistics in parentheses are on an adjusted basis,

using standard errors corrected for clustering at the firm level and heteroskedasticity. Here ***, **, and * denote p-values of less than 1%, 5%, and 10%,

respectively, with F-tests and two-tailed t-tests.

Table 5 Association between Auditor Locality and Client Accrual Quality

Section A

Using AQ1 as the Dependent

Variable

Section B

Using AQ2 as the Dependent

Variable

Expected

Sign

(1a)

DLOCAL =

DMSA

(2a)

DLOCAL =

D100

(1b)

DLOCAL =

DMSA

(2b)

DLOCAL =

D100

DLOCAL

-0.0027

(-1.94*)

-0.0029

(-1.93*)

-0.0031

(-2.27**)

-0.0030

(-2.05**)

LNBGS

-0.0011

(-1.64*)

-0.0020

(-1.67*)

-0.0010

(-1.59)

-0.0010

(-1.63)

DLOCAL*

LNBGS

0.0018

(1.76*)

0.0026

(1.92*)

0.0016

(1.71*)

0.0017

(1.84*)

LNTA

-0.0041

(-17.44***)

-0.0041

(-17.47***)

-0.0006

(-2.54**)

-0.0006

(-2.60***)

BIG4

-0.0009

(-0.83)

-0.0009

(-0.80)

-0.0018

(-1.61)

-0.0017

(-1.60)

OPCYCLE

0.0047

(7.90***)

0.0047

(7.90***)

0.0004

(0.64)

0.0004

(0.63)

STD_CFO

0.1745

(15.78***)

0.1747

(15.81***)

0.0004

(0.04)

0.0005

(0.05)

STD_SALE

0.0227

(8.20***)

0.0226

(8.19***)

-0.0026

(-0.98)

-0.0027

(-1.00)

NAS

0.0023

(1.34)

0.0023

(1.33)

0.0030

(1.77*)

0.0030

(1.76*)

INDSPEC

-0.0025

(-3.78***)

-0.0025

(-3.78***)

-0.0022

(-3.27***)

-0.0021

(-3.23***)

ZMIJ

0.0024

(8.98***)

0.0024

(8.94***)

0.0022

(8.18***)

0.0021

(8.15***)

CFO

0.0112

(3.01***)

0.0112

(3.02***)

0.0222

(6.51***)

0.0223

(6.52***)

Intercept

0.0719

(16.32***)

0.0722

(16.49***)

0.0197

(4.09***)

0.0176

(4.10***)

Ind. & Year

Indicators

Included

6,640

0.4146

0.0430

0.0428

Test for

[

α1

+ (

α3

* 1.0986)] = 0

F = 1.53

F = 1.76

F = 3.30*

F = 2.58

Test for

[

α1

+ (

α3

* 1.6094)] = 0

F = 0.10

F = 0.17

F = 0.73

F = 0.50

Test for

[

α1

+ (

α3

* 2.4849)] = 0

F = 0.36

F = 0.29

F = 0.06

F = 0.09

This table reports the results of the regression in Eq. (5). All variables are as defined in Table 1. All

reported t-statistics in parentheses are on an adjusted basis, using standard errors corrected for clustering

at the firm level and heteroskedasticity. Here ***, **, and * denote p-values less than 1%, 5%, and 10%,

respectively, with F-tests and two-tailed t-tests.

