The Audit Company Governance Code
in the UK
Evidence on its influence on auditor independence and
audit quality
Author: Jeroen de Zoete
Date: 20 June 2016
Student number: 10254323
Supervisor: Prof. Dr. G. Georgakopoulos
MSc Accountancy & Control, specialization Accountancy
Faculty of Economics and Business, University of Amsterdam
2
Statement of Originality
This document is written by Student Jeroen de Zoete, who declares to take full responsibility for
the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other
than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of
the work, not for the contents.
3
ABSTRACT
This research examines whether the audit company governance code has influenced auditor independence and audit quality. More specifically, it examines if it affected the relation between auditor independence and audit quality, on audit quality alone and if the independency of the audit committee strengthen a possible effect of the Code. This is determined using the modified Jones model followed by a multiple regression model, using data between January 2007 and December 2015. It is expected that the Code has a significant effect on the relationship between auditor independence and audit quality. Furthermore, it is expected that the Code resulted in a significant higher degree of audit quality. At last, it is expected that the presence of independent audit committee members increases the expected positive relation between audit quality and the adoption of the Audit Company Governance Code. This study finds no significant evidence that the Code has a significant effect on the relationship between audit fees and discretionary accruals. Furthermore, it did not find evidence for the third hypothesis. However, the second hypothesis is significant supported by the obtained results and the Code resulted in significant higher audit quality.
4
Contents
1
Introduction ... 5
2
Literature Review ... 8
2.1
The Audit Company Governance Code ... 8
2.1.1
Background of the code ... 8
2.1.2
Features of the code ... 9
2.2
Auditor Independence ... 10
2.2.1
Definition of auditor independence ... 10
2.2.2
The threats to auditor independence ... 11
2.3
Audit quality ... 12
2.3.1
Definition of audit quality ... 12
2.3.2
The incentives for audit companies to provide high audit quality ... 13
2.3.3
Audit effort ... 14
2.3.4. Audit company size ... 15
2.4
Relation between auditor independence and audit quality ... 15
2.4.1
Influence of auditor independence on audit quality ... 15
2.5
Hypothesis development ... 16
3
The research methodology ... 19
3.1
Audit quality ... 19
3.2
Regression models ... 21
3.3
Sample selection ... 25
4
Descriptive statistics ... 27
5
Results and discussion ... 31
6
Conclusion ... 35
5
1 Introduction
The UK Financial Reporting Council (FRC) and the Institute of Chartered Accountants in England and Wales (ICAEW) issued a governance code for audit companies in January 2010. The Audit Company Governance project is the result of the recommendation of the October 2007 report of the Market Participants Group (MPG) which stated the following: ‘Every company that audits public interest entities should comply with the provisions of a Combined Code-style best practice corporate governance guide or give a considered explanation’ (MPG, 2007). The MPG was established by the UK Financial Reporting Council (FRC) to advise it on its work on Choice in the UK Audit Market. The ICAEW was invited by the FRC to support the follow-up to Recommendation 14 by drawing up a code and for that purpose formed the independent Audit Company Governance Working Group.
The purpose of the code is to promote confidence and choice in the UK audit market and to provide a formal benchmark of good governance against which companies, which audit listed companies, can report for the benefit of shareholders in such companies (ICAEW, 2010). The code was a response to concerns raised by market participants about the maintenance of a sufficient supply of high quality audits in a highly concentrated market following the collapse of Arthur Andersen (ICAEW, 2010). The code applies to eight audit companies (such as, EY, PwC, KPMG and Deloitte) that together audit about 95% of the companies listed on the main market of the London Stock Exchange, although adoption is not a regulatory requirement. The most significant changes the Code introduced were the appointment of Independent Non-Executives (INEs) and the dialogue between audit companies and shareholders in listed companies (ICAEW, 2010).
The introduction of the Code is seen as a helpful development which should improve audit company governance(ICAEW, 2010). All audit companies agreed the Code should continue to stay principles based and operate on a “comply or explain” basis. Audit quality should be the base of the Code but the majority of the companies were most enthusiastic that it also applied to the governance and oversight of the companies’ non-audit businesses – which collectively are the most important line of business to all of the companies which apply the Code and may therefore impact on the capacity to provide and/or quality of the audit service line (FRC, 2015).
The effectiveness of the Code’s comply or explain approach also depends on the companies’ leadership setting the right ‘tone at the top’ and taking the Code to heart rather than seeing it only as a cost of doing business. Audit companies can demonstrate their
6
commitment to good governance and the public interest by ensuring that their reporting does not become standard(ICAEW, 2010).
There have been numerous papers on auditor independence and audit quality. Nonetheless, the conclusions this past literature made on the latter variables, differ from one another. Some studies found no association between auditor independence and audit quality (e.g. Ashbaugh et al., 2003; Chung & Kallapur, 2003). Other papers concluded that higher audit fees resulted in increased audit effort which increased the audit quality (e.g. Frankel et al., 2002; Hoitash et al., 2007). Furthermore, there are also papers that found that increased audit fees resulted in auditors becoming more economic dependent on their client, which resulted in lower audit quality (e.g. Choit et al., 2010; Gul et al., 2003).
As the briefly discussed literature in the latter paragraph has pointed out, the existing literature has not universally agreed upon the precise effects of audit fees on audit quality. Additionally, there is no past literature that examined the effect of audit fees on audit quality prior and after the Code or what the effect of the Code was on audit quality. The code can be seen as efficient in accomplishing its objectives if either auditor independence or audit quality have positively developed, or both. The efficiency of the Code can be pointed out by an altered relationship between the variables that are examined. This thesis will contribute to current expertise about the subject matter by focusing on the following research question :
What influence did the Audit Company Governance Code of 2010 have on auditor
independence and audit quality?
The research question is examined in three separate parts using the modified jones model, first the effect of the Code on the relationship between auditor independence and audit quality is examined, then the effect on the audit quality and at last the effect of the independency of the audit committee on the possible effect of the Code. To examine these effects this study will look at UK companies listed on the FTSE 350 in the period 2007 till 2015.
From an academic point of view, this thesis will contribute to this area of expertise through various manners. The Audit Company Governance Code in the UK is an unique opportunity to research if voluntarily compliance to its provisions, results in an overall enhancement of auditor independence and audit quality of the Big four companies in the UK and if they should also apply this governance on a global basis since the Big four are all globally integrated organizations. Therefore, the various accounting companies could assess and acquire a knowledge from the governance code issued in the UK and take this into consideration in their own audit governance requirements. Furthermore, since the governance code is only put into action since the beginning of 2010 there is no evidence at
7
hand in today’s literature. As a result, merely theories and thoughts exist about the effects, which the Code is set up on. It would be worth knowing to find out if the FRC and the ICAEW constructed the proper provisions to achieve their planned objectives. Finally, this is also an research of interest, since this area of investigation is a highly talked about subject in present debates in the UK.
