hossain - 2013 - effect of regulatory changes on auditor independence and audit quality

The results show a significant and positive association between auditor-provided non-audit services NAS fees and the propensity to issue a going-concern opinion for a financially distres

Effect of Regulatory Changes

on Auditor Independence and

Audit Quality

Sarowar Hossain

Australian School of Business, University of New South Wales, Australia

This study investigates the impact of CLERP 9 on auditor independence and audit quality Audit quality is measured by performance-adjusted discretionary accruals and the auditor’s propensity to issue a going-concern opinion for a financially distressed company The results show a significant and positive association between auditor-provided non-audit services (NAS) fees and the propensity to issue a going-concern opinion for a financially distressed company post-CLERP 9, but an insignificant association pre-CLERP 9 The results show

a significant and positive association between NAS fees and the performance-adjusted absolute value of discretionary accruals pre-CLERP 9, but this significant association was mitigated by the introduction of CLERP 9 To address the potential impact of the level of abnormal NAS fees on auditor independence, the analysis is extended by incorporating predictions of abnormal NAS fees Abnormal NAS fees are significantly and negatively associated with the propensity to issue a going-concern opinion for a financially distressed company and positively associated with discretionary accruals pre-CLERP 9, but they are not significant post-CLERP 9 The results provide evidence of improved audit quality after the implementation of CLERP 9 The results will be useful to regulators for the justification of regulatory changes The findings provide evidence of the effectiveness of regulatory changes, specifically the CLERP 9, on the improvement of auditor independence and audit quality.

Key words: Going-concern, audit opinion, abnormal non-audit

fees, audit quality, discretionary accruals, non-audit fees, CLERP

9, auditor independence, regulatory changes

SUMMARY This study investigates the effectiveness of regulatory changes, specifically the Corporate Law Economic Reform Program (CLERP) Act 2004 (Cth), on auditor independence and audit quality

Correspondence to: Sarowar Hossain, Australian School of

Business, University of New South Wales, NSW 2052, Australia.

Email: s.hossain@unsw.edu.au

ISSN 1090-6738

© 2013 Blackwell Publishing Ltd

The first measure of audit quality is the propensity

of auditors to issue a going-concern opinion for

a financially distressed company, and the second

is the performance-adjusted absolute value of

discretionary accruals Auditor independence is

measured by the auditor-provided non-audit

services (NAS) fees and abnormal NAS fees The

study uses Australian company data for the period

2002-2007

The results show a significant and positive

association between NAS fees and the propensity

to issue a going-concern opinion for a financially

distressed company post-CLERP 9, but this

association is not significant pre-CLERP 9 NAS

fees are significantly and positively associated

with the performance-adjusted absolute value

of discretionary accruals pre-CLERP 9, but are

not significant post-CLERP 9 To address the

potential impact of fee composition on auditor

independence, the analysis is extended by

incorporating predictions of abnormal NAS fees

Abnormal NAS fees are derived from a fee

estimation model drawn from prior research

negatively associated with the propensity to issue

a going-concern opinion for a financially distressed

company and positively associated with the

absolute value of discretionary accruals pre-CLERP

9, but are not significant post-CLERP 9 The

findings provide evidence of improved audit

quality after the implementation of CLERP 9

The results will be useful to regulators for

justifying regulatory changes The findings provide

evidence of the effectiveness of regulatory changes,

specifically the CLERP 9, on the improvement of

auditor independence and audit quality

INTRODUCTION

This study examines the effectiveness of regulatory

changes, specifically the Corporate Law Economic

Reform Program Act 2004 (Cth) (hereafter CLERP 9)

on auditor independence and audit quality

Auditor-provided non-audit services have been one

of the most debated issues for the last several

decades, with arguments that abnormally high fees

for auditor-provided non-audit services may

compromise auditor independence (DeAngelo,

1981; Teng, 2006) Hoitash, Markelevich and

Barragato (2007) argue that any association

between the fees paid to auditors and audit quality

is an important input to the ongoing debate on

how the accounting profession should be organized

and monitored Following prior studies concerned with auditor independence, this study uses auditor-provided non-audit services (NAS) fees and abnormal NAS fees as measures of auditor independence (e.g., Craswell, 1999; DeFond, Raghunandan & Subramanyam, 2002; Frankel, Johnson & Nelson, 2002; Ashbaugh, LaFond & Mayhew, 2003; Chung & Kallapur, 2003; Kinney,

