MSc Accountancy & Control
Track: Control
Corporate Governance Code
Does compliance with the UK corporate governance code affect the corporate
governance quality?
Name: Jamie Stap
Student number: 10001524
Date: 21-6-2015
Word count: 12.904
Thesis supervisor: Dr. Réka Felleg
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam
Statement of originality
This document is written by student Jamie Stap, 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 Economic and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
I examine whether compliance with the UK Corporate Governance Code is associated with corporate governance quality. The motivation behind this study are the concerns expressed by the audit committee institute that compliance, for compliance‟s sake could harm the board of
directors, business operations and shareholders. .
No significant relation is found between compliance with the UK Corporate
Governance Code and corporate governance quality. However, when looking at compliance with the separate provisions, a significant negative relationship is found when the chairman of the board holds meetings with non-executive directors without the executive directors present and when the audit committee consists solely of independent members. Also, a significant positive relation is found when board performance is externally evaluated at least once every three years by an independent third party.
These results partially confirm the Audit Committee Institute‟s concerns that
compliance, for compliance‟s sake could harm the board of directors, business operations and shareholders.
Table of Contents
1.Introduction ……… 1
2. Literature Review and Hypothesis ...………. 3
2.1 Corporate Governance ………...……….. 3
2.2 The Corporate Governance Code………. 4
2.3 The U.K. Corporate Governance Code………. 5
2.3.1 “Comply or Explain” Principle……….. 6
2.4 Board of Directors………. 7
2.5 Audit Committee……….. 9
2.6 The nomination and remuneration committee……….. 11
2.7 Hypothesis………. 11
3. Methodology………... 12
3.1 Sample………... 12
3.2 Research Design……… 13
3.3 Corporate Governance Quality……….. 16
3.4 Earnings Management………... 17 4. Results………. 19 4.1 Descriptive Statistics………. 19 4.2 Hypothesis Tests……… 23 5. Conclusion………... 28 6. References………... 31
Appendix 1 Examined Provisions………... 34
Appendix 2 Hand-Collecting Process………. 37
Appendix 3 Skewness and Kurtosis Test……… 39
Appendix 4 Shapiro-Wilk Test……… 39
Appendix 5 White‟s General Test………... 39
1 1.Introduction
This paper examines whether compliance with the UK Corporate Governance Code is associated with corporate governance quality. This question derives from the results and statements of prior literature: (1) The UK Corporate Governance Code (2010) with its
standards and provisions regarding good corporate governance quality, which are perceived as “good”, “responsible” and represent a form of “best practice”; (2) The flexibility that the UK Corporate Governance Code offers with the “comply or explain” principle, thereby enabling companies to deviate from the Code if it leads to good corporate governance quality (UK Corporate Governance Code, 2010); (3) Concerns raised by the Audit Committee Institute (2008) regarding the sustainability of the “comply or explain” principle. They find that companies increasingly seek compliance just for compliance‟s sake, which could eventually even harm the board of directors, business operations and shareholders; (4) Bhuiyan et al. (2013) who finds a positive relationship between the level of compliance and corporate
governance quality thereby contradicting the Audit Committee Institute‟s concerns; (5) Mixed results in papers such as Xie et al. (2001), Brown and Caylor (2004), Abbott et al. (2004) and Krishnan (2005) who found a positive relationship between the level of board independence and corporate governance quality which is in line with the UK Corporate Governance Code, as opposed to Vafeas and Theodorou (1998), Turley and Zaman (2004) and Klein (2006) who found no positive relationship between the level of board independence and corporate
governance quality which contradicts the UK Corporate Governance Code. Based on these prior findings I expect to find a relationship between compliance with the UK Corporate Governance Code and corporate governance quality although I could not predict whether this is a positive of a negative relationship.
In order to investigate this relationship, I examine the compliance rate of FTSE 350 listed companies with the UK Corporate Governance Code and compare this with their corporate governance quality. The compliance rate is roughly divided in 2 categories, full compliance versus not full-compliance, and the data regarding the compliance rate is hand-collected. The hand-collection process consisted of manually checking the 2013 annual reports to examine whether they were in compliance with certain provisions, 11 provisions in total, or not. Companies were classified as full-compliance companies if they complied with all the 11 provisions. Non-compliance with any of the 11 provisions resulted in a
classification as not full-compliance company. It is not possible to predict the outcome of the relationship between the compliance rate and corporate governance. According to the UK Corporate Governance Code (2010) compliance should be positively related to corporate
governance quality while it is theoretically also possible that non-complying companies achieve a higher corporate governance quality due to the fact that they designed a unique and effective corporate governance system, one which is not covered by the UK Corporate Governance Code.