The remainder of this study is formed as the following. The second chapter examines the literature about the Audit Company Governance Code, auditor independence, audit quality and the relation between auditor independence and audit quality, ending with the formulation of the hypotheses. The third chapter examines the methodology applied to analyze the three hypotheses, providing a clarification for the sample that is worked with. The fourth chapter provides the descriptive statistics. Chapter five presents the obtained results with the discussion. The study closes with the conclusion in chapter six.
8
2 Literature Review
2.1 The Audit Company Governance Code
2.1.1 Background of the code
The market for large audits in the UK is ruled by four accounting companies and the possible threat of the withdrawal of a major company is of ongoing concern to the UK Financial Reporting Council (FRC) and many others since the collapse of Arthur Andersen. The FRC wanted to address audit problems from the banking crisis. It also hoped to keep clients’ choice of auditor as wide as possible by reducing the risk that a company might leave the market because it lost public trust.
The market participants Group was created by the FRC to advise it on its work on the choices in the UK Audit Market. The group made a recommendation in October 2007, which resulted in the Audit Company Governance Code. The recommendation stated that “every company that audits public interest entities should comply with the provisions of a Combined Code-style best practice governance guide or give a considered explanation’(ICAEW 2010, pp. 2-3).
The Institute of Chartered Accountants in England and Wales (ICAEW) was invited by the FRC to prepare the recommended code of the market participants Group. Following this invitation, the independent Audit Company Governance Working group (the Working Group) was set up by the ICAEW to fulfil and finish this work (ICAEW, 2010).
To accomplish its responsibilities, the Working Group wanted to secure a wide support for the Code through a proportionate way. By slowly seeking support, they wanted to demonstrate a practical application of evidence-based public policy making. The Working Group has conducted two wide-ranging formal consultations: the first to gather evidence on key issues to inform drafting of the Code; and the second to obtain views on a draft of the Code (ICAEW, 2010).
The Working Group decided that for the purposes of the Code, public interest entities should be defined as UK companies listed on the London Stock Exchange’s Main Market (ICAEW, 2010). This is in line with the definition used in UK legislation to implement EU requirements for audit company transparency reports. The Working Group recommends that the Code applies to large and medium-sized audit companies that each audit more than twenty listed companies(ICAEW, 2010).
9
The Code is created to act in the best interest of shareholders in listed companies that make use of the reports made by auditors. The Code helps companies in their aim of producing high quality audits that results in more reliable information to shareholders. The Code states that it is designed to play four major roles (ICAEW, 2010):
• enhance the stature of companies as highly visible exemplars of best practice governance; • enrich companies’ transparency reports;
• encourage changes in governance which improve the way that companies are run; and • strengthen the regulatory regime by achieving transparent and effective governance. 2.1.2 Features of the code
The construction of the code is almost the same as to that of the UK Corporate Governance Code (formerly the Combined Code on Corporate Governance) for listed companies. The Code consists of 20 principles and 31 provisions. In the Code it is stated that the provisions helps companies to implement the principles of the Code but does not relieve companies from their responsibility to take correct measures to apply the principles and embrace the spirit of the Code (ICAEW, 2010).
One of the principles on governance reporting states that a company ‘should publicly report how it has applied in practice each of the principles of the Audit Company Governance Code excluding the principles on shareholder dialogue and on informed voting and make a statement on its compliance with the Code’s provisions or give a considered explanation for any non-compliance.’ This is the approach which is summarised in the Code as the phrase ‘comply or explain’(ICAEW 2010, pp. 6-7).
The Working Group acknowledges that the audit companies are professional practices and already have responsibilities to act in a way that is in the best interest of the public. The Code point out that these responsibilities need to be supported by all members of a company’s management team and governance structures (ICAEW, 2010).
Furthermore, the Working Group acknowledges that audit is subject to extensive regulation and so, where appropriate, the Code makes reference to such regulation, for example in relation to standards on auditing, quality control and ethics and transparency reporting disclosure requirements. Companies are expected to integrate disclosures called for by the Code within the transparency reports published on their websites (ICAEW, 2010). The appointment of independent non-executives by the accounting companies is very important according to the Code. This reflects the view that regulation is not an alternative for effective governance and that good governance accompanies regulation in promoting audit
10
quality(ICAEW, 2010).The reputation and the continued existence of the company’s audit practice can be exposed to significant risks due to their shared operations and brands with businesses that are subject to little or no regulation. The Code foresees that independent non- executives play a role in helping to address those risks, as well as strengthening the confidence in the decision making of the company and make sure that stakeholder concerns are communicated correctly at the highest level(ICAEW, 2010).
Experience of applying the Code will be the basis for improving it and thereby further
enhancing audit company governance. It is with these objectives in mind that companies, listed companies and their shareholders are encouraged to give the Code their full support(ICAEW, 2010).
2.2 Auditor Independence
2.2.1 Definition of auditor independence
Independence is regarded as a vital condition for external auditors in forming their audit opinion. However, throughout the past century no mutually accepted meaning of the term “auditor independence” was decided in the accounting world. Antle (1984) critically argued that the SEC and AICPA guidelines on auditor independence were long and continuously reinterpreted. Following this critique and that of others as well, both regulative bodies stopped their efforts in presenting a mutually accepted definition for the phrase. This decision give rise to various definitions of auditor independence as stated by different academic literature. For example, DeAngelo (1981) stated auditor independence as the joint likelihood that when a material error has been identified, the auditor will also reveal the existing material error. Despite the fact that the Code of ethics states that “independence is not susceptible of precise definition”, the ICAEW in the U.K. still provided a general guide of auditor independence which they presented the following definition: “The independence of the external auditor. It is characterised by integrity and an objective approach to the audit process. The concept requires the auditor to carry out his or her work freely and in an objective manner” (ICAEW 2010, pp.8-9).
11
2.2.2 The threats to auditor independenceAuditor independence can be disturbed by a variety of bad influences. The International Ethics Standards Board for Accountants (IESBA) determined five different threats to the independence of the external auditor (IEASBA, 2013). The first threat according to the IESBA is the self-interest threat. This is the threat that the actions or judgements of the professional accountant is influenced by financial or other kinds of interests. This is one of the biggest concerns of regulators, and is generally involved with audit and non-audit fees. DeAngelo (1981) argues that because of the fees the auditor receives from the client, auditors prefer favouring clients instead than experiencing the possibility of losing the client due to dissatisfaction. Consequentially, this will lead to an economic bond between the external auditor and the client, which in turn endangers auditor independence since the auditor will be extra flexible with auditing for the client to maintain the client.
Next, the self-review threat is stated by the IESBA as the threat that a professional accountant will not appropriately check the results of a previous judgment made or service performed by the professional accountant, or by another individual within the professional accountant’s company or employing organisation, on which the accountant will rely when forming a judgment as part of providing a current service. This is also relevant for non-audit services provided by the auditor to a client, such as bookkeeping or giving advice. For the reason that the auditor is evaluating his own judgement when handling the audit for the same client. The auditor will not look as critical to his past opinion than to that of a third party, which could introduce flaws in the financial statements.