Palmrose & Scholz, 2004; Cahan et al., 2008) and

investigates the extent to which those fees are associated with audit quality in pre- and post-CLERP 9 periods Consistent with prior studies (e.g., Carey & Simnett, 2006; Choi, Kim & Zang, 2006; Chen, Sun & Wu, 2010; Chi, Douthett & Lei, 2010), this study uses the propensity to issue a going-concern opinion for financially distressed companies and the absolute value of discretionary accruals as proxies for audit quality

Corporate governance failures have reduced investor confidence and raised questions regarding the integrity of the information provided to investors (Jain & Rezaee, 2003; Jain, Kim & Rezaee, 2003; Cohen, Dey & Lys, 2005) The highly publicized failures triggered the passage of the Sarbanes Oxley Act (SOX) on 30 July 2002 in the US and the CLERP 9 Act (2004) on 25 June 2004 in Australia Although CLERP 9 introduced sweeping changes, the implications of the reforms are yet

to be ascertained Examination of the impact of the CLERP 9 on NAS fees and their associations with audit quality will provide evidence of the effectiveness of regulatory changes

Effective for the reporting period commencing

on or after 1 July 2004, the CLERP 9 reforms require the mandatory disclosure of specific categories of NAS fees, although the NAS fees have been disclosed in the annual reports for a long period of time for Australian Securities Exchange (ASX) listed companies The Corporations Act 2001 (s300) requires that the audit committee or, in the absence

of an audit committee, the board of directors is required to formally attest to their satisfaction that auditor independence was not compromised by the provision of NAS In annual reports, in accordance with the requirements of s307C of the Corporations Act 2001, the lead audit partner must state that there were: (a) no contraventions of the auditor independence requirements of the Corporations Act 2001; and (b) no contraventions of any applicable code of professional conduct in relation

to the audit CLERP 9 also mandated the rotation

of both the lead and review audit partners every five years due to concerns about the familiarity

between the auditor and audit clients and their

effects on auditor independence There is presently

a lack of empirical evidence on the impact of

CLERP 9 on audit quality Therefore, this study

seeks to provide such evidence

The results show a significant and positive

association between NAS fees and the propensity

to issue a going-concern opinion for a financially

distressed company post-CLERP 9, but this

association is not significant pre-CLERP 9 The

results indicate that auditors are more likely to issue

a going-concern opinion for a financially distressed

company post-CLERP 9 The results also show a

significant and negative association between

abnormal NAS fees and the propensity to issue a

going-concern opinion for a financially distressed

company pre-CLERP 9, but this association is not

significant post-CLERP 9 Further, the association

between positive abnormal NAS fees and the

propensity to issue a going-concern opinion for a

financially distressed company is stronger

pre-CLERP 9 than post-pre-CLERP 9 The results indicate

that auditors were less likely to issue a

going-concern opinion for a financially distressed

company pre-CLERP 9 than post-CLERP 9 when

auditors earn abnormally high NAS fees

The association between NAS fees and the

discretionary accruals is significant and positive

pre-CLERP 9, but this relationship is not significant

post-CLERP 9 The results indicate that auditors

tolerated a larger magnitude of accruals when

they earned higher NAS fees pre-CLERP 9, but

that significant association was mitigated by the

introduction of CLERP 9 Furthermore, abnormal

NAS fees are significantly and positively associated

with the absolute value of discretionary accruals

pre-CLERP 9, but this association is not significant

post-CLERP 9 In addition, the association between

positive abnormal NAS fees and the absolute value

of discretionary accruals is stronger pre-CLERP 9 as

compared to post-CLERP 9 Based on the results of

multiple proxies of auditor independence and audit

quality, the results provide evidence of improved

audit quality after the implementation of CLERP 9

The results documented in this study make

significant contributions to the literature Similar

studies have been conducted in the US; however,

the US legislative response (i.e., SOX) is quite

different from that in Australia Thus, the current

paper address the question: Is the Australian

approach effective? This is an important question

for policy making to achieve an outcome with the

least compliance cost This study provides evidence

as to whether there is an observable impact from the CLERP 9 reforms This has not been examined

