goodwin and seow - 2002 -the influence of cg mechanisms on the quality of financial reporting and auditing in singapore

Accounting and Finance 42 (2002) 195–223
© AAANZ, 2002. Published by Blackwell Publishing.
Blackwell Publishing, LtdOxford, UKACFIAccounting and Finance0810-5391© The Accounting Association of Australia and New Zeland, 2002423November 2002016???ORIGINAL ARTICLEGraphicraft Limited, Hong Kong
The influence of corporate governance mechanisms on the
quality of financial reporting and auditing: Perceptions of
auditors and directors in Singapore
Jenny Goodwin
a
, Jean Lin Seow
b
a
The University of Queensland Business School, Queensland 4072, Australia
b
Nanyang Business School, Nanyang Technological University, Nanyang Avenue,
Singapore, 639798
Abstract
This study uses two hypothetical cases to examine the perceptions of auditors
and directors in Singapore about corporate governance practices relating to the
quality of financial reporting and auditing. In the first case, the strength of the
audit committee, the existence of an internal audit function and the strength
of a corporate code of conduct were manipulated. All three variables were
perceived to have some influence on financial reporting and audit quality.
However, some interesting differences were found between the perceptions of
auditors and directors. Auditors place more weight on the internal audit func-
tion, possibly due to their familiarity with the role that internal audit can play
in reducing audit risk and enhancing controls. Directors have more confidence
in board enforcement of a strong code of conduct, possibly reflecting the view
that this encourages staff to adhere to higher ethical standards. In the second
case, audit partner rotation, outsourcing of internal audit services and whether
the audit firm audited all companies within a group were manipulated. Auditors
believed that their ability to resist management pressure was enhanced when

they audited all companies within the group. No significant differences were
found for the other variables, suggesting that neither group believes that these
practices impair audit independence.
Key words: Corporate governance; Financial reporting quality; Audit quality;
Audit independence
JEL classification: M40; G34
The authors acknowledge the helpful comments of two anonymous reviewers, Margaret
Abernethy, Allen Craswell, Terry O’Keefe, Donald Stokes, and participants at the One-Day
Symposium on Accountability and Performance in the New Millenium, Brisbane, Australia,
February 2000, the 23
rd
Annual Congress of the European Accounting Association, Munich,
Germany, March 2000 and the British Accounting Association Annual Conference 2000, Ex-
eter, England, April 2000.
196 Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223
© AAANZ, 2002
1. Introduction
There has been considerable debate in recent times concerning the need for
strong corporate governance (McConomy and Bujaki, 2000), with countries
around the world drawing up guidelines and codes of practice to strengthen
governance (Cadbury, 1997). The underlying reason for this emphasis lies in
concerns over the integrity of securities markets (Millstein, 1999). Sound
governance by boards of directors is recognised to influence the quality of financial
reporting, which in turn has an important impact on investor confidence (Levitt,
(1998 and 2000a). Strong governance should reduce the adverse effects of
earnings management as well as reduce the likelihood of misstatements arising
from fraud or errors (Beasley, 1996; Dechow et al., 1996; McMullan, 1996).
Traditionally, the external auditor has also played an important role in
improving the credibility of financial information (Mautz and Sharaf, 1961;
Wallace, 1980). A high quality audit is generally regarded as one where the

auditor both discovers misstatements (discovery) and is willing to report those
misstatements (independence) (DeAngelo, 1981a and 1981b; O’Keefe and
Westort, 1992). However, considerable criticism has been directed at the audit
profession for failing in both of these aspects (Knutson, 1994; Rabinowitz,
1996).
The profession has been proactive in attempting to improve audit quality by
issuing standards focused on discovery and independence. For example, in the
United States (US), greater responsibility for the discovery of fraud has been
placed on the auditor by the issue of SAS no. 82. Independence has received the
attention not only of the profession but also regulators and stock exchanges
around the world. Levitt’s (1998 and 2000a) criticisms of the failure of the
auditing profession in the US to curb excessive earnings management, together
with the Security and Exchange Commission’s attempts to regulate audit inde-
pendence have been well publicised. As a result, there has been a concerted
effort to devise ways of enhancing independence (SEC, 2000; Independence
Standards Board, 2000).
The profession has also responded to criticisms on audit quality by emphasis-
ing that, by its nature, the inherent limitations of an audit make it impossible to
eliminate the risk of audit failure (Ricchiute, 1998; IFAC, 1994a). Furthermore,
it is stressed that it is primarily the responsibility of the client to both prevent
and detect the occurrence of fraud, errors or other irregularities in the financial
statements (IFAC, 1994b). This responsibility can be met by putting into place
a strong system of corporate governance (Bishop et al., 2000; Rabinowitz, 1996).
In recognition of this, regulators and professional bodies have imposed require-
ments or made recommendations concerning board structure, audit committees
and internal controls (Blue Ribbon Committee, 1999; Committee on Corporate
Governance, 2001).
The effect of sound governance practices on the quality of financial reporting
has recently received attention from researchers, particularly in the US (McMullen,
Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223 197

© AAANZ, 2002
1996; Beasley, 1996; Beasley et al., 2000; Abbott et al., 2000). The main focus of
these studies has been the relation between audit committees and fraudulent
financial reporting, with results generally supporting a negative relation
between an active audit committee and the likelihood of a company being cited
for fraudulent reporting.
While these results provide evidence from a strong and sophisticated capital
market environment, very little research has been conducted in countries where
capital markets are less developed and where governance mechanisms are still
evolving. However, sound corporate governance practices are equally, if not
more, important in countries that are attempting to gain credibility among
global investors. This is particularly so in South East Asia as the region attempts
to regain investor confidence following the financial crisis of the late 1990s.
This paper addresses the scarcity of research in this environment by explor-
ing governance issues relating to the quality of financial reporting and auditing
in Singapore. In spite of its size, Singapore plays an important role in setting
standards of corporate governance in the South East Asia/Oceania region. Not
only was it one of the first countries to legislate that all listed companies must
have an audit committee, but as part of a government-led drive to become the
dominant regional financial centre, there has been considerable emphasis on the
need for sound governance (SES, 1996b; Ng, 1997; Goodwin and Seow, 1998
and 2000). This study thus enables investigation into the perceived incremental
effectiveness of some of the governance mechanisms proposed by various
global regimes within a regulatory framework that mandates a base level of
corporate governance.
In this study, we examine the perceptions of external auditors and directors
concerning the impact of certain governance mechanisms on the prevention and
detection of control weaknesses, financial statement errors and fraud. While
this provides only an indirect measure of the effectiveness of these mechan-
isms, the findings reflect the considerable experience of these two groups. In any