The following threat is the advocacy threat and concerns a professional accountant that will promote a client’s or employer’s position to the point that the professional accountant’s objectivity is compromised. For example, the auditor could act as an advocate for client in conflicts with third parties or invest in shares of the client (IESBA, 2013). An additional threat is the familiarity threat, due to a strong relationship with a client or employer, the professional accountant will be too compassionate to their concerns or willing to accept their work without proper judgement. For example, if an auditor turns close with a client, it’s possible the auditor does not want to make decisions that are not in the best interest of the client, despite the fact that it would be the most proper outcome.
The fifth and last threat is the intimidation threat, which is the possible situation that a professional accountant will be deterred from acting objectively because of actual or
12
perceived pressures, including attempts to exercise undue influence over the professional accountant. One of these pressures is the client intimidating the auditor with litigation charges or discharging the auditor. A different example is the rejection of a client to approve a non-audit service if the auditor is not in unison with the proposed accounting method of the client (IESBA, 2013).
Figure 1 : The different threats to auditor independence
From: http://kfknowledgebank.kaplan.co.uk/KFKB/Wiki%20Pages/Audit%20and%20compliance.aspx
2.3 Audit quality
2.3.1 Definition of audit quality
The description that is given by Arruñada and Paz-Ares (1997) to audit quality is that it is the result of the combined interaction of two separate variables. One of the two variables is professional competence, this concerns the ability of the auditor to identify material errors in the financial statements of the client. The other variable mentioned by the authors, is independence. This is the willingness of the auditor to communicate the identified material error. The function of an audit is to decrease information asymmetries that occur between managers and different stakeholders through miscommunication, that is accomplished by permitting third parties to concompany the reliability of financial statements (Francis et al., 1988; Watts and Zimmerman, 1981). A separate objective research of the financial statements is provided by the external auditor. The audit by the third party reinforces the value and trustworthiness of the financial statements presented by management, resulting in a higher
13
level of trust in the financial statement by users. The separate objective research also lowers investor risk and as such lowers the cost of capital of the preparer of the financial statements (Becker et al., 1998).
2.3.2 The incentives for audit companies to provide high audit quality
Francis and Wilson (1988) and Simunic and Stein (1987) argued that Big four accounting companies have a greater reputation to uphold and are therefore to a greater extent motivated to provide high quality. The paper of Weber et al. (2008) explains what the motivation is for audit companies to continuously keep audits at high quality. The authors explain that there are two separate motivations, which are the insurance and the reputation rationale. The concept behind the insurance rationale is that the audit company with more financial resources has a higher motivation to keep audits at high quality, so it can rule out the chance of litigation. The reasoning for this assumption, according to the paper of Weber et al. (2008), is that the more financial powerful companies are thought to be more probable to be taken to court, since their finances are better than that of smaller audit companies. The concept behind the second motivation, reputation rationale, is that companies with a great reputation are more probable to maintain their audits at high quality. When these companies provide a low quality audit for a client, the bad publicity following this failed audit may result in a loss in fee premiums and the possibility of losing clients, existing but also possible future clients.
Moreover, Becker et al. (1998) discusses that the Big four are considered to be affected the most by these two motivations, since they control most of the market. The results of the authors showed that financial statements that are audited by the Big four bear less abnormal accruals than those audited by non-Big four companies. Abnormal accruals as a measure of audit quality is a frequently used method in prior literature (e.g. Gul et al., 2003, Carey and Simnett, 2006). The theory behind this method is that management can affect financial statements by using accruals, the competence of the auditor to limit this specific event is considered to be a reliable measure of audit quality.
14
2.3.3 Audit effortThe result and quality of an audit highly depends on the amount of time spend by the auditor on the audit. When audit effort is increased or decreased by the auditor there emerges an effect on the probability that an existing material error is uncovered. However, there is no influence on the probability that the existing material error will be actually reported after an alteration of audit effort (Bell et al., 2008). Several past literature examined the influence of devoting a greater amount of effort and hours in an audit and the effects of such an adjustment on the quality of the audit (e.g. Caramanis & Lennow, 2008; Lobo & Zhao, 2013). The size of the client’s company and other complications were controlled for in this past literature, since the characteristics of the client greatly determines the required effort that is needed for the audit. Moreover, O’Keefe et al. (1994) carried out an empirical research to distinguish the various risk that come along with a specific audit and to show that audit effort is significantly affected by the client’s company size, complicacy, debt risk, internal control risk and company listing status. In addition, Bell et al. (2008) explained that, auditors will work with greater caution as a reaction to increased business risk and as part of their risk mitigation approach, which therefore would lead to an increase in audit effort.
The model that was created by Shibano (1990) links audit quality to the probability that misstatements are uncovered by the auditor. The model proved that the probability of undiscovered misstatements could be reduced by the auditor as a result of exercising additional audit effort. The models that were applied in the studies of Dye (1993) and Hillegeist (1999) concluded that auditors who are more driven and committed, have a higher probability to uncover earnings management. Furthermore, the model that was applied by Matsumura and Tucker (1992) tested the relation between several ranges of audit and fraud risk. They organized an experiment so that they were able to test the forecasts from their model. The results of their experiment showed that there exists a negative relationship between audit fees and fraud that was not discovered.
15
2.3.4. Audit company sizeThe size of the audit company that performs an audit, has a considerable influence in determining the audit quality. DeAngelo (1981) researched the influence of size of auditor companies on the quality of audits and concluded that there is an association between the two variables. Subsequently, Francis and Yu (2009) examined the impact of Big four companies on audit quality. The authors found that Big four companies maintain a higher quality in auditing. However, the results also revealed that there is a major difference between the results of small audit company’s performance against that of Big four companies. Moreover, Sundgren and Svanström (2011) investigated if audit quality and audit pricing differ between Big four audit companies and non-Big four audit companies. The authors used disciplinary sanctions as a measure of audit
quality.
The sample of their study consisted out ofBig
four auditors and non-Big four auditors. The results of the study determined a negative association between the chance of a sanction and the size of the audit office. In addition, the results revealed that audit fees support this pattern meaning that Big four companies maintain to have better quality.2.4 Relation between auditor independence and audit quality
2.4.1 Influence of auditor independence on audit quality
It is widely accepted in prior literature to use audit fees when connecting auditor independence to audit quality. The two explanations that Hay et al. (2006) discusses for this reasoning are the following, asses the competitiveness of audit markets and analysing concerns of contracting and independence that are connected to the audit operation.