Does auditor-client geographic proximity matter? New evidence from an emerging market

Article

Full-text available

Mar 2025
J Account Organ Change

Purpose The purpose of this study is to assess the implications of auditor–client geographic proximity on audit fees, audit report lag and audit quality in the context of an emerging economy, Bangladesh. Design/methodology/approach The auditor–client proximity is gauged in kilometers and travel time, consistent with prior research to assess its association with audit fees, audit report lag and audit quality. Analyzing a data set of 469 firm-year observations from 2018 to 2021 through panel regression, the results are then interpreted in accordance with cluster theory and transaction cost theory. Findings The findings affirm a significant positive association of auditor–client proximity with audit fees and audit report lag. Distant auditors charge lower fees and maintain the timeliness of audit reports to capture and retain distant clients. In addition, the study uncovers a negative association between proximity and audit quality. Geographic proximity can create a familiarity threat between the management team of the client and the local auditor, which can decrease audit quality. These associations are more pronounced in low-risk clients than the high-risk ones. Practical implications These findings underscore the intricate interplay between geographic proximity, communication hurdles and their effects on diverse facets of the audit process that both auditor and client should consider in future audit engagement. Originality/value This research criticizes the existing literature linking auditor–client proximity with audit quality, fees and report lags and provides novel insight from an emerging economy context.

Organizational resilience of audit firms – evidence from the outbreak of the COVID-19

Article

Jan 2024
J Appl Account Res

Purpose This paper examines the organizational resilience of audit firms during the early stages of COVID-19. The unexpected restrictions placed on travel and on-site working created unanticipated barriers for auditors in Hong Kong. The authors expect that auditors with greater organizational resilience can respond to unexpected situations and restore expected performance levels relatively quickly. Design/methodology/approach The authors utilize a sample of 1,008 companies listed on Hong Kong Stock Exchange (HKEX) with a financial year-end of December 31. The authors identify five proxies contributing to organizational resilience: auditor size, industry specialization, diversity, geographic proximity to the client and auditing a new client. The authors use audit report timeliness as this study's main dependent variable. Findings This study's full-sample results suggest that larger auditors, industry specialists and auditors with closer relationships to clients issued more timely audit reports during the pandemic. The analysis of a subsample of companies that initially published unaudited financial statements reveals that industry expertise and longer auditor-client relationships significantly reduced the need for year-end audit adjustments. Finally, the authors find that larger auditors were more likely to offload clients, whereas industry specialists were more likely to retain clients. Research limitations/implications The results of the paper suggests that audit firm characteristics associated cognitive abilities, behavioral characteristics and contextual conditions are associated with audit firm organizational resilience and, consequently, helps auditors respond unexpected changes in the audit environment. Practical implications The findings of the paper are informative for those involved in audit firm management or auditor hiring and retention decisions. Originality/value This study is the first to link organizational resilience to the performance of audit firms in a time of unexpected events. The authors connect three auditor and two auditor-client dimensions to the organizational resilience of the audit firms.

Predicting the Effects of the Financial Crisis, Public Sector Accounting, Corporate Governance Mechanism and Audit Quality on the Financial Performance of Companies Listed on the Tehran Stock Exchange: The Performance of Neural Networks and Machine Learning

Article

Jan 2025

Suhailah Ahmed Hussein

Gambling Preference and Audit Decision-Making—From the Perspective of Key Audit Matters Disclosure

Article

Full-text available

Jan 2025
J BUS ETHICS

Using a novel Chinese gambler conviction database to proxy the local gambling preference, we examine the impact of gambling preference on the disclosure of key audit matters (KAMs). Our findings suggest that the number of KAMs is significantly greater for firms in cities with a strong gambling preference than for firms in cities with a weak gambling preference. Our results are consistent with the view that firms located in areas with a sin culture may improve the quality of their financial reports to overcome their stigmatized image. Additional analysis suggests that large audit firms, shorter audit terms, and closer audit-client distance strengthen the positive relationship between the gambling preference and the disclosure of KAMs. Namely, audit firms with a good reputation, strong independence, and information advantages play a more significant role in “destigmatization”. Finally, we show that auditors charge more audit fee and exert more audit effort towards clients located in areas with a strong gambling preference. This finding suggests that attempts by firms that are affected by a sin culture to achieve “destigmatization” by improving the quality of their financial reports are also costly.