Corporate governance quality is measured by the level of discretionary accruals. Discretionary accruals are a form of earnings management and are calculated with the use of the Modified Jones model (Dechow et al., 1995). The motivation to use discretionary accruals as dependent variable is the implicit assertion by the SEC, the NYSE, and the NASDAQ that discretionary accruals and poor corporate governance quality are positively related (Klein, 2006).
The results of this paper show no relationship between the rate of compliance and corporate governance quality in general. However, when I look at the relationship between the rate of compliance with the separate provisions and corporate governance quality, I find that when the chairman of the board holds meetings with non-executive directors without the executive directors present and when the audit committee consists solely of independent members, the corporate governance quality tends to be lower. This could be explained by the fact that independent board members are from outside the company and are therefore
considered to have less knowledge about the company. This in turn compromises their ability to effectively carry out their oversight function and lowers the corporate governance quality. I also found that companies whose board performance are externally evaluated at least once every three years by an independent third party, show a higher corporate governance quality. A possible explanation for this would be the fact that an independent third party is less likely to provide an unjustified report in favor of the board since it is not economically dependent on the board‟s performance.
My results are important from a scientific point of view since it provides further insight regarding the consequences of the corporate governance code. First, this paper shows that compliance does not necessarily lead to an increased corporate governance quality and that the UK corporate governance code is still eligible for improvement. Second, although in general no negative relationship is found between compliance and corporate governance quality, a negative relationship is found between compliance with certain provisions and corporate governance quality. This confirms the concerns raised by the Audit Committee Institute (2008) that companies are seeking compliance for compliance‟s sake thereby potentially harming the board of directors, business operations and shareholders. Third, my research is one of the first that directly examines the effect that complying versus
non-3
complying has on corporate governance quality in the UK. Thereby hopefully providing a basis for future research.
My results are also important from a societal point of view. First, they contribute to the societal debate about corporate governance, the corporate governance code and its effectiveness/importance. Second, they are useful for several parties, such as shareholders which are affected by the UK Corporate Governance Code and the government which is the standard setter and needs feedback in order to improve the effectiveness of their regulations.
This paper is structured as follows. Section 2 provides an overview of the prior literature and ends with the formulation of the hypothesis. Section 3 introduces the methodology used in this paper. Section 4 presents and discusses the results and section 5 forms the conclusion.
2. Literature Review and Hypothesis 2.1 Corporate Governance
Corporate governance is present-day a frequent occurring and debatable topic in the media. News-sources of all kinds release/publish detailed information about corporate fraud,
accounting scandals, insider trading, excessive compensation, and other organizational frauds (Larcker & Tayan, 2011). All these stories have the same underlying assumption, which is that Corporate Governance is to blame, that is, the system of checks and balances meant to prevent abuse by executives failed (Larcker & Tayan, 2011).
The need for Corporate Governance originates from the idea that when separation is in place between the ownership of a company and its management, self-interested executives have the opportunity to take actions that increase their own wealth while adversely affecting shareholders‟ wealth. This is commonly known as the agency problem where the costs resulting from this problem are known as agency costs. Organizations can adapt certain types of control or monitoring systems in order to decrease their agency costs. Those control or monitoring systems are called corporate governance (Larcker & Tayan, 2011). Larcker and Tayan (2011) define corporate governance as:
“The collection of control mechanism that an organization adopts to prevent or dissuade potentially self-interested managers from engaging in activities detrimental to the welfare of shareholders and stakeholders.” (p. 8)
It must be noted however, that although several social sciences, such as behavioral psychology, have proved that individuals are self-interested, this does not mean that individuals are always uniformly and completely self-interested (Larcker & Tayan, 2011). Individuals could encounter forms of self-restraints based on moral grounds that are not related to economic rewards. This knowledge that certain actions are inherently wrong even if they stay undetected and therefore unpunished is known as moral salience (Larcker & Tayan, 2011). All individuals perceive moral salience in different degrees. The degree of moral salience depends on personality, religion, and personal and financial circumstances.
The need for corporate governance control mechanisms to discourage self-interested behavior of executives therefore depends on the magnitude of the potential agency costs, the ability of the corporate governance control mechanism to mitigate these potential agency costs, and the costs of implementing this corporate governance control mechanism (Larcker & Tayan, 2011).