corporate governance regime, auditors and directors should have considerable
influence over the accountability of management and the integrity of financial
reporting. Directors are in a position to set the ‘tone from the top’ with regard
to strong governance in their companies. External auditors, through their interac-
tions with audit committees and client management, are able to influence their
client’s strength of internal controls and integrity of financial reporting. Thus,
identifying the perceptions of these two key constituents should enhance our
understanding of the importance of corporate governance mechanisms. In addition,
interesting comparisons can be made to highlight any differences in the view-
points of these two parties in order to encourage further debate on what constitutes
strong governance and where the balance of power/responsibility for corporate
governance should lie.
Directors and auditors of listed companies in Singapore participated in two
experimental cases. In the first, the strength of the audit committee, the exist-
ence of an internal audit function and the strength of the company’s code of
198 Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223
© AAANZ, 2002
ethical conduct were varied. Participants were asked to indicate the extent to
which they believed that these corporate governance mechanisms would enable
the company both to prevent and to detect weaknesses in internal control, errors
in the financial statements, management fraud and employee fraud. They were
also asked to indicate the extent to which they believed that the same mechan-
isms would assist the company’s external auditors to perform their audit effect-
ively and to resist pressure from management. In the second case, three pieces
of information relating to the company’s external auditors were varied. These
were audit partner rotation, outsourcing of internal audit services and whether the
same audit firm audits all companies within the group. Participants were asked
to indicate the extent to which they believed that the auditors would be able
to resist pressure from management and also to detect weaknesses in internal
control, material errors in the financial statements, and management and employee

fraud material to the financial statements.
The results of the study indicate that auditors and directors perceive that
the existence of an internal audit function and strict enforcement of a corpor-
ate code of conduct have a significant impact on the company’s ability to
strengthen controls, prevent and detect fraud and financial statement errors and
enhance audit effectiveness. The results relating to a strong audit committee are
mixed, with some impact on preventing and detecting financial statement
errors, detecting management fraud and enhancing audit effectiveness. The
study also finds some interesting differences between the perceptions of auditors
and those of directors. Auditors place more weight on the internal audit
function as a mechanism to detect weaknesses in controls and to prevent and
detect fraud. This may be due to auditors’ greater familiarity with the role that
internal audit can play in reducing audit risk. It may also be due to their preference
for the presence of an internal audit function to enhance the internal controls in
their audit clients. In contrast, directors have more confidence in a strongly
enforced code of conduct as a means of detecting control weaknesses, financial
statement errors and fraud. This finding may reflect directors’ beliefs that
setting the tone from the top encourages staff to maintain high ethical standards,
while auditors may have experienced a lack of commitment to codes of conduct
(Cohen
et al., 2000). The variables in the second case do not give rise to signific-
ant differences in the perceived ability of the auditor to detect control weak-
nesses and fraud. However, when the audit firm audits all companies in a group,
participants perceive that auditors would be more able to resist pressure from
management. Auditors also appear to have more confidence than directors in
their ability to detect errors in financial statements.
The remainder of the paper is structured as follows. The next section pro-
vides the background to the study while the third section outlines the research
questions addressed. The fourth section summarises the research methods used,
while the results are reported and discussed in the fifth section. In the final

section, some conclusions are drawn, the limitations of the study are noted, and
the implications of the findings and further research opportunities are discussed.
Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223 199
© AAANZ, 2002
2. Background
As with any experimental design, the number of variables that we can exam-
ine is constrained by many interacting factors such as model complexity, the
length of the research instrument and the likely response rate. From the numerous
checks and balances that have been suggested, we selected three client-related
mechanisms that could influence the quality of the external audit and hence the
quality of the financial statements. We also selected three mechanisms relating
to the external auditor which could affect audit quality and independence. Our
selection was based on a review of the relevant literature, the results of an
earlier survey of investors, auditors and directors (Goodwin and Seow, 2000)
and the listing requirements of the Stock Exchange of Singapore (SES, 1998).
1
The independent variables investigated in our study are of particular interest as
prior research into these variables has produced mixed results.
2.1. Client-related mechanisms
2.1.1. The strength of the audit committee
All listed companies in Singapore must have an audit committee, comprising
at least three members, the majority of whom must not be executive directors
(Companies Act, 1994). The literature suggests that an effective audit commit-
tee should play an important role in strengthening the financial controls of an
entity (Collier, 1993; English, 1994; Vinten and Lee, 1993). A number of
studies have found that companies with an audit committee, particularly when
that committee is active and independent, are less likely to experience fraud
(Beasley et al., 2000; Abbott et al., 2000; McMullen, 1996) and other reporting
irregularities (McMullen, 1996; McMullen and Raghunandan, 1996). Findings
also suggest that audit committees are effective in reducing the incidence of

earnings management that may result in misleading financial statements
(Defond and Jiambalvo, 1991; Dechow et al., 1996; Peasnell et al., 2000). Further,
they should enhance the effectiveness of both internal and external auditors
(Simnett et al., 1993). However, Cohen et al. (2000) report that a number of
audit practitioners involved in exploratory interviews expressed concern over
the effectiveness of audit committees, with some partners suggesting that audit
committees are not powerful enough to resolve conflicts with management.
It is generally agreed that, for an audit committee to be effective, a majority
if not all members should be independent (Cadbury, 1992) and they should have
1
The SES was combined with the Singapore International Monetary Exchange in December
1999 and a new company, the Singapore Exchange Ltd (SGX), was formed. The Listing
Manual is now known as the SGX-ST Listing Manual. Because this research was under-
taken prior to December 1999, our references are made to the SES.
200 Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223
© AAANZ, 2002
an understanding of accounting, auditing and control issues (Cohen et al., 2000;
Goodwin and Seow, 2000; Hughes, 1999; Lear, 1998). Further, the committee
should meet regularly (Collier, 1993; Hughes, 1999) and strong channels of
communication should exist with both the external and internal auditors
(Cadbury, 1992; Deloitte and Touche, 1998). Goodwin and Seow (2000) found
that investors, auditors and directors all believe that a strong effective audit
committee assists external auditors to perform the audit.
2.1.2. Internal audit
While there is no mandatory requirement in Singapore for listed companies
to have their own internal audit function, the benefits of such a function have
been noted by the SES (1996b). Furthermore, various corporate governance
committees have recommended that companies should establish an internal
audit function (Cadbury, 1992; COSO Report, 1992; Committee on Corporate
Governance, 2001). Only a few studies have examined the benefits of internal

audit and results have been mixed. Beasley et al. (2000) report that companies
which experience fraud are less likely to have an internal audit function. Cohen
et al. (2000) found that 90 per cent of their respondents indicated that internal
audit could affect corporate governance but concern was expressed over the
strength of many internal audit departments. In Singapore, while Goodwin and
Seow (2000) found both investors and auditors strongly in favour of an internal
audit function, the results of a survey conducted by Goodwin et al. (1998)
indicated that many companies listed on the SES did not have their own internal
audit function.
2.1.3. Corporate code of conduct
Cadbury (1992) recommends that the board of directors should draw up a
code of ethics or a statement of business practice to be published both inter-
nally and externally. Research findings in the US indicate that corporate codes
of ethics, accompanied by training and monitoring programs, have an impact on
employee behaviour (Pickard, 1995). There is also evidence that companies that
disclose in their annual report a commitment to ethical behaviour or compliance
with their code of conduct outperform those companies without such disclos-
ures (Stoffman, 1991; Verschoor, 1997 and 1998). The Working Group on Corporate
Practices and Conduct (1993) in Australia argues that a code of ethical conduct
should enhance the company’s reputation for fair trading and help to maintain
high standards of behaviour throughout the organisation. It should also give
employees a clear idea of the company’s goals, help to develop pride among staff
and give a focus to the organisation as a whole (Hicks, 1999).
Merely having in place a code of conduct is not, however, considered to be
sufficient (Howard, 1996). Cohen et al. (2000) report that their interviewees
Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223 201
© AAANZ, 2002
felt that, as such codes are generally given only superficial attention, they have
limited effect on reducing the potential for misstatements and fraud. To be effect-
ive, codes must be communicated throughout the organisation and be enforced