The findings of Hoitash et al. (2007) suggested that there exist two effects in which audit quality is affected by audit fees. The authors discussed that audit effort rises when the auditor receives a higher amount of fee from the client, this in turn will proceed in a raise of the quality of the audit. The second effect that arises, is that the auditor is in a greater extent economically bounded to the client when earning a high audit fee. The high audit fee earned by the auditor is an important part of the total fees earned by the auditor. This results in the auditor being less independent because he wants to prevent losing this client and this will proceed into a lower audit quality (Hoitash et al., 2007; DeAngelo, 1981).
16
Furthermore, the study of Antle et al. (2006) examined the combined relationship between audit fee, non-audit fees and abnormal accruals. The authors discovered a significant positive relationship between audit fees and abnormal accruals. This suggests that if the external auditor earns a higher amount of fees, it will accept a higher amount of abnormal accruals from the client. In addition, the authors also concluded that there is not a positive significant relationship between non-audit fees and abnormal accruals. This result was in contrast with the findings of various alternative papers on the subject matter. For example, Frankel et al. (2002) stated that the independence of the auditor is in jeopardy in the situation where the auditor does not report or demands restatement when the client is able to manage earnings.
2.5 Hypothesis development
The various scandals in the previous century have raised the attention concerning the impact of audit fees on the quality of financial reports. Subsequently, this became the main inspiration for most of the researches concerning this discussion. The papers discussed in the previous chapter reveals that the various studies reached contrasting conclusions about the issues surrounding auditor independence and audit quality. Krishnan and Visvanathan (2008) and Lobo & Zhao (2013) concluded that audit effort is a positive influence on the quality of the audit, which would imply that audit fees have a positive influence on the quality of the audit. Meanwhile, various other papers that examined the subject matter, stated that audit fees have a negative relationship with auditor independence. These papers concluded this due to the fact that an economic bond could be created between the auditor and his client when audit fees increase and the auditor becomes more dependent on the increasing income since it represents a larger part of its total income (Gul et al., 2003).
The focus of this study is to examine whether the Audit Company Governance Code has had an influence on auditor independence and audit quality. More specifically, it examines if it influenced the relation between auditor independence and audit quality, on audit quality alone and if the independency of the audit committee strengthen a possible effect.
Since there is as of yet no conclusive knowledge about the exact effect of auditor independence on audit quality, or whether the Code was effective in achieving its goals, the first hypothesis is stated as follows
:
17
H1: The audit company governance code did have an influence on the relationship between auditor independence and audit quality.
This hypothesis is intended to examine the potential positive or negative effect of the Code on the relationship between auditor independence and audit quality. The Code could enhance the trust of the public and listed companies and provide more business(higher audit fees) or prevent a company from acting as auditor of an entity when there are relationships between an independent non-executive and that entity that a company audits (lower audit fees).
Furthermore, various gains on audit quality are expected from the Code. First off, the appointment of independent non-executives by accounting company, who through their involvement will collectively enhance shareholder confidence in the public interest aspects of the company’s decision making, stakeholder dialogue and management of reputational risks including those in the company’s businesses that are not otherwise effectively addressed by regulation. The audit quality is expected to increase through improved communication and discussion of information that is relevant. The communication among the auditor and the people imposed with the responsibility about governance is also expected to be improved by the Code. As a result of more communication of important audit concerns that are disclosed in the financial report of the auditor. An additional gain of the Code is the increase in professional scepticism through the increase attention of the auditor on stakeholder’s concerns that should be reported. This results in the second hypothesis:
H2: The audit company governance code has a positive influence on audit quality.
One of the provisions in the Audit Company Governance Code states that the company should establish board or other governance structures, with matters specifically reserved for their decision, to oversee the activities of the management team. Furthermore, the company’s governance structures and management team and their members should be subject to formal, rigorous and on-going performance evaluation and, at regular intervals, members should be subject to re-election or re-selection.
Several papers in the past have concluded that various governance characteristics have reduced the amount of earnings management (Klein, 2002; Xie et al., 2003). In addition,
18
a strong corporate governance system, such as the independence of the board, decreases management incentives and so the amount of earnings management. Bhuiyan et al. (2013) discussed that the importance of the audit committee in reducing the amount of earnings management will significantly increase in the future, and that effective monitoring will be most secure when the whole audit committee consists of independent directors since they will
not bias their judgement with personal objectives. These independent directors will submit to the objectives of the Code, as such, the hypothesis regarding the controlling effect of corporate governance on earnings management is as
follows :
H3: The presence of independent audit committee members increases the expected
positive relation between audit quality and the adoption of the Audit Company Governance Code.
19
3 The research methodology
3.1 Audit quality
Prior literature worked with various proxies for audit quality. The various methods that are applied to proxy for audit quality are the following: size of the audit company (Francis et al., 2009), the restatement of financial figures (Raghunandan et al., 2004), the declaration of going concern (Knechel and Vanstraelen, 2007) and earnings management (Klein, 2002; Frankel et al., 2002; Caramanis et al., 2008).
This study will measure audit quality using the modified Jones Model defined in Dechow et al. (1995). This is an earnings management model and Dechow et al. (1995) stated that the absolute value of the abnormal component is a method for the quality of earnings. Furthermore, the value of discretionary accruals has an negative relationship with the quality of the audit. The modified Jones Model states that all alterations in credit sales in the event period is due to earnings management. The model (Dechow et. al., 1995) applied in this paper is estimated as follows:
NDAi,t /Ai,t = (1/Ai,t-1) + (∆REVi,t – ∆RECi,t / Ai,t) + (PPEi,t)
Wherein the abbreviations mean the following:
Variables
Description
NDAi,t Non-discretionary accruals of company i in year t
Ai,t-1 Total assets at t-1
∆REVi,t Changes in revenue of company i from year t to year t-1.
∆RECi,t Changes in receivables of company I from year t to year t-1
PPEi,t Gross property, plant and equipment of company I in year t.