The corporate governance system should consist, at least, of a board of directors who oversee the management and an external auditor which provides a professional opinion about the reliability of the financial statements (Larcker and Tayan, 2011). Most corporate
governance systems are affected by a larger group of stakeholders, including shareholders, creditors, unions, suppliers, regulators and many more. This paper solely focuses on the corporate governance as imposed by so called corporate governance codes designed by the government.
2.2 The Corporate Governance Code(s)
Until today, there are no universally accepted corporate governance standards that determine good governance. Instead, every country has developed their own standards regarding good corporate governance and incorporated this in so called corporate governance codes. Seidl et al. (2009) define a Corporate Governance Code in general as a non-binding set of principles, standards and best practices, issued by a collective body and relating to the internal
governance of corporations.
Although no single corporate governance code is identical, they could be roughly categorized as either rules-based or principle-based standards. Rules-based standards contain detailed rules for how accounting standards should be practiced. Countries such as the United States and Japan adopted the rules-based standards. The idea behind rule-based standards is that it cannot lead to situations in which professional judgments, made in good faith, result in different interpretations for similar transactions and events thereby creating comparability
5
problems (Larcker & Tayan, 2011). In addition, the risk of law suits based on wrong
accounting is high and gives accountants a strong incentive to ask for even more rules-based standards (Benston, 2006).
However, a man learns from experience that rules-based standards often provide a way to avoidance of the objectives inherent in the standards (SEC, 2003). This can result in
financial reporting that is not representational faithful to the underlying economic substance of transactions and events. In a rules-based system, financial reporting may be seen as an act of compliance rather than an act of communication (SEC, 2003). Some opponents say that there is no such thing as a “one size fits all” system. They prefer principles-based standards which offers some flexibility in the application of the standards set in the code (Seidl et al., 2009). The flexibility is meant to prevent companies from being forced into inflexible regulations. It is not intended that companies should follow all provisions. Rather, where individual rules do not fit the unique organizational settings, companies are expected to depart from the standards in order to apply what fits best their particular organizational setting (Seidl et al., 2009). This idea is incorporated in the so called “comply or explain” principle whereby the explanation should justify the unique circumstances of a company (Arcot & Bruno, 2006), whilst avoiding an inflexible “one size fits all” approach (Seidl et al, 2009). See section 2.3.1 for an detailed explanation of the “comply or explain” principle. Countries such as the United Kingdom and many other European nations adopted principles-based standards (Larcker & Tayan, 2011). The first serious Corporate Governance Code originates from the Cadbury Committee report in 1992. It contained a set of standards addressed to the boards of directors of listed companies in the U.K. from which many standards are still in use today. The
Cadbury Code of 1992 was also the first code which introduced the “comply or explain” principle (Seidl et al, 2009). For that reason, the UK is seen as a pioneer in the field of corporate governance.
This paper focuses on the UK Corporate Governance Code since it has the longest history of principles-based standards and therefore it is assumed that they have one of the best developed principles-based Corporate Governance Code that is currently operative.
2.3 The U.K. Corporate Governance Code
The U.K. corporate governance model is mostly referred to as the Anglo-Saxon model. Historically, the government of the U.K. has followed a hands-off approach to regulation. That is why this model relies on market mechanisms to determine governance standards instead of strict regulations. This resulted in a flexible position towards the corporate body in
the development of corporate governance standards and gave the U.K. a leading role in governance reforming (Larcker & Tayan, 2011).
The U.K. has known, throughout the years, several Corporate Governance Codes starting with the Cadbury Code from 1992 until 1998, the Combined Code from 1998 until 2010, to the Revised Corporate Governance Code starting from 2010 up to now (last revised in 2014)(Larcker & Tayan, 2011). However, in this paper, the attention is focused on the UK Corporate Governance Code of 2010 since that code emphasizes more attention on the board and audit committee characteristics. Although the Corporate Governance Code of 2010
covers, for a great extent, the content of the Cadbury Code of 1992. It is noticed that, since the accounting scandals, within the corporate governance code the role of the audit committee has rapidly increased in importance and expanded in scope (Audit Committee Guidance, 2011).