by management (Brooks, 1992; McNamee, 1992). Further, the board of directors
needs to ensure that checks are performed to ensure that corporate values are
adhered to (Howard, 1996). Goodwin and Seow (2000) found strong support for
a formal corporate code of ethical conduct to which staff should adhere. How-
ever, there has been very little empirical research into the effectiveness of a strongly
enforced code of conduct as a corporate governance mechanism.
2.2. Auditor-related mechanisms
2.2.1. Rotation of audit partner
Mandatory rotation of audit firms has been suggested as a means of enhanc-
ing auditor independence (Brody and Moscove, 1998). However, considerable
concern has been raised about the adverse effects of such a requirement, par-
ticularly on audit quality and audit fees (AICPA, 1978, 1992; Cadbury, 1992;
Arrunada and Paz, 1997). As the AICPA (1992) argues, problem audits are more
likely to occur when the auditor lacks a sound understanding of the client’s
business, operations and systems. As a result of these concerns, a within-firm
rotation has been recommended, involving a periodic change of partners in
charge of an audit (Cadbury, 1992; Wendell et al., 1998). In Singapore, the SES
(1998) requires
2
that auditors of listed companies practise partner rotation at least
once every five years. However, Goodwin and Seow (2000) found mixed sup-
port for partner rotation amongst auditors and the effectiveness of this practice
has not been tested empirically.
2.2.2. The provision of internal audit services to the audit client
The impact on audit independence of the provision of other services to an
audit client has been the subject of intense debate (Briloff, 1987; Cadbury, 1992;
Pany and Reckers, 1988; Wallman, 1996; Levitt, 2000b). Wallman (1996) sees
a positive impact on audit quality resulting from the more in-depth understand-
ing of the client obtained by providing other services. He argues that staff who
provide internal audit services are more likely to communicate problems to the

auditor if they are employees of the audit firm, ‘thereby reducing the possib-
ility of investor losses caused by fraud’ (Wallman, 1996, p.87).
However, others are concerned that issues such as fee dependency and an
involvement in management decision making could lead to a compromise of
2
This requirement remains in force as part of the SGX-ST Listing Manual (2000).
202 Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223
© AAANZ, 2002
audit independence (Briloff, 1987; Gul, 1991; Knapp, 1985). Acciani (1995,
p.50) argues that, when internal audit services are provided by the external
auditor, ‘the common checks and balances that exist between internal and
external auditors’ are eliminated, thereby compromising external auditor inde-
pendence and increasing the risk of unreported control weaknesses and fraud.
In an empirical study, Lowe et al. (1999) found that loan officers’ perceptions
of audit independence and financial statement reliability were adversely
affected when the external auditor assumed management responsibilities with
respect to internal audit. However, perceptions were favourable when the audit
firm performed internal audit activities using personnel not involved in the
external audit. Swanger and Chewning (2001) found that financial analysts’
perceptions of audit independence were lower when the external auditor per-
formed internal audit services for the client compared to when another audit
firm performed those services or when the client employed its own internal
audit staff. They also found, however, that perceptions of independence were
higher when different staff of the audit firm performed internal audit services
for the client compared to when the same staff provided those services. The
recent ruling by the SEC (2000) allows an audit firm to perform no more than
40 per cent of a client’s internal audit work. It also stresses that the auditor
must not engage in management decision making with respect to internal audit
work performed for a client.
2.2.3. Auditing all companies within a group

The SES (1996a) argues that, if different audit firms audit different compa-
nies within a listed group, the auditors of the group may not obtain ‘an accurate
and comprehensive view of the entire group’s affairs’ (para.5.2). It is there-
fore suggested that the same auditor audit all companies in the group to help
to prevent frauds by directors and managers of listed companies. However, in
recognition of the difficulties that could occur if such a requirement was
mandated for all subsidiaries and associate companies both in and outside of
Singapore, the SES (1998) recommends
3
that the same audit firm should be
appointed to audit the parent company and its Singapore-based subsidiaries and
associates.
4
Overseas subsidiaries and associated companies are required to be
audited by a reputable accounting firm, although no guidance is given on how
reputability should be evaluated.
There is little discussion in the academic literature on the advantages or
otherwise of the same audit firm auditing all companies within a group.
3
This requirement remains in force as part of the SGX-ST Listing Manual (2000).
4
Exceptions to this requirement are permitted if a listed company’s board and audit com-
mittee are satisfied that the standard and effectiveness of the audit will not be compromised.
Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223 203
© AAANZ, 2002
However, Moizer et al. (1986) discuss the possible conflicts of interest that
can exist between primary and secondary auditors involved in the audit of a group
of companies, and the potential for client management to play off one audit
firm against the other. While Moizer et al. (1986) concentrate their discussion
on the issue of audit fees and audit tenure, their arguments suggest that the

client is more able to conceal important information from the auditor when
the firm does not audit all companies in the group. Discovery of irregularities
may therefore be more difficult. Jubb and Houghton (1999) further suggest
that auditor-client relationships will be enhanced and audit efficiencies
obtained when a common auditor audits all companies within a group. These
arguments imply that the transaction costs to the client associated with replac-
ing the auditor are likely to be greater when a single audit firm audits all com-
panies within the group. Emby and Davidson (1997) also suggest that when
transaction costs are high, the auditor has more power to resist management
pressure. They find empirical support for the argument that parent company
auditors have greater power to resist pressure from management of other com-
panies within the group because of their relationship with the controlling
entity’s management. This suggests that audit independence can be enhanced
when the audit firm audits all companies in the group but no empirical studies
have addressed this.
3. Research questions
The following research questions are addressed in the study:
RQ1: Are auditors’ and directors’ perceptions of a company’s ability to prevent
and detect control weaknesses, fraud and errors in the financial statements
influenced by:
(a) the strength of the audit committee
(b) the existence of a strong internal audit function
(c) board enforcement of a strict code of ethical conduct for employees?
RQ2: Are auditors’ and directors’ perceptions of the ability of a company’s external
auditors to perform the annual audit effectively and to resist pressure from man-
agement influenced by:
(a) the strength of the audit committee
(b) the existence of a strong internal audit function
(c) board enforcement of a strict code of ethical conduct for employees?
RQ3: Are auditors’ and directors’ perceptions of the ability of a company’s external

auditors to detect control weaknesses, fraud and errors affecting the financial
statements influenced by:
204 Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223
© AAANZ, 2002
(a) rotation of the audit partner in charge of the audit
(b) the provision of internal audit services by the external auditors
(c) whether the audit firm audits all companies within the group?
RQ4: Are auditors’ and directors’ perceptions of the ability of a company’s external
auditors to resist pressure from management in the event of a dispute influenced
by:
(a) rotation of the audit partner in charge of the audit
(b) the provision of internal audit services by the external auditors
(c) whether the audit firm audits all companies within the group?
4. Research methods
To answer the research questions, participants were asked to role-play as
‘director’ or ‘auditor’ on a review committee focused on corporate govern-
ance issues. This design was used because it simulates the typical approach
taken by the Singapore government when addressing important issues per-
taining to the business community. Review committees have been established
comprising representatives from various interest groups to put forward the
views of that group on such matters as the corporate regulatory framework,
disclosure standards and corporate governance.
5
Thus, asking participants
to role-play as members of a committee was designed to add realism to the
experiment and encourage participants to take the tasks more seriously. To
help ensure that respondents focused on their respective roles as directors or
auditors, a separate covering letter was used for each group. The letter to dir-
ectors stressed that we were concerned with issues that directors considered
important for strong corporate governance and the covering letter to auditors