20
The following model is used for calculating total accruals:
TACi,t = (1/Ai,t-1) + (∆REVi,t) + (PPEi,t) + νt
Wherein the abbreviations mean the following:
Variables
Description
TACi,t Total accrual
s scaled by lagged total
assets
Ai,t-1 Total assets at t-1
∆REVi,t Changes in revenues of company I from year t to year t-1
PPEi,t Gross property plant and equipment of company I in year t
νt The residual, which represents the company-specific discretionary portion of
total accruals
The following model is used for calculating discretionary accruals:
DAi,t = (TACi,t /Ai,t-1) – (NDAi,t /Ai,t-1)
Wherein the abbreviations mean the following:
Variables
Description
DAi,t Discretionary accruals of company i in year t
TACi,t Total accrual
s scaled by lagged total
assets
Ai,t-1 Total assets at t-1
NDAi,t Non-discretionary accruals of company I in year t
21
3.2 Regression models
There are three different regression models applied to test the various hypotheses. In all regressions discretionary accruals are applied as a proxy for audit quality. The following regression model is applied to test the first hypothesis:
DAi,t=α+β1CODE+β2AUDITFEE+β3LEV+β4BIG4+ β5CFO + β6LOSS+ β7ROA+ β8SIZE + ε (1)
wherein the abbreviations mean the following:
Variables
Description
DAi,t Discretionary
accruals of company i in
year t
CODE Dummy variable one if the audit report confirms with the Audit Company Governance Code and zero otherwise AUDIT FEE Logarithm of the audit fee that was paid by
the client
LEV The ratio of total liabilities divided by total assets in
year t
BIG 4
Dummy variable one if audit company
is a Big 4
audit company and zero otherwiseCFO Scaled cash flow from operations (= cash flow from operations / total assets) LOSS Dummy variable equals one if the
company’s net income is negative, otherwise zero
ROA Return on Assets
SIZE The size of the company based on the natural logarithm of total assets of
company i
ε Error term
To examine the effect of the Audit Company Governance Code on the relationship between audit quality and auditor independence, the dummy variable CODE will be used. The dummy variable displays ‘one’ if the auditor’s report implemented the Audit Company Governance
22
Code from 2010 and ‘zero’ if the audit report does not implement the Code. Furthermore, the variable AUDIT FEE will be used for the examination of the first hypothesis. The natural log of the audit fees is taken so this variable is normally distributed.
There are various elements that could potentially alter the relationship between auditor independence and audit quality. For this reason, multiple control variables are applied to be sure if the statistical relationship is significant. The first control variable is leverage (LEV). Leverage is defined by Ashbaugh et al. (2003) as the ratio of total liabilities to total assets in year t. This control is applied to the regression because according to Watts (2003) companies with an increased value of leverage are identified as companies that have more difficulties with debt-holder and shareholder clashes. The assumption made in this study is that the value of LEV increases discretionary accruals and therefore decreasing the audit quality. Since, there is a possibility that discretionary accruals are exploited to control external financing, and that a more optimistic asset structure and company performance can be presented to the stakeholders(Watts, 2003).
The second control variable relates to whether a Big four company did the audit of a company (BIG 4). Prior literature (Francis and Yu, 2009; Caramanis & Lennow, 2008) have presented that the audit quality of big four audit companies is greater than of non-big four companies. Moreover, Big four audit companies devote a large amount of capital to their reputation and maintain their reputation through producing high-quality audits. Therefore, a dummy variable is applied that is ‘one’ in the case that the company was audited by a Big four audit company and ‘zero’ otherwise. This study expects that a Big four auditor will decrease discretionary accruals.
CFO is calculated as the cash flow from operations divided by the total assets of company i in year t (Hoitash et al.,2007). Numerous study include the variable in order to control for companies with high cash flow (Hoitash et al.,2007; Knechel et al., 2003). A negative relation is expected between discretionary accruals and the cash flow from operations. Since companies with high cash flows are more prone to beat earnings benchmarks and purchase greater outside consulting services (Hoitash et al.,2007; Knechel et al., 2003).
The LOSS variable is a dummy variable, which represents ‘one’ if company i reports a negative net income in year t and otherwise ‘zero’. This study expects that the value for LOSS lowers discretionary accruals, on account of companies with negative earnings are discouraged to manage earnings according to the study of Francis and Yu (2009).
The control variable return on assets (ROA) expresses the performance of a company in the efficiency of the usage of assets. A better ROA expresses better company performance
23
since it shows that the companies’ assets are well used. There are no incentives, to exploit discretionary accruals in order to present a more optimistic asset structure and company performance when the return on assets are already high. Therefore, this study expects that the level of ROA has a negative influence on discretionary accruals (Carcello and Li, 2013). The final control variable concerns the size of the company (SIZE), which is
presented as the natural log of the total assets. This control variable was also used in several previous studies (Becker et al., 1998; Gul et al., 2003). It is expected that the size of the company will increase the amount of discretionary accruals. Since bigger companies are more prone to manage earnings because these companies undergo greater public scrutiny than smaller companies (Gul et al., 2003; Hay et al., 2006).
The second hypothesis is examined through the following model:
DAi,t = α + β1CODE + β2LEV + β3BIG 4+ β4CFO + β5LOSS+ β6ROA + β7SIZE + ε (2)
wherein the abbreviations mean the following:
Variables
Description
DAi,t Discretionary
accruals of company i in
year t
CODE Dummy variable ‘one’ if the audit report confirms with the Audit Company Governance Code and ‘zero’ otherwise AUDIT FEE Audit fee that was paid by the client
BIG 4
Dummy variable ‘one’ if audit company
is a Big 4
audit company and ‘zero’ otherwiseCFO Scaled cash flow from operations (= cash flow from operations / total assets) LOSS Dummy variable equals ‘one’ if the
company’s net income is negative, otherwise ‘zero’
ROA Return on Assets (=Net income / total assets)
SIZE The size of the company that is based on the natural logarithm of total assets
24
To examine the effect of the Audit Company Governance Code on audit quality, the dummy variable CODE is applied. The regression is nearly identical as the one stated above apart from the fact that the variable AUDIT FEE is excluded.
The third hypothesis is examined through the following model:
DAi,t = α + β1CODE + B2CG + β3LEV + β4BIG 4 + β5CFO + β6LOSS + β7ROA + β8SIZE + ε (3)
Wherein the abbreviations mean the following:
Variables
Description
DAi,t Discretionary accruals of company i in year t
CODE Dummy variable ‘one’ if the audit report concompanies with the Audit Company
Governance Code and ‘zero’ otherwise CG Dummy variable for corporate governance,
which is ‘one’ in the presence of both the proxies audit committee (AUD) and the independence of the members of the audit
committee (AUDIN) and ‘zero’ otherwise AUD Dummy variable for audit committee,
which is ‘one’ in the presence of an audit committee and ‘zero’ otherwise AUDIN This reflects if the members of the audit
committee are independent, which is ‘one’ if present and ‘zero’ otherwise
LEV Ratio of total liabilities divided by total assets in year t
BIG 4 Dummy variable ‘one’ if audit company is a Big 4 audit company and ‘zero’ otherwise CFO Scaled cash flow from operations (= cash
flow from operations / total assets) LOSS Dummy variable equals ‘one’ if the
company’s net income is negative, otherwise ‘zero’
ROA Return on Assets(=Net income / total assets)
SIZE The size of the company based on the natural logarithm of total assets
25
To examine the effect of the presence of independent audit committee members on the relationship between audit quality and the adoption of the Audit Company Governance Code, the dummy variable CG will be used. This variable is reflected by the presence of the audit committee (AUD) and the independence of the members of the audit committee (AUDIN). The CG variable will have a value of ‘one’, when the two proxies are present and a value of ‘zero’ otherwise. Klein (2003) states that the audit committee has an impact on the performance of the company, which makes it more probable that the structure of a board committee will affect management’s desire to manage earnings. Therefore, it is expected that the variable CG will have a negative influence on discretionary accruals. The remainder of the regression is identical to the regression for the second hypothesis.