in prior Australian studies; therefore, the paper provides new evidence on the success of the CLERP 9 approach to addressing problems with audit practice

The remainder of the paper is organized as follows: The next section discusses the background

of the regulatory changes, followed by presentation

of the hypotheses The fourth section describes the research method and provides information about the sample The findings are reported in the fifth section, and this is followed by conclusions in the final section

REGULATORY CHANGES During the last decade, corporate collapses throughout the world have received a great deal

of media attention (Fargher & Jiang, 2006), questioning the adequacy of regulations to improve auditor independence and audit quality Following these corporate failures, the adequacy of the Australian regulations that aimed to ensure auditor independence were questioned, which resulted in a review of the independence of Australian company auditors (Fargher & Jiang, 2006) This review resulted in the publication of the Ramsay Report

in October 2001 (Ramsay, 2001), and subsequently the CLERP 9 draft was proposed in September

2002 In the US, corporate legislation, SOX (2002), was passed in response to a series of financial scandals (Chambers & Payne, 2008) The debates in the Ramsay Report (2001) and the CLERP 9 proposal focused on the joint provision of audit and non-audit services (Fargher & Jiang, 2006) To improve auditor independence and audit quality in the joint provision of audit and non-audit services, the Joint Committee of Public Accounts and Audits (JCPAA, 2002) and the ASX Corporate Governance Council issued reports on best practice guidelines for listed companies (ASX, 2003) Australian reporting entities adopted Australian equivalents

of International Financial Reporting Standards (IFRS) for the reporting period commencing on or after 1 January 2005 In annual reports, in accordance with the requirements of s307C of the Corporations Act 2001, the lead audit partner has

to declare whether there are: (a) any contraventions

of the auditor independence requirements of the Corporations Act 2001; and (b) any contraventions

of any applicable code of professional conduct in

relation to the audit

CLERP 9 does not ban any particular NAS

provided by the auditor However, this Act stresses

that the auditor must identify and evaluate threats to

independence and apply safeguards to reduce any

threats to an acceptable level (Ramsay, 2001) Where

the provision of NAS to a client poses a threat that

cannot be reduced to an acceptable level, CLERP 9

prohibits the provision of that service (Ramsay,

2001) As part of the CLERP 9 reforms, the

Corporation Act 2001 was amended to require the

mandatory disclosure of fees paid for the specific

categories of NAS provided in companies’ annual

reports for the reporting period commencing on or

after 1 July 2004 Moreover, the reforms stipulated

that the financial statements must include a

statement by the audit committee (or board in its

absence) that it is satisfied that the provision of

NAS is compatible with auditor independence

(Corporation Act 2001, s300) This disclosure should

include an explanation of why certain problematic

NAS, such as IT systems services and internal audit

services, if contracted, did not compromise auditor

independence (CLERP 9, 2004)