2.3.1 “Comply or Explain” Principle
The “comply or explain” principle requires that companies should state in their report and accounts whether they comply with the Corporate Governance Code and provide reasons for any areas of non-compliance. It is recognized that an alternative to following a provision may be justified in particular circumstances if good governance can be achieved by other means (UK Corporate Governance Code, 2010). Adding this principle allows companies to be in conformance with the code as a whole while it is still possible to deviate from individual rules. But, a condition for doing so is that the reasons for it should be explained clearly and carefully to shareholders. The explanation should illustrate how its actual practices are consistent with the principle to which the particular provision relates, contribute to good governance and promote delivery of business objectives. It should also set out the
background, provide a clear rationale for the action it is taking, and describe any mitigating actions taken to address any additional risk and maintain conformity with the relevant principle (UK Corporate Governance Code, 2010).
According to the Audit Committee Institute (2008), the strength of the “comply or explain” principle is its flexibility. However, they raised some questions as to whether the “comply or explain” principle is sustainable. They foresee a rising demand of institutional shareholders to comply with every code provision since the non-compliance explanations are currently no more than statements of the blindingly obvious. This can eventually lead to compliance for compliance‟s sake which will not be good for the board of directors, business operations and shareholders. Seidl et al (2009) examine the way the “comply or explain” principle gets applied in the UK and Germany. They find that a significant number of
7
companies are not providing full and proper justifications for deviating from code provisions, thereby confirming the Audit Committee Institutes observations.
Faure-Grimaud et al (2010) ask whether the U.K. combined code of corporate governance led to too much compliance and too few explanations. They found that the code encouraged compliance, especially in the areas not covered by its forerunner, the Cadbury Code (1992). These provisions are mainly provisions regarding the board and committee characteristics such as independent non-executive directors, senior non-executive director and audit committee. Since the accounting scandals the role of the audit committee within the corporate governance code has rapidly increased in importance and expanded in scope (Audit Committee Guidance, 2011).
Results contradicting the Audit Committee Institute‟s concerns are found by Bhuiyan et al. (2013). They investigate the effect of better compliance with corporate governance regulation on discretionary accruals and their results show that better compliance with
corporate governance regulation reduces the discretionary accruals. This is consistent with the proposition that firms complying with corporate governance regulation have more efficient monitoring compared to low compliance firms (Bhuiyan et al., 2013).
2.4 Board of Directors
The Organization for Economic Cooperation and Development (2004), which originates from article 1 of the Convention signed in 1960, provides a vision of the responsibilities of the board of directors. They state that:
“The corporate governance framework should ensure the strategic guidance of the company, the effective monitoring of management by the board, and the board’s accountability to the company and the shareholders.” (p. 24)
All directors must act in what they consider to be the best interest of the company, consistent with their statutory duties (UK Corporate Governance Code, 2010). In order to do this, the board of directors is expected to fulfill two functions: an advisory and an oversight function (Larcker & Tayan, 2011). Although the board‟s responsibilities are inter-related, they have a fundamentally different focus. In the advisory function, the board of directors consults with management regarding the strategic and operational direction of the company, hereby trying to balance the risks and rewards of the operations. In the oversight function, the board of directors is obliged with monitoring management and making sure that it is acting in the
interest of its shareholders. The duty of directors is different than those of management since they are expected to advise on strategy but do not develop it, just as they are expected to ensure the integrity of the financial statements while they do not prepare the financial
statements themselves (Larcker & Tayan, 2011). In other words, the board of directors is not an addition to the management. It should be seen as a governing body which represents the interests of shareholders.
In order to properly fulfill the advisory and oversight function, the board of directors needs to possess some independence. Larcker and Tayan (2011) describe independence as:
“The degree to which a director is free from conflicts of interest that might
Compromise his or her ability to act solely in the interest of the firm.” (p. 69)
Independence is therefore seen as critical since it, if necessary, enables directors to oppose the management. However, independence may be easily compromised. Factors such as
background, personal experiences, values and relation to management are all negatively influencing the independency of a director (Larcker & Tayan, 2011). In the absence of direct laws which address the independence of directors, the UK Corporate Governance Code (2010) has introduced several provisions regarding the board of directors and, subsequently, its committees. These provisions are perceived as best practice and, once in place, are expected to secure the board‟s independency.
Xie et al. (2001) examined the relation between earnings management and the
structure, background, and composition of a firm‟s board of directors. Their results show that a lower level of earnings management is associated with higher non-executive representation on the board. This goes together with the fact that active boards, as proxied by the number of board meetings, are also associated with lower levels of earnings management.