3.3 Sample selection
The final sample consists of 1384 observations from 173 companies in the United Kingdom, from the London Stock Exchange FTSE 350 index. The FTSE 350 is chosen because this index contains the largest capitalized companies of the London Stock Exchange and it is most probable that these companies provide the required data for this research. As the Audit Company Governance Code is applied to audit companies that have clients that are listed on the Stock Exchange. The database ’Compustat’ is used for the collection of financial data, which is used for the measure of discretionary accruals and control variables data regarding audit fees and corporate governance are collected from the database Datastream. The database Compustat can be seen as the beginning source for the collection of data. Finally, the software STATA is used to combine the different databases.
The initial sample consisted of 3500 observations on the Compustat database from January 1, 2006 till January 1, 2016. Previous related studies do not include data that is from the financial sector or companies that provide insurance, because their financial ratios are not similar to that of companies that are in the non-financial sector, because it’s possible that this would lead to biased results (Chung and Kallapur, 2004; Carey & Simnett, 2006). Therefore, companies in the financial sector are not included in this research. The remaining sample consisted of 2290 observations with 229 companies after deducting 1210 observations
26
of 121 companies in the financial sector from the previous sample.
Earnings management is a proxy for audit quality that according to facts does not have any limits to which companies can be included. Almost all companies can have discretionary accruals and therefore there is no need for specific company characteristics. The collection of the sample is therefore for the main part confined by the availability of data and the observations that did not disclose the necessary parts for the variables in the regressions were excluded. This resulted in a remaining sample of 1730 observations from 173 companies. For the variable CODE, a pre-Code and post-Code had to be determined. As the Code was issued in January 2010 after a recommendation in October 2007, the pre-Code period is set at 2007-2009 and the post-Code period is set at 2011-2015. The exclusion of the year 2010 after the installment of the Code is intended to make certain that there are no problems which may arise in the first year after installment of the Code due to companies not fully acquainted with the objectives of the Code. Therefore, 346 observations from the remaining 173 companies were deleted which resulted in a final sample of 1384 observations of 173 companies.
TABLE 1
Final sample set for all Hypotheses
Data origin Companies Observations
FTSE 350 350 3500
Financial sector (-121) (-1210) Data missing in Compustat (-56) (-560)
Data from 2006 and 2010 0 (-346)
Final sample 173 1384
27
4 Descriptive statistics
The tables 2a and 2b offers an overview of the descriptive statistics of the discretionary accruals, the independent variables and the control variables for respectively the samples before and after the implementation of the code. All variables have gone through winsorization at the 1st and 99th percentile. Furthermore, the tables below show that the whole sample is composed of 1384 observations, 519 observations (37.5%) are made up of data before the issuance of the Code and the remaining 865 observations (62.5%) are made up of data after the issuance of the Code.
TABLE 2a
Descriptive statistics before the Code (2007-2009) Before the code (n=519)
Variables Min Max Mean Std. Dev.
DAi,t (-0,347) 0,299 0,013 0,047 CG 0,000 1,000 0,971 0,168 AUDIT FEE 4,443 11,399 7,359 1,273 LEV 0,021 1,418 0,625 0,202 BIG 4 0,000 1,000 0,983 0,131 CFO (-0,095) 0,550 0,118 0,078 LOSS 0,000 1,000 0,095 0,294 ROA (-0,786) 0,622 0,067 0,083 SIZE 3,594 12,684 7,579 1,590 TABLE 2b
Descriptive statistics after the Code (2011-2015) After the code (n=865)
Variables Min Max Mean Std. Dev.
DAi,t (-0,142) 0,232 0,010 0,032 CG 0,000 1,000 0,930 0,255 AUDIT FEE 2,773 11,102 7,347 1,275 LEV 0,027 1,254 0,585 0,191 BIG 4 0,000 1,000 0,983 0,131 CFO (-0,111) 0,546 0,111 0,075 LOSS 0,000 1,000 0,067 0,251 ROA (-0,606) 0,458 0,072 0,070 SIZE 4,760 12,795 7,867 1,503
At first glance, the samples in the tables above appear to be very much alike. However, there are some deviations between the two periods that are given. Such as, the average amount of discretionary accruals decreased in the time period 2011-2015 compared to 2007-2009.
28
Furthermore, the minimum discretionary accrual during 2011-2015 was almost three times the minimum than during the former time period. The latter observations are compatible with the alteration that occurred in the fees, since the average audit fee declined as well in the latter time period and the minimum was nearly the half compared to that of 2007-2009. This was to some extent anticipated because the Code hinders a company from operating as auditor of an entity when there are relationships between an independent non-executive and that entity that a company audits (lower audit fees) and the communication among the auditor and the people imposed with the responsibility about governance is also expected to be improved by the Code.
An additional worth knowing remark is that the amount of companies that realized a loss declined from 9,5% during 2007-2009 to 6,7% during 2011-2015. Furthermore, the amount of companies audited by a Big four auditor remained the same during the whole time period of the sample. This result could be clarified by the fact that all companies in the sample are listed on the FTSE350 index, which is almost completely done by Big four auditors.
In addition, the tables 3a and 3b present an overview of the Pearson correlations in which the coefficients are significant at a significance level of 1%, 5 % or 10%.