stressed that we were concerned with issues that auditors considered
important.
6
Participants were told to assume that, in order to assess the level of agree-
ment on corporate governance issues, the chairman of the review committee
5
Committees of this nature were established by the Ministry of Finance together with the
Monetary Authority of Singapore and the Attorney-General’s Chambers in 1999.
6
While it could be argued that participants were not reviewing the cases as directors and
auditors per se but rather as members of a review committee, the study was framed to cap-
ture their views as either directors or auditors. Furthermore, it can be argued that, even
though role-playing, participants’ views would have been formulated from their individual
experiences as either directors or auditors.
Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223 205
© AAANZ, 2002
had developed two hypothetical cases.
7
These cases differed in their focus, with
one addressing the strength of the company’s corporate governance and the
other addressing aspects relating to the external auditors.
8
Participants were
asked to read the cases and respond to a series of questions. In each case, three
variables were manipulated in a 2 × 2 × 2 between-subjects design. This
resulted in eight versions of each case.
9
4.1. Experimental case one
The first case described a listed company with diverse interests. The same
international audit firm had audited the company since its listing in 1989 and

the partner-in-charge had held that position for the last five years. There were
three executive directors, an independent chairman and two other non-executive
directors. Participants were informed that they had no reason to doubt the integ-
rity of board members or the management team, but that they did not know
them personally. They were also told that they had no direct knowledge of the
company’s control systems but had no reason to believe that there were any
unusual weaknesses in controls. This scenario was designed to portray a fairly
neutral picture of a company with no noticeable weaknesses in corporate gov-
ernance other than with respect to the variables being manipulated.
4.1.1. Independent variables
The first variable manipulated was the strength of the audit committee. All
participants were advised that the company’s audit committee comprised three
directors. In the strong condition, all directors were independent and all had
accounting and auditing expertise. The committee met every two months and
had regular meetings with the external auditor. The committee was well resourced
and the members were known to take their responsibilities very seriously. In the
weak condition, the committee comprised two independent directors and one
7
The research instrument comprised an instruction sheet, the two cases and some biograph-
ical questions. The instrument was pre-tested using post-graduate students and staff with
business and auditing experience at a major university in Singapore. Minor modifications
were made to the final instrument based on the feedback received.
8
It could be argued that presenting participants with two separate cases might lead to con-
fusion. However, we believe the advantages of this approach outweigh the disadvantages.
Primarily, we were able to test more variables with a limited number of participants. Had we
used only one case with six independent variables, not only would we have required a much
greater number of participants but the design could have generated complex interactions
that would have been difficult to interpret.
9

To avoid demand effects, a participant could receive any one of the eight versions in the
first experimental case together with any one of the eight versions in the second experi-
mental case. This resulted in 64 different versions of the instrument.
206 Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223
© AAANZ, 2002
executive director. Only the executive director had any accounting or auditing
expertise. The committee met two or three times a year for the purpose of com-
plying with statutory and listing requirements.
The second independent variable involved the internal audit function. In one
condition, participants were told that the company had a strong and active internal
audit function led by an experienced internal audit manager who had direct access
to the audit committee. In the second condition, participants were advised that
the company did not have its own internal audit function as management felt
that it could not justify spending the necessary resources on such a function.
The third variable concerned the company’s commitment to a written code of
ethical conduct. In the strong condition, participants were informed that the
board had developed a comprehensive code of ethical conduct with which staff
were required to be familiar and to comply. The board regularly monitored
management’s enforcement of the code. In the weak condition, participants
were advised that, while the company had a written code of ethical conduct,
this had not been updated for many years. Further, the board did not monitor
management’s enforcement of the code and little emphasis was placed on the
need for staff to be aware of its contents.
4.1.2. Dependent variables
Participants were asked to respond to ten questions. Four concerned the extent to
which participants believed that the mechanisms would prevent the occurrence
of weaknesses in internal control, errors in the financial statements, manage-
ment fraud and employee fraud respectively. A further four questions related to
the extent to which the mechanisms would enable the company to detect each
of these problems should they occur. The final two questions related to the

extent that the corporate governance mechanisms discussed in the case would
assist the company’s external auditors to perform their annual audit effectively
and to resist pressure from management. For all questions, an eleven-point
scale ranging from zero to ten was provided, with zero being described as ‘to a
very limited extent’ and ten described as ‘to a very great extent’.
4.2. Experimental case two
The scenario developed for this case was also designed to portray a company
with no noticeable weaknesses in corporate governance. Some background
information was provided about a second listed company with similar board
and management structure to the company in the first case. The company
had subsidiaries and associate companies in Singapore, Malaysia, China and
Taiwan. The company’s audit committee comprised the finance director and two
non-executive directors. It met regularly and appeared to take its responsibil-
ities seriously. The company did not have its own internal audit function but
outsourced internal audit activities when necessary. Participants were again
Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223 207
© AAANZ, 2002
advised that they had no reason to believe that there were any unusual weak-
nesses in control systems and no reason to doubt the integrity of management.
4.2.1. Independent variables
For this case, participants were told that the same international audit firm had
audited the company since its listing in 1992. Three variables relating to this audit
firm were manipulated. The first concerned the issue of audit partner rotation.
In one condition, the partner in charge of the audit had been responsible for the
audit since the firm acquired the client. Partner rotation in accordance with
Singapore listing requirements would only be implemented in a few years’ time.
In the other condition, the firm had recently implemented partner rotation and
the partner in charge had taken over the audit in the present year. Prior to this,
another partner had been in charge of the audit since the client was acquired.
The second variable involved the provision of internal audit services to the

client. In one condition, participants were advised that internal audit activities
were provided by the internal audit division of the same audit firm while in the
other condition, they were told that another international audit firm provided
these services.
The final manipulation concerned the firm auditing all companies in the
group. In one condition, the firm audited all subsidiaries and associate com-
panies while in the other condition, the overseas subsidiaries and associate
companies were audited by local firms in their respective countries of operation.
4.2.2. Dependent variables
Participants were asked to respond to five questions with the same scales as
in the first experimental case. Four questions related to the extent that the exter-
nal auditors, in the course of a normal statutory audit, would be able to detect
weaknesses in internal control, material errors in the financial statements, and
both management and employee fraud. The last question concerned the extent
to which the auditors would be able to resist pressure from management.
4.3. Participants and procedures
Participants in the study were directors and auditors of listed companies in
Singapore. Research instruments were mailed to some 400 directors of listed
companies, randomly selected from corporate information available on the SES
website. Responses were returned in a stamped-addressed envelope directly to
the researchers. Usable responses were received from 71 directors.
10
The great
10
While this represents a response rate of only 18 per cent, this rate is comparable to other
studies involving directors in Singapore (Goodwin and Seow, 2000) and also in the U.S.
(Hermanson, 2000).
208 Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223
© AAANZ, 2002
majority of respondents were males (70) and more than 84 per cent (60) were

over the age of 40. Approximately 69 per cent (49) of respondents indicated that
they held more than one directorship and almost 65 per cent (47) were members
of an audit committee.
For the auditor group, a number of audit partners were contacted and asked to
assist in distributing the instrument to employees above the level of audit senior.
Approximately 200 copies were distributed and 63 usable responses were mailed
directly to the researchers.
11
Of these respondents, 21 (34 per cent) were partners,
26 (42 per cent) were managers and 15 (24 per cent) were supervisors. There
were 38 males and 24 females and almost 73 per cent (45) were aged 40 or less.
Analysis of variance (ANOVA) was used to test for differences in responses
due to experience, age and gender. The tests indicated that these biographical
differences had no significant impact on the results.
While low response rates are a possible cause of concern in the study, there
are a number of indicators that suggest that non-response bias is unlikely to be
a problem. First, ANOVA tests revealed no significant differences between early
and late respondents. Second, the various versions of the research instrument
were distributed evenly and were returned in reasonably similar proportions.
12
Third, with regard to the director group, the biographical data supplied by
respondents is consistent with ad hoc evidence about the characteristics of
Singaporean directors
13
, thereby giving no reason to suggest that respondents
were not ‘typical’ directors of Singapore listed companies.
5. Results and discussion
5.1. Experimental case one (research questions 1 and 2)
Research questions 1 and 2 are explored using multivariate analyses of vari-
ance (MANOVA)