Regulators and legislators throughout the world

apparently presume that providing NAS impairs

auditor independence, which may lead to lower

audit quality (Kinney et al., 2004) After the

implementation of SOX (2002), most NAS were

banned in the US, and SOX also requires prior

approval by a registrant’s independent audit

committee of any NAS allowed by law (Kinney

et al., 2004) Using pre- and-post SOX data, prior

studies (e.g., Hoitash et al., 2007; Li, 2009) find a

significant association between NAS fees and audit

quality during the pre-SOX period and the

association is not significant post-SOX In

Australia, CLERP 9 has not banned NAS; however,

both the auditors and the audit committee

(or board in its absence), must declare that

the provision of NAS has not compromised

auditor independence Therefore, the objective

of the current study is to investigate whether

the implementation of CLERP 9 mitigates the

association between NAS fees and audit quality

DEVELOPMENT OF HYPOTHESES

Association between auditor-provided NAS

fees and audit quality

The objective of passing CLERP 9 was to improve

auditor independence and audit quality CLERP 9

and the Corporations Act 2001 require auditors to sign a declaration in the annual report stating that the independence of the auditor has not been compromised due to NAS The audit committee (or board in its absence) must also declare that the external auditor is working without impaired independence CLERP 9 does not prohibit

mandates disclosure of the types of NAS fees in the annual report It is expected that non-audit fees will decrease due to the additional auditing pronouncements and disclosure of types of NAS and the prospect of greater regulatory reporting requirements by companies following those legislative reforms In addition, after the implementation of SOX in 2002, most NAS fees were banned in the US Big 4 audit firms applying the international ruling can have the same effect in Australia because Big 4 audit firms are international audit firms and may apply the same policy in auditing their Australian clients

Auditor provided non-audit services have long been regarded as a threat to auditor independence

by the regulators both in Australia and overseas (Craswell, 1999) Investigating the effect of the

legal and regulatory changes in China, Chen et al.

(2010) find a negative association between client importance at the audit partner level and modified opinions during the pre-regulatory change period and a positive association during the post-regulatory change period The findings of Chen

et al (2010, p 127) suggest that ‘institutional

improvements prompt auditors to prioritize the cost of compromising quality over economic benefits gained from important clients.’ Reynolds and Francis (2001) find that client importance is positively associated with the issuance of going concern reports for Big N clients Craswell, Stokes and Laughton (2002) argue that if fee dependence affects auditors’ independent judgements, then auditors are less likely to issue qualified audit opinions Using Australian data, Craswell (1999) reports that auditors’ decisions to qualify audit opinions are not affected by the provision of non-audit services Other Australian studies (e.g., Wines, 1994) find a significant and negative association between NAS and qualified opinions and argue that NAS creates perceptions of independence problems In contrast, Barkess and Simnett (1994) do not find a significant association between NAS and qualified opinions Using US data, Li (2009) finds that client importance is not significantly associated with the issuance of

going-concern opinions during the pre-SOX

period, but that there is a positive association

post-SOX Based on the prior discussion, I expect a

significant positive association between NAS fees

and the propensity to issue going-concern opinions

pre-CLERP 9 and no or a significant negative

association post-CLERP 9 Thus, I propose the

following hypothesis:

H1a:There is a positive association between NAS fees

and the propensity to issue a going-concern opinion

for a financially distressed company post-CLERP 9

but not pre-CLERP 9.

Regulators and legislators apparently presume

that auditor-provided non-audit services impair

auditor independence, which leads to lower quality

audits (Kinney et al., 2004) The findings of prior

studies regarding the associations between NAS

fees and earnings management-based audit

quality measures are mixed Frankel et al (2002)

hypothesized that NAS increases the economic

bonding between auditors and clients and

therefore impairs auditor independence Cahan

et al (2008) argue that faster growth and longer

time period over which non-audit services are

purchased from auditors may reduce auditor’s

independence However, they did not find any

support of a relationship between fee growth rates

and length of purchasing non-audit services and

discretionary accruals Studies by Krishnan, Sami

and Zhang (2005) and Francis and Ke (2006) find

lower earnings response coefficients for firms

paying higher NAS fees Ferguson, Seow and

Young (2004) find that client importance is

positively associated with the absolute value of

abnormal accruals Ghosh, Kallapur and Moon

(2009) find that client importance is negatively

associated with earnings response coefficients

Using pre- and-post SOX data, Hoitash et al (2007)