TABLE 3a
Pearson correlation before the Code (2007-2009)
The dependent variable is the discretionary accrual as computed by the modified Jones model, where *, ** and *** denote significance at the 10%, 5% and 1% levels, respectively
Variables DAi AUDIT FEE CG LEV BIG 4 CFO LOSS ROA SIZE
DAi 1 AUDIT FEE (-0,064)* 1 0,0941 CG 0,0747** (-0,029) 1 0,0496 0,4521 LEV 0,0491 0,1322*** (-0,008) 1 0,1974 0,0005 0,8325 BIG 4 0,0377 0,1662*** (-0,023) 0,1423*** 1 0,3224 0,0000 0,5473 0,0002 CFO (-0,047) (-0,111)*** (-0,045) (-0,0388) 0,0358 1 0,2148 0,0035 0,2403 0,3077 0,347 LOSS (-0,105)*** (-0,058) 0,056 0,0547 (-0,070)* (-0,199)*** 1 0,0057 0,1313 0,141 0,1505 0,066 0,0000 ROA 0,0795** (-0,069)* (-0,013) (-0,1538)*** 0,0464 0,5721*** (-0,530)*** 1 0,0394 0,077 0,7469 0,0001 0,2299 0,0000 0,0000 SIZE (-0,075)** 0,7815*** (-0,034) 0,1352*** 0,0866** (-0,187)*** 0,017 (-0,1902)*** 1 0,0474 0,0000 0,3725 0,0004 0,0227 0,0000 0,6561 0,0000
29
TABLE 3b
Pearson correlation after the Code (2011-2015)
The dependent variable is the discretionary accrual as computed by the modified Jones model, where *, ** and *** denote significance at the 10%, 5% and 1% levels, respectively
Variables DAi AUDIT FEE CG LEV BIG 4 CFO LOSS ROA SIZE
DAi 1 AUDIT FEE (-0,0574)* 1 0,0929 CG (-0,0083) (-0,0216) 1 0,8084 0,5265 LEV 0,0143 0,2499*** (-0,0399) 1 0,6747 0 0,2424 BIG 4 0,0189 0,2101*** (-0,0364) 0,1688*** 1 0,5793 0 0,2858 0 CFO (-0,0686)** (-0,1391)*** (-0,0071) (-0,0819)** 0,0772** 1 0,0439 0 0,8351 0,0161 0,0234 LOSS (-0,153)*** 0,0998*** 0,0371 0,1127*** 0,0357 (-0,154)*** 1 0 0,0034 0,277 0,0009 0,2945 0 ROA 0,0431 (-0,2125)*** 0,0272 (-0,262)*** (-0,021) 0,6429*** (-0,483)*** 1 0,2114 0 0,4311 0 0,5419 0 0 SIZE (-0,121)*** 0,7834*** (-0,0614)* 0,2118*** 0,0974*** (-0,223)*** 0,1252*** (-0,3018)*** 1 0,0004 0 0,0715 0 0,0042 0 0,0002 0
Table 3a and 3b displays Pearson correlation matrices for the variables applied in the regressions. The coefficients between the variables in the tables above present the extent to which they are dependent with each other. Francis et al. (1988) pointed out that when the correlation coefficient between two variables exceed 0,80, there is a high possibility of multicollinearity and this would bias the regression. Multicollinearity is stated as a phenomenon in statistics in which two or more predictor variables in a multiple regression model are strongly correlated, implying that one can be linearly predicted from the others with a non-trivial degree of accuracy (Francis et al., 1988). However, none of the variables in the tables show a correlation above 0,8 and suggest that there are no multicollinearity problems in this study.
The tables display that there occurs a strong correlation between the variables AUDITFEE and SIZE with a coefficient of respectively 0,7815 and 0,7834. This high correlation suggests that bigger companies in general are associated with paying a much higher amount in audit fees in comparison to companies with a relative small value for SIZE. More strong correlated variables by comparison in either matrices are CFO and ROA with a coefficient of respectively 0,5721 and 0,6429, only this was anticipated, because both variables denote a somewhat similar variable as ratio of total assets. Furthermore, both latter variables have as well a strong correlation with LOSS. However, this is likewise anticipated
30
since income is the essential feature of these three variables. The dependent variable is in either table not strongly correlated with any of the other independent variables that are displayed, none of the coefficients shown are greater than 0.20. The highest correlation in table 3A is with ROA with 0,0795 and (-0,1210) in table 3B with SIZE, what demonstrates a highly powerless to insignificant correlation.
In both matrices,
the positive correlation of LEV follows the results of previous studies since the likelihood of not being able to meet debt contract requirements is one of the primary incentives to control external financing and present a more optimistic asset structure and company performance to the stakeholders in the accounting studies (Watts, 2003). Moreover, The positive coefficient between the ROA and the dependent variable, and the negative coefficient between SIZE and the dependent variable are not following the results of previous studies. Actually, previous studies concluded that there is no necessity, and therefore no incentives, to use discretionary accruals in order to control external financing when the values of ROA are already high (Bhuiyan et al. 2013). Additionally, O’Keefe et al. (1994) stated that there is a positive correlation between SIZE and DAi because bigger companies are more prone to manage earnings because these companies undergo greater public scrutiny than smaller companies (Gul et al., 2003; Hay et al., 2006)31
5 Results and discussion
The first hypothesis is tested with a regression that includes the variable AUDIT FEE. By looking at the explanatory powers of the Code in the ascertainment of the discretionary accruals, the regression results assist in analysing the first hypothesis to check whether the Audit Company Governance Code has an effect on the relationship between audit quality and auditor independence.
Table 4a Regression Hypothesis 1
The dependent variable is the discretionary accrual as computed by the modified Jones model. The regression results are obtained with OLS regressions, where *, ** and *** denote significance at the 10%, 5% and 1% levels, respectively.
Table 4a gives an overview of the outcomes of the regression output. The R2 of the regression is 0,0451 and this low value implies that the explanatory power of the regression is limited as it only explains 4,51% of the DAi. A negative value of -0.004 is found for the dummy variable CODE when taking control variables into account. This indicates that degree of earnings management lowered with -0.40% during the period after the implementation of the Code. This observation is in line with the expectations resulting from the literature review. The negative value is significant at 10%. Furthermore, the variable AUDIT FEE indicates an increase of 0,00013 with the amount of audit fee. Since the combined coefficients of the latter two variables results in a negative effect of 0,387%((0,004-0,00013)*100) to the discretionary accruals, this would indicate that the Code had an effect on the relationship between audit quality and auditor independence. Decreasing the negative effect that audit fees have on the level of earnings management. However, the variable AUDIT FEE is highly insignificant with a p-value of 0,920. Which means that there was no significant relationship
Hypothesis 1
Variables Coef. p-value
CODE (-0,004)* 0,065 AUDIT FEE 0,00013 0,920 LEV 0,013** 0,019 BIG 4 0,011 0,155 CFO (-0,082)*** 0,000 LOSS (-0,015)*** 0,000 ROA 0,047** 0,017 SIZE (-0,003)** 0,012 Constant 0,024*** 0,009 Number of Observations 1384 R-squared 0.0451
32
between audit fee and earnings management in both periods of time. The first hypothesis is therefore rejected, the Code did not have an effect on the relationship between the two variables. There is no empirical evidence that the Code influenced the relationship between auditor independence and audit quality.
The regression used for the second hypothesis excludes the variable AUDIT FEE. This regression tests the second hypothesis, whether the Audit Company Governance Code has had a significant positive effect on the audit quality.
Table 4b Regression Hypothesis 2
The dependent variable is the discretionary accrual as computed by the modified Jones model. The regression results are obtained with OLS regressions, where *, ** and *** denote significance at the 10%, 5% and 1% levels, respectively.
Table 4b gives an overview of the outcomes of the regression output. The R2 of the regression is 0.0451 and this low value implies that the explanatory power of the regression is limited. The explanatory power of this regression is the same as the regression used for the first hypothesis due to the same amount of observations used. The relation between CODE and DAi is statistically significant at the 10% level with a coefficient of -0,004 and a p-value of 0.052 when taking control variables in to account. This result indicates that the Code decreases the amount of discretionary accruals and accordingly decreasing earnings management. The Code has a significant positive influence on the level of audit quality after its implementation in 2009 and the second hypothesis is accepted. The obtained result corresponds to the findings of the literature review. Furthermore, it contemplates with the several benefits that were expected from the Code in improving audit quality. Such as, the appointment of independent non-executives by accounting company, the enhanced communication and discussion between the auditor and those charged with governance and the revived attention of the auditor on concerns to be reported, which indirectly results in an increase in professional scepticism, among other contributors to audit quality.