14
supported by univariate analysis and multiple comparison
tests where appropriate.
15
The MANOVA results indicate that the internal audit
11
As not all of the instruments may have been distributed by the partners, the actual
response rate is likely to be greater than suggested by these figures.
12
The mean cell size was 16.4, with relatively small standard deviations of slightly less than
two for both scenarios.
13
Media reports indicate that there are very few female directors in Singapore and most
directors are older males holding multiple directorships (see for example Lee, 2002; Lee
and Chin, 2002).
14
MANOVA is used because the dependent variables are correlated, with Pearson correla-
tion scores ranging from .353 to .733.
15
Bonferroni and Tukey multiple comparison tests were undertaken for all significant inter-
action effects.
Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223 209
© AAANZ, 2002
variable is highly significant (p = .001) while the code of conduct and the respond-
ent group variables are marginally significant (p = .068 and .088 respectively).
The audit committee variable is not significant.
16
These results support the need
for separate analyses of variance (ANOVA) for each of the dependent variables
and the results of these are reported in Tables 1−5. Panel A of each table gives

the mean responses (standard deviations) for all participants, for auditors and
for directors for each condition of the manipulated variables. Panel B reports
the results of the ANOVA for the various dependent variables, showing the
main effects and the significant first order interactions.
17
Table 1 focuses on the prevention and detection of weaknesses in internal
control. Panel A shows that all mean responses are higher for the strong condi-
tions compared to the weak conditions. From Panel B it can be seen that the
presence or absence of an internal audit function has a significant impact on
respondents’ perceptions regarding both the prevention and detection of control
weaknesses. Board enforcement of a strict code of ethical conduct for employees
is significant at the .05 level for the detection of control weaknesses but only
marginally significant (p = .08) for the prevention of weaknesses. The strength
of the audit committee is not significant.
There is a significant interaction between the code of conduct variable and
the respondent group for the detection of control weaknesses. Multiple compar-
ison tests indicate that, while the mean responses for the auditor group do not
differ significantly, the mean responses for directors are significantly different.
Thus, directors appear to have greater confidence than do auditors that a strictly
enforced code of conduct will help to detect control weaknesses. The ANOVA
also reveals a marginally significant difference between the internal audit variable
and the respondent group. However, multiple comparison tests do not identify a
significant interaction between these two factors.
Table 2 reports the results of the ANOVA concerning the prevention and detec-
tion of errors in the financial statements. From Panel B, it can be seen that the
internal audit variable is highly significant for both dependent variables. The
audit committee variable is significant at the .05 level with regard to the detec-
tion of errors and marginally significant (p = .067) for the prevention of errors.
The code of conduct variable is also significant for the detection of errors but
16

Factor analysis indicates that the ten dependent variables can be collapsed into a single
construct. ANOVA using a single construct score (calculated by totalling the scores for the
dependent variables) reveals that the three manipulated variables are all significant (audit
committee at p = .042, internal audit at p = .000 and code of conduct at p = .001). How-
ever the respondent group variable is not significant. These findings further support the need
for univariate analysis on each of the dependent variables to determine which variables
are driving the results.
17
As none of the higher order interactions are significant, details of these effects have not
been provided.
210 Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223
© AAANZ, 2002
not for error prevention. Further examination of the significant interaction
effect between the code of conduct variable and respondent group reveals that
this is again driven by the perceptions of the director group.
Tables 3 and 4 report the results relating to the prevention and detection of
fraud. Both tables show in Panel A that respondents are somewhat sceptical
Ta
bl
e
1
The extent to which the client-related mechanisms would help prevent and detect weaknesses in
internal controls
Panel A
Descriptive statistics
Panel B
Results of ANOVA
All
Range
(0–10)

Auditors Directors
n Mean (s.d.) n Mean (s.d.) n Mean (s.d.)
Prevent weaknesses in internal controls
Audit committee S 72 5.28 (2.29) 0–9 36 5.03 (2.20) 36 5.53 (2.38)
W 62 5.03 (2.11) 1–8 27 4.63 (1.86) 35 5.34 (2.26)
Internal audit Y 67 5.78 (2.12) 1–9 33 5.70 (1.88) 34 5.85 (2.35)
N 67 4.55 (2.13) 0–8 30 3.93 (1.86) 37 5.05 (2.24)
Code of conduct S 68 5.49 (2.15) 1–9 33 5.00 (1.92) 35 5.94 (2.28)
W 66 4.83 (2.22) 0–9 30 4.70 (2.22) 36 4.94 (2.25)
Detect weaknesses in internal controls
Audit committee S 72 5.69 (2.22) 0–9 36 5.44 (2.27) 36 5.90 (2.16)
W 62 5.08 (2.13) 1–9 27 5.07 (2.34) 35 5.09 (1.99)
Internal audit Y 67 6.27 (1.94) 1–9 33 6.39 (1.82) 34 6.15 (2.08)
N 67 4.53 (2.08) 0–8 30 4.07 (2.15) 37 4.91 (1.98)
Code of conduct S 68 5.82 (1.94) 2–9 33 5.30 (2.14) 35 6.31 (1.61)
W 66 4.96 (2.35) 0–9 30 5.27 (2.48) 36 4.71 (2.25)
Prevent weaknesses in
internal controls
Detect weaknesses in
internal controls
Main effects F-Value Probability F-Value Probability
Audit committee 0.730 .395 2.699 .103
Internal audit 11.298 .001 26.942 .000
Code of conduct 3.117 .080 5.714 .018
Respondent group* 2.665 .105 0.541 .464
Significant interaction effects (only those significant at p = .1 or less are shown)
Internal audit × group NS NS 3.196 .076
Code of conduct × group NS NS 4.963 .028
S = Strong; W = Weak; Y = Yes; N = No; NS = Not Significant
*Auditors versus directors

Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223 211
© AAANZ, 2002
of the extent to which the corporate governance mechanisms in the scenario
would help prevent and detect fraud.
18
However, mean responses are higher for
the strong conditions than for the weak conditions.
Ta
bl
e
2
The extent to which the client-related mechanisms would help prevent and detect errors in financial
statements
Panel A
Descriptive statistics
Panel B
Results of ANOVA
All Auditors Directors
n Mean (s.d.) Range (0–10) n Mean (s.d.) n Mean (s.d.)
Prevent errors in financial statements
Audit committee S 72 5.57 (2.39) 0–9 36 5.36 (2.36) 36 5.78 (2.44)
W 62 4.94 (1.65) 2–9 27 4.85 (1.51) 35 5.00 (1.77)
Internal audit Y 67 5.81 (1.82) 2–9 33 5.52 (1.60) 34 6.09 (1.99)
N 67 4.75 (2.23) 0–9 30 4.73 (2.39) 37 4.76 (2.13)
Code of conduct S 68 5.41 (1.87) 2–9 33 5.06 (1.64) 35 5.74 (2.03)
W 66 5.14 (2.31) 0–9 30 5.23 (2.43) 36 5.06 (2.24)
Detect errors in financial statements
Audit committee S 72 5.88 (2.44) 0–9 36 5.50 (2.42) 36 6.25 (2.43)
W 62 5.03 (2.01) 1–9 27 5.15 (2.27) 35 4.94 (1.81)
Internal audit Y 67 6.04 (2.00) 2–9 33 5.97 (1.83) 34 6.12 (2.17)

N 67 4.93 (2.42) 0–9 30 4.67 (2.67) 37 5.14 (2.21)
Code of conduct S 68 5.88 (2.07) 2–9 33 5.27 (2.08) 35 6.46 (1.92)
W 66 5.08 (2.43) 0–9 30 5.43 (2.64) 36 4.78 (2.23)
Prevent errors in
financial statements
Detect errors in
financial statements
Main effects F-Value Probability F-Value Probability
Audit committee 3.429 .067 4.842 .030
Internal audit 8.008 .005 9.394 .003
Code of conduct 0.882 .349 5.196 .024
Respondent group* 0.746 .390 0.380 .539
Significant interaction effects (only those significant at p = .1 or less are shown)
Code of conduct × group NS NS 5.400 .022
S = Strong; W = Weak; Y = Yes; N = No; NS = Not Significant
*Auditors versus Directors
18
Five of the 16 scores with respect to management fraud and only two of the 16 scores
with respect to employee fraud exceeded the midpoint of five.
212 Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223
© AAANZ, 2002
From Panel B of Table 3, it can be seen that the code of conduct variable is highly
significant for the prevention of management fraud while the internal audit variable
is marginally significant (p = .09). For the detection of management fraud, all three
manipulated variables are significantly different at the .05 level while the group
variable is marginally significant at p = .071. Further tests of the interaction effect
between code of conduct and respondent group again indicate that it is directors
who believe that a strongly enforced code affects both the prevention and detection
Ta
bl

e
3
The extent to which the client-related mechanisms would help prevent and detect management fraud
Panel A
Descriptive statistics
Panel B
Results of ANOVA
All Auditors Directors
n Mean (s.d.) Range (0–10) n Mean (s.d.) n Mean (s.d.)
Prevent management fraud
Audit committee S 72 4.92 (2.42) 0–9 36 4.83 (2.04) 36 5.00 (2.78)
W 62 4.44 (2.05) 1–10 27 4.22 (1.83) 35 4.60 (2.21)
Internal audit Y 67 5.06 (2.34) 0–9 33 5.06 (2.03) 34 5.06 (2.63)
N 67 4.33 (2.14) 0–10 30 4.03 (1.75) 37 4.57 (2.40)
Code of conduct S 68 5.25 (2.18) 1–10 33 4.73 (2.04) 35 5.74 (2.23)
W 66 4.12 (2.21) 0–9 30 4.40 (1.89) 36 3.89 (2.45)
Detect management fraud
Audit committee S 72 5.06 (2.21) 0–9 36 4.50 (1.96) 36 5.61 (2.34)
W 62 4.24 (2.02) 1–8 27 4.15 (2.20) 35 4.31 (1.91)
Internal audit Y 67 5.22 (2.16) 0–9 33 5.18 (1.83) 34 5.27 (2.47)
N 67 4.13 (2.03) 0–9 30 3.43 (1.92) 37 4.70 (1.96)
Code of conduct S 68 5.21 (2.11) 1–9 33 4.49 (2.03) 35 5.89 (1.97)
W 66 4.14 (2.09) 0–8 30 4.20 (2.11) 36 4.08 (2.10)
Prevent management fraud Detect management fraud
Main effects F-Value Probability F-Value Probability
Audit committee 2.460 .119 5.604 .020
Internal audit 2.929 .090 10.172 .002
Code of conduct 7.329 .008 8.420 .004
Respondent group* 0.703 .404 3.311 .071
Significant interaction effects (only those significant at p = .1 or less are shown)

Internal audit × group NS NS 3.422 .067
Code of conduct × group 3.508 .064 4.956 .028
S = Strong; W = Weak; Y = Yes; N = No; NS = Not Significant
*Auditors versus directors
Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223 213
© AAANZ, 2002
of management fraud. In contrast, auditors see little impact of a strongly enforced
code, with all means falling between 4.20 and 4.73. Tests of the marginally significant
interaction between the internal audit variable and respondent group indicate
that the auditor group considers the existence of an internal audit function to
have a significant impact on the company’s ability to detect management fraud.
Table 4 shows that, with regard to employee fraud, the code of conduct vari-
able is significant at the .05 level for both the prevention and detection of fraud,
Ta
bl
e
4
The extent to which the client-related mechanisms would help prevent and detect employee fraud
Panel A
Descriptive statistics
Panel B
Results of ANOVA
All Auditors Directors
n Mean (s.d.) Range (0–10) n Mean (s.d.) n Mean (s.d.)
Prevent employee fraud
Audit committee S 72 4.79 (2.22) 0–9 36 4.94 (2.22) 36 4.64 (2.24)
W 62 4.45 (2.19) 1–10 27 4.52 (1.87) 35 4.40 (2.43)
Internal audit Y 67 4.99 (2.30) 0–8 33 5.46 (2.11) 34 4.53 (2.42)
N 67 4.28 (2.06) 0–10 30 4.00 (1.76) 37 4.51 (2.27)
Code of conduct S 68 5.12 (2.12) 1–10 33 4.91 (1.99) 35 5.31 (2.25)

W 66 4.14 (2.19) 0–8 30 4.60 (2.18) 36 3.75 (2.16)
Detect employee fraud
Audit committee S 72 4.88 (2.13) 0–9 36 4.56 (2.09) 36 5.20 (2.15)
W 62 4.44 (2.09) 1–9 27 4.30 (2.25) 35 4.54 (1.98)
Internal audit Y 67 5.00 (2.17) 0–9 33 5.12 (2.03) 34 4.88 (2.33)
N 67 4.35 (2.01) 0–9 30 3.70 (2.05) 37 4.87 (1.84)
Code of conduct S 68 5.24 (2.10) 1–9 33 4.67 (2.18) 35 5.77 (1.91)
W 66 4.10 (1.98) 0–7 30 4.20 (2.12) 36 4.01 (1.87)
Prevent employee fraud Detect employee fraud
Main effects F-Value Probability F-Value Probability
Audit committee 0.786 .377 1.442 .232
Internal audit 3.273 .073 4.001 .048
Code of conduct 5.291 .023 9.446 .003
Respondent group* 0.313 .577 1.682 .197
Significant interaction effects (only those significant at p = .1 or less are shown)
Internal audit × group 3.791 .054 4.726 .032
Code of conduct × group NS NS 3.929 .050
S = Strong; W = Weak; Y = Yes; N = No; NS = Not Significant
*Auditors versus Directors
214 Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223
© AAANZ, 2002
while the internal audit variable is significant for the detection of fraud and
marginally significant for the prevention of fraud (p = .073). The audit committee
variable is not significant. Tests of the interaction effects reveal that it is the auditor
group that believes that the presence or absence of an internal audit function
has an impact on both the prevention and detection of employee fraud while the
director group believes that a strongly enforced code has a greater impact.
Table 5 concerns perceptions of whether the corporate governance mechanisms
would assist external auditors to perform the audit effectively and to resist
Table 5