Vafeas and Theodorou (1998) found contradicting results. They examined the relationship between board structure and firm performance by analyzing the importance of several board characteristics such as percentage of non-executive board members,
independent chairman of the board, monitoring value of board committees. Their results reveal an insignificant relationship between the percentage of non-executive board members, leadership structure and committee composition to firm performance. These results are in line with the results from Klein (2006) who did not find a significant relationship between board independence and earnings management either.
9
Not all corporate matters are addressed by the full board of directors, instead some are delegated to committees. Directors are assigned to committees based on the experience and knowledge (Larcker & Tayan, 2011).
2.5 Audit Committee
The audit committee is responsible for monitoring (oversight function) the company‟s external audit and is the primary contact between the auditor and the corporation. This reporting responsibility is in place to prevent management manipulation of the audit (Larcker & Tayan, 2011).
As stated earlier, the interest and focus on the audit committee is not new. Since the 1940‟s, regulatory bodies and interest groups have actively promoted the idea that the establishment of audit committees would improve financial reporting, or strengthen the corporate governance (Walker, 2004). However, over the years much more attention appears to have been devoted to advocating the establishment of audit committees rather than to analyze what they should do (Walker, 2004). Walker (2004) gives several points of criticism
regarding the audit committee and their role as it was in the early 21st century.
Firstly, Walker (2004) states that the label „audit committee‟ is misleading since these committees do not do any auditing. Audit committees may spend time talking to auditors and may even be formally assigned the responsibility of selecting and engaging the external auditor (SOX, 2002). But audit committees do not audit in their own right or are they responsible for the audit opinions on financial statements.
Secondly, audit committees are unlikely to contribute to improvement in the quality of the external audit function. At best, audit committees can focus external auditors‟ attention on what members believe are issues of concern and play a role in securing value from the audit process. As stated by Flint (1980) only auditors can improve the quality of auditing. But arguably auditing cannot be entirely improved without changing the accounting framework that would ensure that information presented in financial statements is capable of independent corroboration (Wolnizer, 1995).
Thirdly, the performance of audit committees heavily depends on the people involved, their knowledge, skills, critical capacities, skepticism and determination (Walker, 2004).
In recent years there have been significant changes in expectations about what audit committees can contribute. Initially they were seen as subcommittees of boards that were to meet with and deal with matters raised by the auditors. But now such committees will also be
responsible for over sighting many aspects of the management of a corporation which can lead to an improved corporate governance (Sox, 2002; EC Directive, 2006/43).
Prior studies regarding the relationship between the audit committee and corporate governance have focused on several characteristics of audit committee. Krishnan (2005) examined the association between audit committee quality and the quality of corporate internal control. The results indicate that independent audit committees and audit committees with financial expertise are significantly less likely to be associated with the incidence of internal control problems. This is consistent with recent policy emphasis on audit committee independence and expertise (Krishnan, 2005).
Abbott et al. (2004) addresses the impact of certain audit committee characteristics on the likelihood of financial restatement. They find that the independence and activity level of the audit committee exhibit a significant and negative association with the occurrence of restatements. They also document a significant negative association between an audit
committee that includes at least one member with financial expertise and restatement (Abbott et al., 2004).
Carcello and Neal (2000) examines the relation between the composition of financially distressed firms‟ audit committees and the likelihood of receiving going-concern reports. For firms experiencing financial distress during 1994, they find that the greater the percentage of affiliated directors on the audit committee, the lower the probability the auditor issues a going-concern report. These results support regulators‟ concern about financial-reporting quality and the recent calls for more independent audit committees (Carcello & Neal, 2000).
Most prior research supports the recent policy emphasis on audit committee and their influence on audit quality but Turley and Zaman (2004) argues that there is only limited and mixed evidence of effects to support claims and perceptions about the value of audit
committees for corporate governance. He concludes that there is no automatic relationship between the adoption of audit committee structures or characteristics and the achievement of particular governance effects (Turley & Zaman, 2004).
The results from Klein‟s (2006) research are even contradicting the regulators‟
concerns. She examines the relationship between audit committee characteristics and earnings management. Her results show a non-linear negative relation between audit committee
independence and earnings manipulation which is only significant when the audit committee has less than a majority of independent directors. In contrast to the new regulations, she finds no significant association between earnings management and the more stringent requirement of 100% audit committee independence.