Hypothesis 2
Variables Coef. p-value
CODE (-0.004)* 0,052 LEV 0,012** 0,022 BIG 4 0,011 0,147 CFO (-0,078)*** 0,000 LOSS (-0,016)*** 0,000 ROA 0,044** 0,025 SIZE (-0,003)*** 0,000 Constant 0,025*** 0,006 Number of Observations 1384 R-squared 0,0451
33
The regression used for the third hypothesis excludes the variable AUDIT FEE and includes the dummy variable CG(corporate governance). This regression tests the third hypothesis, whether the presence of independent audit committee members increases the expected positive relation between audit quality and the adoption of the Audit Company Governance Code.
Table 4c Regression Hypothesis 3
The dependent variable is the discretionary accrual as computed by the modified Jones model. The regression results are obtained with OLS regressions, where *, ** and *** denote significance at the 1%, 5% and 10% levels, respectively.
Table 4c gives an overview of the outcomes of the regression output for the third hypothesis. The R2 of the regression is 0,0457 and this low value implies that the explanatory power of the regression is limited. The explanatory power of this regression is 0,005% higher than the regression used for the first and second hypothesis due to the probable effect that CG is added which is a combined result of AUD and AUDIN. However, the distinction in the R-squared is so modest that it doesn’t provide a remarkable upper hand to the explanatory power of the latter regression. A positive value of 0,004 is found for the dummy variable CG when taking control variables into account. This would indicate that the presence of an independent audit committee would increase discretionary accruals with 0,4% and thus the level of earnings management, thereby diminishing the positive effect that the Code has. This observation would not be in line with expectations resulting from the literature review. However, the coefficient for CG is insignificant with a value of 0,321 and the third hypothesis is rejected. There is no empirical evidence that the presence of independent audit committee members increases the expected positive relation between audit quality and the adoption of the Audit Company Governance Code.
Additionally, it was expected that the control variable LEV would have a positive effect on DAi. The significant positive coefficient in all regressions indicates that companies with an increased value of leverage are identified as companies that have more difficulties with
Hypothesis 3
Variables Coef. p-value
CODE (-0,004)* 0,065 CG 0,004 0,321 LEV 0,012** 0,021 BIG 4 0,011 0,141 CFO (-0,078)*** 0,000 LOSS (-0,016)*** 0,000 ROA 0,043** 0,000 SIZE (-0,003)*** 0,028 Constant 0,021** 0,045 Number of Observations 1384 R-squared 0,0457
34
debt-holder and shareholder clashes. (Watts, 2003).The negative coefficients for CFO in all regressions are in line with the negative relation that was expected between discretionary accruals and the cash flow from operations (Hoitash et al.,2007; Knechel et al., 2003). The companies with high cash flows are more prone to beat earnings benchmarks and purchase greater outside consulting services.
Furthermore, the negative coefficients found for LOSS in all regressions correspond with the expectation that a loss in the net income will have a negative effect on discretionary accruals, because companies with negative earnings are discouraged to manage earnings (Francis and Yu, 2009).
The control variable ROA was expected to have a negative impact on discretionary accruals (Carcello and Li, 2013). However, all regressions show significant positive coefficients for ROA. This effect indicates that as the return on assets are higher, the discretionary accruals and thus the amount of earnings management will increase before the adoption of the Code. This effect is moreover not in line with the rationale that, there are no incentives to use discretionary accruals in order to present a more optimistic asset structure and company performance when the return on assets are already high (Bhuiyan et al., 2013). The final control variable is SIZE and shows for every regression a negative value, which is line with expectations risen from the literature review (Becker et al., 1998; Carey and Simnett, 2006 and Gul et al., 2003). This indicates that bigger companies are more prone to manage earnings because these companies undergo greater public scrutiny than smaller companies (Gul et al., 2003; Hay et al., 2006)
35
6 Conclusion
The UK Financial Reporting Council (FRC) and the Institute of Chartered Accountants in England and Wales (ICAEW) issued a governance code for audit companies in January 2010. The purpose of the code is to promote confidence and choice in the UK audit market and to provide a formal benchmark of good governance against which companies which audit listed companies can report for the benefit of shareholders in such companies. The code was a response to concerns raised by market participants about the maintenance of a sufficient supply of high quality audits in a highly concentrated market following the collapse of Arthur Andersen. The existing literature has not universally agreed upon the precise effects of audit fees on audit quality. Additionally, there is no past literature that examined the effect of audit fees on audit quality prior and after the Code or what the effect of the Code was on audit quality. More specifically, it examines if it influenced the relation between auditor independence and audit quality, on audit quality alone and if the independency of the audit committee strengthen a possible effect.
The modified Jones Model in the study has been applied to compose the discretionary accruals for 1384 company observations in the period 2007-2015. The obtained DA’s are used as dependent variable in the regressions to find evidence in support for the three hypotheses. For the first hypothesis, this study found that the Code had a positive effect on the relationship between auditor independence and audit quality. However, this result is not significantly different from zero. Based on this finding the first hypothesis is rejected. There is no empirical evidence that the Code influenced the relationship between auditor independence and audit quality.
Secondly, the regression for the second hypothesis found empirical evidence that the Code has had a positive influence on audit quality. The study revealed that the suggested benefits of the Code resulted in a rise in audit quality. The appointment of independent non-executives by the accounting companies, the enhanced communication and discussion between the auditor and those charged with governance and the revived attention of the auditor on concerns to be reported, which indirectly results in an increase in professional scepticism, among other contributors to audit quality.
Finally, the results for the third hypothesis indicated that the presence of an independent audit committee would increase discretionary accruals with 0,4% and thus the level of earnings management, thereby diminishing the positive effect that the Code has. This observation is not in line with expectations resulting from the literature review. However,
36
this result was not statistically different from zero. Based on this finding the third hypothesis is rejected. There is no empirical evidence that the presence of independent audit committee members increases the expected positive relation between audit quality and the adoption of the Audit Company Governance Code.
This study has several limitations. Firstly, the regression analysis for every hypothesis had a small explanatory power, this significantly lowers the quality of this research. Increasing the amount of control variables and working with a different proxy for audit quality would possibly significantly enhance the power of the model. Additional alternative models can be applied to determine the discretionary accruals, such as for example, the DeAngelo model. The second limitation is that there does not exist a possibility to analyze data of additional years since the issuance of the Code. It is probable that the audit companies and others have to become more acclimate to the new governance code and require a longer period of time to familiarize with them before it will result in higher audit quality and auditor independence. This could be addressed in future research.
Future research could also focus more on additional countries carrying out the same kind of governance regulations and examining the dissimilarities of effect in various economic and political environments.