The Geographic Proximity Among Firms, Auditor Office and SEC office and Audit Fees

Article

Nov 2024

This paper investigates the geographic distances among the auditors, audit clients, and the regulatory agency Securities & Exchange Commission (SEC) and their impact on audit quality. Consistent with the exacerbated information asymmetry issue caused by distance, we find the audit quality is negatively associated with the distance between the client/auditor and the SEC offices. We further find that auditors charge higher fees when they are closer to the SEC office, irrespective of the client’s distance to the SEC. This suggests that the distance between auditor and the SEC has a more prominent influence than the client’s location on the audit quality.

Does audit committee heterogeneity matter in auditor selection?

Article

Feb 2025

Purpose Although various stakeholder groups frequently advocate and call for greater heterogeneity among directors and managers, it remains unknown whether team heterogeneity can be beneficial for audit committee to exercise the auditor selection functions. This study aims to address this question. Design/methodology/approach Drawing on a sample of domestically listed nonfinancial A-share firms in China from 2008 to 2022, the authors empirically examine whether and how firm’s audit committee heterogeneity associates with the selection of auditors. Findings Firms with higher levels of audit committee heterogeneity are more likely to be associated with lower-quality auditors. Further examination reveals the mediating role of risk-taking: higher levels of heterogeneity are associated with higher levels of risk-taking, influencing firms to employ lower-quality auditors. Moreover, the authors document that increased audit committee heterogeneity is associated with more audit committee meetings and lower audit efficiency, and that hiring lower-quality auditors can influence the market value of firms with high audit committee heterogeneity. Originality/value To the best of the authors’ knowledge, this study is the first to examine whether and how audit committee erogeneity associates with the selection of auditors. Moreover, because China is a high-power distance, collectivism-oriented, more relations-based (i.e. guanxi-based) than rules-based society, it is critical to examine the influence of team heterogeneity based on the unique cultural context and transitional nature of China’s business environment.

Geographical distance and stock price synchronization: evidence from China

Article

Full-text available

Mar 2025

The effects of geographic factors on information dissemination among investors have been extensively studied; however, the relationship between the geographical distance and stock price synchronization remains unclear. Grounded in information asymmetry theory, this study investigates the impact of geographical distance on stock price synchronization in the Chinese stock market. Using the data from the Shanghai and Shenzhen Stock Exchanges, we find that a greater geographical distance between mutual funds and firms considerably increases stock price synchronization, highlighting a strong positive relationship. Additional analysis show that firms in the regions with better external and internal governance, benefit more from reduced information asymmetry, than those in less regulated or transparent regions. These results have key implications for institutional investors and policymakers aiming to enhance information dissemination and market integration in China.

Does options trading affect audit pricing?

Article

Aug 2024
J BUS FINAN ACCOUNT

We examine the impact of options trading on audit pricing for a sample of US firms over the period from 2004 to 2021. We find that options trading is significantly and negatively related to audit fees, indicating that firms characterized by higher options trading incur lower audit fees. Auditors spend a lower number of days auditing firms with higher options trading and firms with higher options trading experience lower probabilities of lawsuits, and misstatements, and lower likelihood of material weaknesses and auditor opinion on internal controls. The impact of options trading on audit fees is stronger when the auditor is located further away from the audited firm, for firms with non‐specialized auditors, higher information asymmetry problems, poorer earnings and lower governance quality. Overall, our findings underscore the significance of options trading in improving a firm's information environment and reducing litigation risk, resulting in lower audit fees.