The extent to which the client-related mechanisms would assist the external auditors to perform effect-
ively and resist management pressure
Panel A
Descriptive statistics
Panel B
Results of ANOVA
All Auditors Directors
n Mean (s.d.) Range (0–10) n Mean (s.d.) n Mean (s.d.)
Perform effectively
Audit committee S 72 6.30 (2.44) 0–10 36 6.17 (2.35) 36 6.42 (2.55)
W 62 5.40 (1.89) 2–9 27 5.19 (1.88) 35 5.57 (1.90)
Internal audit Y 67 6.49 (1.99) 1–10 33 6.46 (1.82) 34 6.53 (2.16)
N 67 5.27 (2.32) 0–10 30 4.97 (2.34) 37 5.52 (2.30)
Code of conduct S 68 6.49 (1.68) 3–9 33 6.18 (1.76) 35 6.77 (1.57)
W 66 5.26 (2.56) 0–10 30 5.27 (2.55) 36 5.26 (2.61)
Resist pressure
Audit committee S 72 6.22 (2.39) 0–10 36 6.39 (2.43) 36 6.06 (2.37)
W 62 5.53 (2.02) 0–9 27 5.33 (2.22) 35 5.69 (1.88)
Internal audit Y 67 6.36 (2.14) 0–10 33 6.55 (2.28) 34 6.18 (2.02)
N 67 5.45 (2.27) 0–10 30 5.27 (2.35) 37 5.60 (2.22)
Code of conduct S 68 6.52 (1.89) 0–10 33 6.39 (2.11) 35 6.63 (1.68)
W 66 5.27 (2.42) 0–10 30 5.43 (2.60) 36 5.14 (2.28)
Perform effectively Resist pressure
Main effects F-Value Probability F-Value Probability
Audit committee 5.806 .018 3.518 .063
Internal audit 10.623 .001 4.624 .034
Code of conduct 12.569 .001 10.190 .002
Respondent group* 0.723 .397 0.008 .930
No significant interaction effects
S = Strong; W = Weak; Y = Yes; N = No

*Auditors versus Directors
Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223 215
© AAANZ, 2002
pressure from management in the event of a dispute. All manipulated variables
are significant at the .05 level for both dependent variables, with the exception
of the audit committee variable which is marginally significant at p = .063 for
resisting management pressure. None of the interaction effects are significant.
The results of this experimental case show that, overall, the existence of an
internal audit function is regarded as having the most impact of the three vari-
ables, being significant or marginally significant for all ten questions. This is
closely followed by the code of conduct variable, which is significant or margin-
ally significant for all but one of the questions. This is an important contribution
in view of the paucity of research concerning the effectiveness of these variables
as corporate governance mechanisms. The audit committee variable is perceived
as having the least impact, being insignificant for five of the ten questions.
In spite of prior research supporting the impact of a strong audit committee on
fraud and irregularities in financial statements, the findings of the present study
are consistent with those of Cohen et al. (2000) where audit practitioners have
expressed concern over the effectiveness of audit committees.
The findings also highlight some interesting differences between the percep-
tions of auditors and directors. Auditors have greater confidence than directors
in the relationship between the existence of an internal audit function and
the prevention and detection of employee fraud and the detection of manage-
ment fraud. This may reflect external auditors’ close association with internal
auditors and their experience in assessing the impact on audit risk of a strong
internal audit function. Further, it is likely that auditors prefer governance
mechanisms that address their concerns for a reduction in the risk of fraudulent
financial reporting. Auditors may also regard the existence of an internal audit
function as an indication that directors and management are committed to
strong corporate governance. Directors, on the other hand, may have had only

limited contact with internal auditors, particularly if their company does not
have its own internal audit function. This finding suggests that more convincing
evidence of the benefits of internal audit may be needed rather than simply
encouraging boards to develop such a function. In view of the fact that 65 per
cent of director respondents were also members of audit committees, it may
also suggest that audit committees might not be taking advantage of the full
potential of the internal audit function with regard to fraud prevention and
detection.
Another major difference is that directors are more convinced than auditors
that a strongly enforced code of conduct plays a role in detecting control weak-
nesses and financial statement errors. They also have more confidence that a
strongly enforced code will help prevent and detect fraud. Directors appear to
believe that, if the board is actively involved in monitoring management’s
enforcement of the code, the tone set from the top is likely to encourage all
staff to be more committed to achieving corporate objectives and more focused
on maintaining high ethical standards. This, in turn, should lead to improved
financial reporting quality and lower incidence of fraud. This finding may also
216 Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223
© AAANZ, 2002
reflect directors’ preferences for mechanisms that enable them to monitor and
control management and employees. Consistent with Cohen et al. (2000), how-
ever, auditors may have experienced a lack of commitment to codes of conduct
amongst their audit clients, resulting in a more sceptical attitude towards their
impact.
5.2. Experimental case two – research questions 3 and 4
Mean responses and standard deviations for the second case are reported in
Table 6.
It can be seen from this table that the means fall between 5.19 and 7.24, sug-
gesting a moderately high level of confidence in the auditors’ ability to perform
an effective and independent audit. Detailed ANOVA results are not presented

for this experimental case as few of the F statistics are significant.
19
The only
significant difference in means at the p = .05 level relates to auditing all com-
panies within the group (F = 5.191, p = .024). Respondents believe that auditing
the entire group assists the auditor to resist pressure from management. There
is a marginally significant difference (F = 3.155, p = .078) for the impact of
this variable on the auditor’s ability to detect errors in the financial statements,
suggesting that the detection of errors may be more difficult if the auditor does
not audit the entire group of companies. A marginally significant difference be-
tween the means of auditors and directors (F = 3.498, p = .064) for this variable
indicates that auditors have more confidence in their ability to detect financial
statement errors than do directors.
The results indicate that the provision of internal audit services by the audit
firm does not have a significant impact on respondents’ perceptions of the auditor’s
ability to detect control weaknesses, fraud and financial statement errors or to
resist pressure from management. This finding is important in light of the recent
audit independence debate (SEC, 2000) which suggests that the provision of
non-audit services to audit clients leads to a compromise of independence and
a reduction in audit effectiveness.
The results also suggest that auditors and directors do not believe that audit
partner rotation has a significant impact on any of the dependent variables.
Again, this finding is important in view of regulatory changes that require audit
firms to implement partner rotation.
These results may be due to a belief that neither the provision of internal audit
services nor audit partner rotation has an impact on actual audit independence
19
As the dependent variables are correlated (with Pearson correlation scores ranging from .417
to .788), MANOVA is also performed. Results indicate that the Wilks’ Lambda statistics are not
significant at p = .05. Factor analysis indicates that the five dependent variables can be collapsed