find a significant and positive association between

NAS fees and discretionary accruals pre-SOX and

no significant association post-SOX

In contrast, Reynolds and Francis (2001) find that

client importance is negatively associated with the

absolute value of abnormal accruals The intention

of regulatory changes is to improve the quality of

financial statement audits and audit quality, which

will also influence earnings reliability (Chambers

& Payne, 2008) In Australia, effective after 1 July

2006, lead audit partner rotation was mandated

(Fargher & Jiang, 2008), and this new mechanism of

mandating audit partner rotation has given audit

standards legislative backing (Carey & Simnett,

2006) The reforms also included proportionate liability for auditors and claims capping, which provided auditors with some certainty on liability claims (Fargher & Jiang, 2008) This could well affect auditor behaviour regarding earnings quality measures As reduced earnings management is used as one of the indicators of improved audit quality (Wong, Jubb & Chalmers, 2007) vis-à-vis

financial reporting quality (see Becker et al., 1998)

and audit quality is affected by independence, it follows that if the implementation of CLERP 9 increased auditor independence, it may have consequently improved earnings management-based measures of financial reporting quality Therefore, I expect a significant positive association between NAS fees and the absolute value of performance-adjusted discretionary accruals pre-CLERP 9 and no or a significant negative association post-CLERP 9 Thus, I propose the following hypothesis:

H1b: There is a positive association between NAS fees and the absolute value of discretionary accruals pre-CLERP 9 but not post-CLERP 9.

Association between abnormal NAS fees and audit quality

The objectivity and independence of auditors is more likely to be influenced by the level of client fees in excess of the auditor’s expectation of normal fees, i.e., abnormal fees, rather than expected fees

(Choi et al., 2006) Choi et al (2006) argue that when

an auditor earns more than the expected level of NAS fees, the perceived net benefits become greater than the associated costs, which increases economic bonding and may decrease audit quality Kinney and Libby (2002) argue that strong economic bonding between the auditor and the client will reduce the quality of the reported earnings through the auditor’s reduced willingness

to resist client-induced biases in reported accounting information Auditors may also be reluctant to issue a going-concern opinion due to the possibility of losing the client Using pre- and

post-SOX data, Hoitash et al (2007) find a

significant positive association between abnormal NAS fees and accruals pre-SOX and no significant

association post-SOX Hoitash et al (2007) provide

evidence consistent with the view that clients with higher abnormal NAS fees may have more influence on auditors’ decisions that lead to

impaired auditor independence Choi et al (2006)

suggest that auditors’ incentives to compromise

audit quality differ depending on whether the

client pays more than or less than the normal level

of fees, which, in turn, leads to the audit fee–audit

quality association being based on the sign of

abnormal audit fees Thus, I propose the following

hypotheses:

H2a: There is a negative association between

abnormal NAS fees and the propensity to issue a

going-concern opinion for a financially distressed

company pre-CLERP 9 but not post-CLERP 9.

H2b: There is a positive association between

abnormal NAS fees and the absolute value of

discretionary accruals pre-CLERP 9 but not

post-CLERP 9.

RESEARCH METHOD

Measurement of abnormal NAS fees

Following the procedures of Choi et al (2006) and

Hoitash et al (2007) of decomposing fees into two

components, namely the expected component

(normal fees) and the abnormal component

(abnormal fees), it is necessary to specify an

expectation model linking actual fees with their

determinants On the basis of economic bonds,

client size, client complexity and client-specific

risk from prior studies (e.g Simunic, 1980;

Francis & Stokes, 1986; Craswell et al., 1995;

DeFond et al., 2002; Frankel et al., 2002; Whisenant

et al., 2003; Choi et al., 2006; Hay et al., 2006) and

independence measures reported by Ashbaugh

et al (2003) and Ruddock, Taylor and Taylor

(2006), this study uses the following ordinary

least squares (OLS) model to calculate abnormal

NAS fees for each year separately The abnormal

NAS fees are the residuals of the estimated

model:

LnNAS LnTA BIG EQUITY

MERGACQS ROA LEVERAG

4

E NEG_ROA GROWTH MB

LnSUBS FOROPS USLIST

+

β

113∑INDUSTRY+ε (1)

where

LnNAS = natural log of auditor-provided non-audit

services fees;1

Control variables

LnTA = natural log of total assets;