Political Incentives and Pollution Reduction in China: Evidence from Firm-level Emissions Data

Article

Dec 2024
ECON MODEL

Use of videoconferencing in audit committee–auditor communication during the COVID-19 pandemic: evidence from Korea

Article

Dec 2024

Purpose The purpose of this study is to examine whether the use of videoconferencing for communication between the audit committee (AC) and auditors affects the quality of client firms’ audits. Design/methodology/approach This paper analyzes the mandatory disclosure information on AC–auditors communication using 1,065 Korean listed firm-years for the fiscal years 2020 and 2021. The details of AC–auditor communication (i.e. the extent of firms’ use of videoconferencing) are manually collected from audit reports. Findings This study finds that videoconferencing has a negative impact on audit quality, suggesting that it is not an effective communication medium between AC and auditors. The results are robust to alternative research designs (e.g. entropy-balanced sample, propensity score matching analysis and change analysis) that address endogeneity concerns. This study also finds that while the negative effect of videoconferencing is mitigated by holding more frequent AC meetings, neither AC independence nor expertise mitigates this effect. Research limitations/implications This paper suggests that videoconferencing may affect audit quality by hurting the discussion between the AC and auditors. Practical implications The findings that videoconferencing impairs the effectiveness of ACs and thus lowers audit quality have practical implications as the COVID-19 pandemic has significantly changed how AC members and auditors interact. This study offers timely and valuable insights into the potential implications of these pandemic-induced changes on audit environments. Originality/value This study provides large-sample empirical evidence that directly examines the effect of videoconferencing on audit quality, enhancing the understanding of the communication dynamics between the AC and auditors. This study also contributes to the literature on the role of ACs in emerging markets by highlighting the information processing role of the AC.

Do Non-audit Service Fees Impair Auditor Independence? Evidence from Going-concern Audit Opinions

Article

Jan 2002

Local Institutional Investors, Information Asymmetries, and Equity Returns

Article

Jan 2009

Auditor conservatism, asymmetric monitoring and earnings management

Article

Jan 2003
CONTEMP ACCOUNT RES

Audit quality attributes: The perceptions of audit partners, preparers, and financial statement users

Article

Jan 1992
AUDITING-J PRACT TH

Earnings manipulation risk, corporate governance risk, and auditors' planning and pricing decisions

Article

Apr 2004

This paper investigates auditors' assessments of earnings manipulation risk and corporate governance risk, and their planning and pricing decisions in the presence of these identified risks. To conduct this investigation, we use engagement partners' assessments of their existing clients made during the participating public accounting firm's client continuance risk assessment process. We find that auditors plan increased effort and billing rates for clients with earnings manipulation risk, and that the positive relationships between earnings manipulation risk and both effort and billing rates are greater for clients that also have heightened corporate governance risk. These findings provide evidence that auditors assess situations involving both an aggressive management and inadequate corporate governance, and that there is a relationship between those assessments and auditors' planning and pricing decisions.

The future of accounting, part III: Reliability and auditor independence

Article

Dec 1996
Account Horiz

S.M.H. Wallman

In this third Commentary on the Future of Accounting and Financial Reporting, the author examines auditor independence. Acknowledging the importance of auditor independence to the reliability of accounting information, the author argues that the current paradigm for assessing independence issues needs to be made both more comprehensive and more comprehensible in order to advance the public interest. He asserts that the current approach is both under and over inclusive because it does not adequately address independence concerns related to traditionally accepted audit and related activities while it precludes, without good reason, various nontraditional services, including services that might advance the public interest. The author advocates an alternative model for assessing independence issues that focuses on auditor dependence. Under the proposed approach, the focus would be on (1) the individual, office, or other unit of the firm making audit decisions, (2) whether a relationship or activity has the potential to create a dependency that could bias auditing judgments, (3) whether the potential for dependence is such that independence can nevertheless be maintained through reasonable measures (in which case, such measures would be required and the activity permitted) and, if it cannot be, (4) whether the benefits to the audit process of the relationship or activity outweigh the potential independence concerns. Compared to the current approach, this alternative model is more clearly articulated, more comprehensive and easier to instill as a culture in a firm. Moreover, in an era of rapid change in the business world, the alternative should result in more reliable accounting information. It will also allow firms to provide their clients with cost-effective services that do not truly raise independence concerns. In sum, this new approach would effect a change in thinking about independence issues from what is allegedly consistent with a public duty, to what is in the public interest.