11 2.6 The nomination and remuneration committee
The nomination committee is responsible for the identification, evaluation and eventually nomination of new directors when seats on the board become available. The CEO succession- planning is also covered by the nomination committee (Larcker & Tayan, 2011). Vafeas (1999) examined the nature of board nominating committees and their role in corporate governance. His results show that the nominating committee can influence the independence of outside directors and are thereby partly consistent with the statement that with the use of nomination committees board quality can be improved. Brown and Caylor (2004) created a broad measure of corporate governance which they relate to firm performance. Their results show that an independent nomination committee is associated with good firm performance. The remuneration committee sets the compensation plan for the CEO and advises the CEO on the compensation plans of other executive directors (Larcker & Tayan, 2011). Gregory-Smith (2012) tests the impact of the remuneration committee independence on CEO pay. He examines whether remuneration committees facilitate optimal contracting or whether CEOs dominate the pay-setting process thereby increasing their own remuneration. No relationship is found between CEO pay and independence of remuneration committee. Weir and Laing (2001) examine the relationship between independence of the remuneration committee and company performance. Their results show little relationship between the proportion of non-executive directors serving on the remuneration committee and company performance.
2.7 Hypothesis
Corporate governance issues mainly arise in organizations whose ownership and control are separated. Shareholders who invest in such organizations are not able to control whether the money is used in their interest (agency problem). This is the main problem that corporate governance should address from an agency perspective point of view (Goncharov et al., 2006). This agency problem results from the fact that managers have incentives to increase their own utility, by ways of consumption of capital or empire building, but not necessarily by maximizing the return on capital invested by the shareholders (Goncharov et al., 2006). A possible solution for the agency problem would be the use of optimal contracting. However, prior research has shown that optimal contracting is practically impossible since the
transaction cost would be too high (Hart, 1995). This means that there is a need for control mechanisms that an organization can adopt to prevent or dissuade potentially self-interested managers from engaging in activities which are not in the best interest of the shareholders and
stakeholders (corporate governance systems)(Larcker & Tayan, 2011). The U.K. Corporate Governance Code (2010) puts several corporate governance mechanisms in place and claims to consist of principles that are “good”, “responsible” and represent a form of “best practice”. If this is true, then it should be observable that companies which comply to a higher degree with the U.K. Corporate Governance Code have a higher corporate governance quality (and vice versa). In other words, a high degree of compliance should be positively related to the corporate governance quality of a company. However, this is not necessarily true. It is theoretically possible that non-complying companies could achieve a higher corporate governance quality than complying companies, due to the fact that they designed a very unique and effective internal corporate governance system, one which is not covered by the UK Corporate Governance Code, while complying companies are just following the crowd, complying for compliance‟s sake, and do not significantly improve their internal corporate governance system.
I expect to find a relationship between the rate of compliance with the UK Corporate Governance Code and the corporate governance quality, but, that relationship can be both positive or negative. Therefore I do not predict the coefficient of the relationship between the rate of compliance and corporate governance quality. The non-directional hypothesis
examined in this paper is:
Compliance with the U.K. Corporate Governance Code is associated with corporate governance quality
3. Methodology
A quantitative archival database research is used to test the hypothesis. This method is broadly used in this research area and is therefore assumed to be an appropriate method for this paper. Since this method is broadly used in similar studies it also enhances the
comparability of this paper with other papers.
3.1 Sample
The sample in this research is based on archival data and is partially hand-collected. The selection process is performed in such a way that the observations of the sample are not compromised by external factors. In order to achieve some form of comparability within the sample, it includes all the companies from the UK FTSE350 since these companies encounter the same economic and political environment and are all obliged to report under the same
13
corporate governance code (The UK Corporate Governance Code 2010). Another reason to use a sample based on UK companies instead of US companies is the fact that the studied condition is not observable in the US due to different legislations. This paper only includes observations from 2013 in order to examine the recent emphasis on board and audit
committee characteristics which are incorporated in the UK Corporate Governance Code of 2010. However, to calculate the discretionary accruals financial data of 2012 is also required. The financial dataset is obtained from the COMPUSTAT global database. Financial
institutions are excluded from the dataset since the overall regulatory environment for those companies differs from that of non-financial companies. It is perceived that those regulations, although not part of the corporate governance code, may influence its provisions and bring implications for corporate governance and performance evaluation (Arcot et al., 2010). The financial dataset, after exclusion of the financial institutions, consisted of 228 companies.