BIG4 = 1 if the audit firm is a Big 4 firm, 0 otherwise;

EQUITY = 1 if the company issued new shares in the current year, 0 otherwise;

MERGACQS = 1 if the company has merger or acquisition activities during the current year, 0 otherwise;

ROA = earnings before interest and taxes divided

by total assets;

LEVERAGE = total liabilities divided by total assets;

NEG_ROA = 1 if the company reports a negative return on assets in the current year, 0 otherwise; GROWTH = change of assets from prior year;

MB = market-to-book value;

LnSUBS = natural log of number of subsidiaries; FOROPS = 1 if the company has any foreign subsidiaries, 0 otherwise;

USLIST = 1 if the company is cross-listed in the US,

0 otherwise;

∑INDUSTRY = there are 10 indicator variables for

11 Global Industry Classification Standard (GICS) industry groups

Definitions of the variables are provided in the Appendix for ease of reference The natural log of total assets (LnTA) is used as a proxy for client size Demand for non-audit services is likely to increase

with firm size (Palmrose, 1986; Raghunandan et al.,

2003) and, therefore, it is expected that NAS fees are positively associated with LnTA Non-audit services fees are likely to be higher for clients with more complex business operations (Raghunandan

et al., 2003) LnSUBS, MERGACQS and FOROPS

are used to proxy for client complexity, and it is expected that the variables representing client complexity are positively associated with NAS fees NEG_ROA, LEVERAGE and ROA are used to proxy for a client’s risk characteristics Since an auditor charges higher fees for riskier clients (Simunic & Stein, 1996), the coefficients of NEG_ROA and LEVERAGE are expected to be positive and that of ROA to be negative The control variable, BIG4, is used to capture the capacity of providing non-audit services and a positive coefficient is expected with NAS fees The demand for non-audit services is greater for high-growth firms than low-growth firms (Choi & Wong, 2006) EQUITY, GROWTH and MB are used

to capture the effect of a client firm’s growth on NAS fees, and positive associations with fees for EQUITY, GROWTH and a negative association

with MB are expected A control variable for the

US cross-listing (USLIST) is used owing to the

additional requirements in US regulations Similar

to Ashbaugh et al (2003) and Ruddock et al (2006),

this study uses industry dummies to control for

cross-industry differences in fees INDUSTRY is an

indicator variable equal to 1 if the company belongs

to an appropriate industry group, 0 otherwise

Audit opinion model

The following logistic regression model is

estimated for the pre- and-post-CLERP 9 periods to

test the association between NAS fees, abnormal

NAS fees and the propensity to issue a

going-concern opinion for a financially distressed

company after controlling for the variables used by

Carey and Simnett (2006):

OPINION LnFEE PQUAL PBANK

LnTA LnAGE LEVERA

CLEVERAGE ROA LLOSS INVESTMENTS BIG

CF

+

β

12

4

O+β13MINING+ε (2) where

Dependent variable

OPINION = OPINION is coded as 1 if an auditor

issues a going-concern opinion for a financially

distressed company in the current year and as 0

otherwise;

Experimental variable

FEE = alternative measures of fees, such as: LnNAS

is the natural log of auditor-provided non-audit

services fees;2ABNAS fees are estimated from the

residuals of the LnNAS fee model (Equation 1);

Control variables

PQUAL = 1 if the auditor issued an opinion other

than an unqualified one in the previous year, 0

otherwise;

PBANK = probability of bankruptcy, as measured

by adjusted Zmijeswki score;3

LnAGE = natural log of number of years the

company has been listed on the Australian

Securities Exchange (ASX);

CLEVERAGE = change in leverage during the year;

LLOSS = 1 if the client reported a loss in the

previous year, 0 otherwise;

INVESTMENTS = short- and long-term investment

securities (measured as current assets minus

debtors and inventory) divided by total assets;

CFO =operating cash flow deflated by total assets;