into one construct. ANOVA reveals that this construct is marginally significant (p = .056)
for auditing all companies within the group but is not significant for the other factors. The
results of these tests are consistent with the univariate analysis reported in this section.
Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223 217
© AAANZ, 2002
Table 6
The effectiveness of auditor-related mechanisms:
Means (standard deviations)
Panel A Detect weaknesses in IC
All Auditors Directors
Audit partner rotation Y 5.84 (1.90) 5.86 (1.61) 5.82 (2.15)
Currently practised N 6.25 (1.79) 6.56 (1.58) 6.00 (1.93)
Internal audit services Y 5.91 (1.81) 6.06 (1.63) 5.78 (1.96)
Provided by auditor N 6.17 (1.92) 6.29 (1.63) 6.07 (2.18)
Audit of all companies in group Y 6.07 (1.78) 6.27 (1.77) 5.89 (1.80)
N 5.97 (1.94) 6.03 (1.45) 5.92 (2.29)
Panel B Detect fin. statement errors
All Auditors Directors
Audit partner rotation Y 6.71 (1.65) 7.17 (1.44) 6.28 (1.92)
Currently practised N 6.76 (1.87) 7.00 (1.66) 6.56 (1.81)
Internal audit services Y 6.71 (1.75) 7.11 (1.64) 6.35 (1.59)
Provided by auditor N 6.76 (1.75) 7.07 (1.41) 6.48 (2.19)
Audit of all companies in group Y 6.97 (1.55) 7.24 (1.70) 6.71 (1.36)
N 6.49 (1.91) 6.93 (1.34) 6.11 (2.23)
Panel C Detect management fraud
All Auditors Directors
Audit partner rotation Y 5.71 (1.92) 5.67 (1.93) 5.74 (1.94)
Currently practised N 5.53 (1.82) 5.19 (1.71) 5.81 (1.89)
Internal audit services Y 5.47 (1.78) 5.20 (1.88) 5.70 (1.68)
Provided by auditor N 5.83 (1.98) 5.79 (1.77) 5.87 (2.19)

Audit of all companies in group Y 5.85 (1.62) 5.67 (1.88) 6.02 (1.34)
N 5.39 (2.09) 5.23 (1.79) 5.53 (2.32)
Panel D Detect employee fraud
All Auditors Directors
Audit partner rotation Y 5.85 (1.92) 5.83 (1.94) 5.87 (1.94)
Currently practised N 5.54 (1.74) 5.63 (1.67) 5.47 (1.81)
Internal audit services Y 5.60 (1.75) 5.51 (1.88) 5.68 (1.64)
Provided by auditor N 5.86 (1.96) 6.04 (1.71) 5.71 (2.18)
Audit of all companies in group Y 5.90 (1.67) 5.85 (1.97) 5.94 (1.35)
N 5.53 (2.00) 5.63 (1.65) 5.44 (2.27)
218 Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223
© AAANZ, 2002
Panel E Assist auditors to resist management pressure
All Auditors Directors
Audit partner rotation Y 6.29 (1.86) 6.75 (1.61) 5.87 (1.99)
Currently practised N 6.10 (2.06) 6.04 (2.23) 6.16 (1.87)
Internal audit services Y 6.01 (1.87) 6.17 (1.95) 5.88 (1.81)
Provided by auditor N 6.46 (2.02) 6.79 (1.93) 6.16 (2.08)
Audit of all companies in group Y 6.52 (1.85) 6.94 (1.90) 6.11 (1.73)
N 5.89 (2.00) 5.90 (1.88) 5.89 (2.12)
Y = Yes; N = No
Table 6
Continued
(as opposed to the appearance of independence). Alternatively, respondents may
believe that these variables do affect audit independence but that a lack of inde-
pendence does not necessarily lead to a low quality audit. Prior research suggests
that loan officers and financial analysts believe that independence is compromised
only when the same staff are involved in both internal audit services and the external
audit and when the audit firm provides internal audit services of a managerial nature
to the audit client (Lowe et al., 1999; Swanger and Chewning, 2001). Furthermore,

Swanger and Chewning (2001) found the provision of internal audit services by the
external auditor had no effect on the ultimate investment decision of financial
analysts. Taken together, these findings have important implications for regula-
tors who may need to re-examine their requirements in these areas.
6. Conclusion
By examining the views of two groups who are heavily involved with corporate
governance in Singapore, this study has provided insights into the perceived
effectiveness of certain corporate governance mechanisms in enhancing both
financial reporting and audit quality. We have contributed to the literature in
this area by focusing on mechanisms where prior research has been scarce or
where research findings have been mixed. The results of the study have implica-
tions for regulators and others who are concerned with establishing guidelines
and listing rules pertaining to corporate governance, especially in developing
capital markets.
A strong audit committee was found to have a significant impact on audit
effectiveness, on errors in financial statements and on the detection of manage-
ment fraud. However, this variable was not significant with regard to strength of
internal control or fraud prevention. Both the existence of a strong internal
audit function and a strongly enforced code of conduct are perceived to have a
significant impact on the strength of internal controls, on fraud and error in
financial statements and on audit effectiveness. However, external auditors have
Jenny Goodwin, Jean Lin Seow / Accounting and Finance 42 (2002) 195–223 219
© AAANZ, 2002
greater confidence in a strong internal audit function while directors see more
benefit from a strongly enforced code of conduct. It appears that more emphasis
may need to be placed on encouraging companies to establish their own internal
audit functions and to ensure that a strictly enforced code of conduct is in place.
However, the different perceptions of auditors and directors should be borne in
mind when making recommendations. The preferred governance mechanisms
may reflect a different emphasis on control issues, with auditors favouring

mechanisms that reduce the risk of fraudulent financial reporting and directors
favouring mechanisms that enable them to more directly monitor and control the
behaviour of management and employees.
The results of the study also suggest that, provided an internationally reput-
able firm is appointed as external auditor, neither directors nor auditors perceive
issues such as partner rotation and the provision of internal audit services as
impacting on audit effectiveness and independence.
A limitation of the study is that the research design asked participants to
role-play as members of a review committee and it is therefore possible that they
did not respond as directors or auditors respectively. As previously discussed,
this is unlikely in the Singapore context where review committees are made up
of representatives from interested parties whose responsibility is to put forward
the views of that party. However, as any bias resulting from role-playing would
presumably work against finding significant differences between auditors and
directors, a possibility remains that the design did not fully capture the true extent
of differences between the two groups.
There are a number of other limitations which should be borne in mind
when interpreting the findings of the study. Some of the non-significant results
relating to the audit committee may stem from the fact that, even in the weak
condition, the audit committee still met the minimum criteria set down in the
legislative and listing requirements of Singapore. As a result, the manipulation
of this variable may not have been strong enough. In addition, the small sample
size may have led to insignificant interaction effects between the variables
tested. On the other hand, multiple testing of the dependent variables may have
resulted in some significant findings being reported purely by chance. There is
also a risk of non-response bias, although tests suggest that this is likely to be
minimal. The experiment was not undertaken in a controlled environment and
this could have resulted in demand effects from participants either guessing the
manipulations or comparing instruments with peers. Some respondents may
not have taken the study seriously or may not have completed the instrument

themselves. It is hoped that this problem would be minimised by the fact that par-
ticipants are professionals whom we assume would act with integrity. As
participation in this study involved effort and time, the most obvious outcome
if a person did not take the study seriously would be a non-response. Further,
the respondents may not be representative of their respective groups and there
may also be cultural characteristics associated with auditors and directors in
Singapore that limit the generalisability of the results to other jurisdictions.