The next step was to hand-collect the variables regarding the compliance rate with the corporate governance code. The hand-collection process consisted of manually checking the annual reports of 2013, for the final 228 companies, to examine whether they were in
compliance with certain provisions or not. The compliance rate was checked for a total of 11 provisions. Appendix 1 provides a list of these provisions and their corresponding
descriptions. All provisions were separately examined in the annual report and coded as either “in compliance” (one) or “not in compliance” (zero). Appendix 2 provides a list of these provisions and their corresponding criteria used for coding. In the situation where the annual report did not mention a certain provision, this provision was coded as “not in compliance” (zero). Also, the company‟s own in compliance statement was neglected and therefore did not automatically lead to an “in compliance” coding for all provisions.
After the hand-collection process, the 2 datasets were combined and observations with missing variables were dropped. This decreased to sample size from 228 companies to 181
companies. Winsorisation on the 1st and 99th percentiles was used to control for any outliers
and 1 more observation was dropped because of its major outlier in the residuals. This reduced the sample size to 180 companies which in the end became my final sample size.
3.2 Research Design
This paper examines the relation between compliance with the UK Corporate Governance Code and corporate governance quality. Compliance with the U.K. Corporate Governance Code is tested with the use of a dummy variable which is 1 if a company is in full-compliance and 0 if a company is not in full-compliance. Companies are coded as full-compliance
companies if they comply with all the 11 provisions as stated in appendix 2. Non-compliance with any of these 11 provisions results in a coding as not full-compliance company. Corporate governance quality is measured by earnings management, which is explained in section 3.3 and 3.4. Since other factors could potentially influence the relationship between compliance and corporate governance quality it is necessary to include several control variables. The natural log of the total assets (SIZE) of the specific company is used to control for size differences between the companies (Klein, 2006; Bhuiyan et al., 2013). Current ratio
(CURR), Debt to Assets (LEV), Inventory to total Assets (INV), Receivables to total Assets (REC), Cash flows to total Assets (CFO) and Return on Assets (ROA) are all variables to control for complexity (Klein, 2006; Bhuiyan et al., 2013). The control variable EEP is added to control for the effect of extreme earnings performance (Klein, 2006). The last control variable added is a dummy variable (LOSSD), which is 1 if the net income is negative and 0 if the net income is not negative, created to control for the effect of a negative net income. I performed a skewness and kurtosis test to check if the control variables were normally distributed (appendix 3). The results showed that I needed to change some of the control variables in order to improve their normal distribution. However, only the following control variables were eligible for improvement. SIZE became 1/sqrt(SIZE), CURR became