The Geography of Investment: Informed Trading and Asset Prices

Article

Aug 2001

Applying a geographic lens to mutual fund performance, this study finds that fund managers earn substantial abnormal returns in nearby investments. These returns are particularly strong among funds that are small and old, focus on few holdings, and operate out of remote areas. Furthermore, we find that while the average fund exhibits only a modest bias toward local stocks, certain funds strongly bias their holdings locally and exhibit even greater local performance. Finally, we demonstrate that the extent to which a firm is held by nearby investors is positively related to its future expected return. Our results suggest that investors trade local securities at an informational advantage and point toward a link between such trading and asset prices.

The Quality of Accruals and Earnings: The Role of Accrual Estimation Errors

Article

Jan 2002

This paper suggests a new measure of one aspect of the quality of working capital accruals and earnings. One role of accruals is to shift or adjust the recognition of cash flows over time so that the adjusted numbers (earnings) better measure firm performance. However, accruals require assumptions and estimates of future cash flows. We argue that the quality of accruals and earnings is decreasing in the magnitude of estimation error in accruals. We derive an empirical measure of accrual quality as the residuals from firm-specific regressions of changes in working capital on past, present, and future operating cash flows. We document that observable firm characteristics can be used as instruments for accrual quality (e.g., volatility of accruals and volatility of earnings). Finally, we show that our measure of accrual quality is positively related to earnings persistence.

Auditor Conservatism, Asymmetric Monitoring, and Earnings Management

Article

Jun 2003
CONTEMP ACCOUNT RES

In this paper, we investigate whether, and how, audit effectiveness differentiation between Big 6 and non-Big 6 auditors is influenced by a conflict or convergence of reporting incentives faced by corporate managers and external auditors. In so doing, we incorporate into our analysis the possibility that managers self-select both external auditors and discretionary accruals, using the two stage “treatment effects” model. Our results show that only when managers have incentives to prefer income-increasing accrual choices are Big 6 auditors more effective than non-Big 6 auditors in deterring/monitoring opportunistic earnings management. Contrary to conventional wisdom, we find Big 6 auditors are less effective than non-Big 6 auditors when both managers and auditors have incentives to prefer income-decreasing accrual choices and thus no conflict of reporting incentives exists between the two parties. The above findings are robust to different proxies for opportunistic earnings management and different proxies for the direction of earnings management incentives.

The Differential Effects of Auditors' Nonaudit and Audit Fees on Accrual Quality

Article

Jun 2007
CONTEMP ACCOUNT RES

This paper examines linkages between audit and nonaudit fees and accrual quality. We measure accrual quality by the Francis, Lafond, Olsson, and Schipper 2005 modification of the Dechow and Dichev 2002 measure. We posit that in settings where audit quality is compromised by a loss of auditor independence, managers use accruals more opportunistically and thereby drive down the accrual quality. Conversely, higher audit effort and quality translate to better accrual quality. Our dependent variables are the relative magnitude of nonaudit fees to audit fees and the absolute magnitudes of audit, nonaudit, and total fees. Results show that accrual quality has a significant negative association with the magnitude of nonaudit fees and a significant positive association with audit fees. This latter result is consistent with the proposition that higher audit fees reflect higher audit effort and better judgements about the propriety of accruals, but is not consistent with the proposition that audit fees are associated with economic bonding.

Geographic Proximity between Auditor and Client: How Does It Impact Audit Quality?

Abstract

Recommended publications

A Low-Cost Methodology for Correcting the Distressed Sales Bias in a Downward Spiraling Housing Mark...

Migration aspects of the urbanization in post-socialism: Bratislava case

The layered city: pedestrian networks in downtown Santiago and their impact on urban vitality

Did the 1998 Merger of Price Waterhouse and Coopers & Lybrand Increase Audit Quality?: A Firm- and O...