MINING = 1 if the company belongs to the mining industry, 0 otherwise;

e = error term

All other variables are as defined in Equation (1) Prior year opinion (PQUAL) is a good predictor

of the current year’s opinion because the going-concern problems may continue for a longer period (Monroe & Teh, 1993) Higher values of

bankruptcy (Carey & Simnett, 2006) LnTA is included because larger companies are less likely

to end up in bankruptcy, due to their financial stability and greater negotiating power (Carey & Simnett, 2006) LnAGE is controlled because younger companies are more likely to encounter financial distress (Carey & Simnett, 2006) Similar

to Carey and Simnett (2006), I use LEVERAGE and CLEVERAGE to control for the risk of insolvency ROA and LLOSS capture the profitability of the companies, as companies incurring a loss are more likely to receive a going-concern opinion (Carey & Simnett, 2006) CFO and INVESTMENTS proxy for the liquidity risk of a company, and a liquidity crisis may end in bankruptcy (Carey & Simnett, 2006) BIG4 is included in order to differentiate the propensity of issuing going-concern opinions

by Big 4 and non-Big 4 audit firms MINING is controlled because the sample includes a large number of mining companies, which have different financial characteristics (Butterworth & Houghton, 1995; Carey & Simnett, 2006)

Earnings management model

Similar to Kothari, Leone and Wasley (2005), this study estimates non-discretionary accruals using the performance-adjusted modified Jones (1991) model, and estimates abnormal accruals as the residuals from the model Subject to a minimum of

10 observations (excluding financial sector 40) in each industry category for each year, this model is estimated cross-sectionally for each two-digit GICS (a six-digit for the Metals & Mining sector) industry group in each of the years 2002–2007 as follows:

PPE LagROA

α β

β2 1 β3 ε

where TACC = total accruals, being the difference between operating income (OI) and cash flow from operations;

DREV = change in revenues from period t - 1 to period t;

DREC = change in accounts receivables from period

t - 1 to period t;

PPE = gross value of property, plants and

equipment;

LagROA = one-year lag of ROA;

e = error terms

All variables, including the intercept (other than

LagROA), are scaled by the lagged total assets

The model for examining the

association between test variables and

earnings management

This study uses the following model to examine the

association between NAS fees, abnormal NAS fees

and the absolute value of performance-adjusted

discretionary accruals for the pre- and post-CLERP

9 periods after controlling for the variables used in

Carey and Simnett (2006):

DACC LnFEE BIG PBANK

OPINION LnTA LEVERAGE

+ββ710 β8+β11 β9+β12 +ε

LLOSS ROA LnAGE

GROWTH CFO MINING

(4) where

Dependent variable

DACC = discretionary accruals estimated from the

residuals of Equation (3);

Experimental variable

FEE = alternative measures of fees, such as: LnNAS

is the natural log of auditor-provided non-audit

services fees; ABNAS fees is estimated from the

residuals of the LnNAS fee model;

Control variables

OPINION = 1 if the company has a modified

opinion during the current year, 0 otherwise;

All other variables are defined in Equations (1) and

(2)

Consistent with the arguments of Carey and

Simnett (2006), a number of variables (BIG4,

PBANK, OPINION, LnTA, LEVERAGE, LLOSS,

ROA, LnAGE, CFO, MINING) used in predicting

the propensity to issue a going-concern opinion are

expected to be associated with the level of

discretionary accruals GROWTH is controlled for

changes in a company’s financial position and their

association with discretionary accruals

Sample selection

CLERP 9 (2004) was enacted on 30 June 2004

Hence, the first reflection of CLERP 9 is seen in

the annual reports prepared for the financial period 2004–05 Therefore, this study uses data from 2002, 2003, 2004 and balance dates between 1 January and 29 June 2005 for the pre-CLERP 9 and 2005 (balance date on 30 June), 2006 and 2007 for the post-CLERP 9 Data were collected up to

2007 because of the Global Financial Crisis in

2008, which might affect the results The initial sample consists of 9,736 firm-year observations whose financial data were available in Aspect’s FinAnalysis database for fiscal years from 2002 to

2007 NAS fees and audit opinions data were hand-collected directly from the annual reports Consistent with Carey and Simnett (2006), the sample for the audit opinion model is restricted

to financially distressed companies only I identified financially distressed companies based

on either the company reporting a loss or a negative cash flow from operations during a given year The sample size is further reduced because of data requirements for predicting the type of audit opinions The final sample consists

of 4,961 firm-year observations for the OPINION model

The sample of discretionary accruals models was derived after excluding financial sector (GICS 40) and data requirements for estimating discretionary accruals The final sample is 6,656 firm-year observations for the DACC model Table 1 shows how the sample sizes are derived for this study Table 2 shows the industry representation of the sample companies The current study uses a two-digit GICS code (a six-two-digit code for Metals & Mining) There are 11 industry groups Similar to prior studies (e.g., Coulton, Taylor & Taylor, 2005), the current study also excludes industry groups for a particular year if the group has fewer than

10 companies when estimating discretionary accruals

Table 1: Sample selection (2002–2007)

Firm-year observations OPINION DACC

Financial (4010 to 4040) – 563 Missing data and data

Non-distressed

DATA ANALYSIS AND DISCUSSION

OPINION

Descriptive statistics

Table 3 shows the descriptive statistics for the

sample The average size of the sample companies

post-CLERP 9 the average size was $1,142.609

million and the difference is significant The

average NAS fees are higher pre-CLERP 9

($128,812.616) than post-CLERP 9 ($67,502.291)

and the difference is significant Other than LLOSS

and PQAUL, mean differences for all the control

variables are significantly different between the

pre-CLERP 9 and the post-CLERP 9 periods The

maximum VIF value is 2.708, indicating that

multicollinearity is not a significant concern

Association between NAS fees and OPINION

Table 4 provides the logistic regression results for

the association between NAS fees and the

propensity to issue a going-concern opinion for a

financially distressed company during the pre- and

post-CLERP 9 Consistent with Carey and Simnett

(2006), the sample is restricted to financially

distressed companies The model is well-fitted with

pseudo R2s ranging from 0.239 to 0.252 The results

(Table 4, columns 2 and 3) show that the association

between the LnNAS fees and the propensity to

issue a going-concern opinion for a financially

distressed company is significant and positive

post-CLERP 9 (coefficient = 0.049, p = 0.006)

(one-tailed) and is not significant pre-CLERP 9 (coefficient = -0.021, p = 0.104). The findings indicate that auditors are more likely to issue a going-concern opinion for a financially distressed company post-CLERP 9 when providing NAS fees This finding is consistent with the US study

by Li (2009), who finds that client importance is not significantly associated with the issuance

of going-concern opinions pre-SOX, but that association is significant and positive post-SOX The significant association between LnNAS fees indicates the improvement of auditor independence and audit quality post-CLERP 9 Auditors earning more than the expected level of NAS fees may compromise audit quality

To test whether the association between abnormal

going-concern opinions remained the same before and after the implementation of CLERP 9, this study estimated abnormal NAS fees based on the NAS fee model (Equation 1) Table 4 (columns

4 and 5) shows that the association between the signed abnormal NAS fees is significantly and negatively associated (coefficient = -0.036,

p = 0.005) with OPINION pre-CLERP 9, but is not significant (coefficient = 0.017, p = 0.196)

post-CLERP 9 A separate analysis (not tabulated) shows that the association between positive abnormal NAS fees and OPINION is significant

and negative (coefficient = -0.090, p = 0.003) and

stronger pre-CLERP 9 than post-CLERP 9

(coefficient = -0.173, p = 0.040) These findings also

indicate that auditor independence and audit

Table 2: Industry representation of the sample companies (2002–2007)

Industry Group (GICS code) OPINION DACC

Firm-year observations Per cent observations Firm-year Per cent

Two-digit GICS industry classification (6-digit for Metals & Mining)

Table