log(CURR) and REC became sqrt(REC).
Model 1 examines the impact of full-compliance on corporate governance quality.
𝑫𝑨𝑪 = 𝜶 + 𝜷𝟏 𝑭𝑪 + 𝜷𝟐 𝑹𝑶𝑨 + 𝜷𝟑 𝑳𝑬𝑽 + 𝜷𝟒 𝑬𝑬𝑷 + 𝜷𝟓 𝑰𝑵𝑽 + 𝜷𝟔 𝑪𝑭𝑶 + 𝜷𝟕 𝑺𝑰𝒁𝑬 + 𝜷𝟖 𝑪𝑼𝑹𝑹 + 𝜷𝟗 𝑹𝑬𝑪 + 𝜷𝟏𝟑 𝑳𝑶𝑺𝑺𝑫 + 𝜺
(1) Where:
𝐷𝐴𝐶 = 𝐷𝑖𝑠𝑐𝑟𝑒𝑡𝑖𝑜𝑛𝑎𝑟𝑦 𝐴𝑐𝑐𝑟𝑢𝑎𝑙𝑠
𝐹𝐶 = 𝐹𝑢𝑙𝑙 𝐶𝑜𝑚𝑝𝑖𝑎𝑛𝑐𝑒: 𝑑𝑢𝑚𝑚𝑦 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 'one' if in full compliance, otherwise 'zero' 𝑅𝑂𝐴 = 𝑅𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝐴𝑠𝑠𝑒𝑡𝑠: 𝑁𝑒𝑡 𝐼𝑛𝑐𝑜𝑚𝑒 𝑑𝑖𝑣𝑖𝑑𝑒𝑑 𝑏𝑦 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠 𝐿𝐸𝑉 = 𝑇𝑜𝑡𝑎𝑙 𝐷𝑒𝑏𝑡 𝑑𝑖𝑣𝑖𝑑𝑒𝑑 𝑏𝑦 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠 𝐸𝐸𝑃 = 𝐸𝑥𝑡𝑟𝑒𝑚𝑒 𝑒𝑎𝑟𝑛𝑖𝑛𝑔𝑠 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒: 𝑐𝑎𝑛𝑔𝑒 𝑖𝑛 𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠 𝑐𝑜𝑚𝑝𝑎𝑟𝑒𝑑 𝑡𝑜 𝑝𝑟𝑒𝑣𝑖𝑜𝑢𝑠 𝑦𝑒𝑎𝑟 𝑑𝑖𝑣𝑖𝑑𝑒𝑑 𝑏𝑦 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠 𝐼𝑁𝑉 = 𝐼𝑛𝑣𝑒𝑛𝑡𝑜𝑟𝑦 𝑑𝑖𝑣𝑖𝑑𝑒𝑑 𝑏𝑦 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠 𝐶𝐹𝑂 = 𝑁𝑒𝑡 𝐶𝑎𝑠 𝐹𝑙𝑜𝑤 𝑓𝑟𝑜𝑚 𝑂𝑝𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑠 𝑑𝑖𝑣𝑖𝑑𝑒𝑑 𝑏𝑦 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠 𝑆𝐼𝑍𝐸 = 𝑁𝑎𝑡𝑢𝑟𝑎𝑙 𝑙𝑜𝑔𝑎𝑟𝑖𝑡𝑚 𝑜𝑓 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠 𝐶𝑈𝑅𝑅 = 𝐶𝑢𝑟𝑟𝑒𝑛𝑡 𝑅𝑎𝑡𝑖𝑜: 𝐶𝑢𝑟𝑟𝑒𝑛𝑡 𝐴𝑠𝑠𝑒𝑡𝑠 𝑑𝑖𝑣𝑖𝑑𝑒𝑑 𝑏𝑦 𝐶𝑢𝑟𝑟𝑒𝑛𝑡 𝐿𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑖𝑒𝑠 𝑅𝐸𝐶 = 𝑅𝑒𝑐𝑒𝑖𝑣𝑎𝑏𝑙𝑒𝑠 𝑑𝑖𝑣𝑖𝑑𝑒𝑑 𝑏𝑦 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠
15
Model 2 is created to perform a more in depth analysis of the impact that compliance has on corporate governance quality. Instead of using 1 dummy variable indicating full-compliance or not, multiple dummy variables are introduced to address the different provisions of the U.K. Corporate Governance Code to which companies can comply. This model examines the impact of these separate provisions on corporate governance quality. The provisions that are included in this model are: separation of CEO and Chairman (sCEOCoB), independent Chairman (IndCoB), senior Non-Executive Director (sNED), role Chairman (rCoB), Non-Executive Directors (NEDs), Nomination Committee (Nom), Performance Evaluation (Eva), Audit Committee (Audit), Remuneration Committee (Rem), Service Contracts (SeCo) and Performance Statement (Pstat). For a detailed description of these provisions together with their corresponding requirement criteria see Appendix 2.
𝑫𝑨𝑪 = 𝜶 + 𝜷𝟏 𝒔𝑪𝑬𝑶𝑪𝒐𝑩 + 𝜷𝟐 𝑰𝒏𝒅𝑪𝒐𝑩 + 𝜷𝟑 𝒔𝑵𝑬𝑫 + 𝜷𝟒 𝒓𝑪𝒐𝑩 + 𝜷𝟓 𝑵𝑬𝑫𝒔 + 𝜷𝟔 𝑵𝒐𝒎 + 𝜷𝟕 𝑬𝒗𝒂 + 𝜷𝟖 𝑨𝒖𝒅𝒊𝒕 + 𝜷𝟗 𝑹𝒆𝒎 + 𝜷𝟏𝟎 𝑺𝒆𝑪𝒐 + 𝜷𝟏𝟏 𝑷𝒔𝒕𝒂𝒕 + 𝜷𝟏𝟐 𝑹𝑶𝑨 + 𝜷𝟏𝟑 𝑳𝑬𝑽 + 𝜷𝟏𝟒 𝑬𝑬𝑷 + 𝜷𝟏𝟓 𝑰𝑵𝑽 + 𝜷𝟏𝟔 𝑪𝑭𝑶 + 𝜷𝟏𝟕 𝑺𝑰𝒁𝑬 + 𝜷𝟏𝟖 𝑪𝑼𝑹𝑹 + 𝜷𝟏𝟗 𝑹𝑬𝑪 + 𝜷𝟐𝟎 𝑳𝑶𝑺𝑺𝑫 + 𝜺 (2) Where: 𝐷𝐴𝐶 = 𝐷𝑖𝑠𝑐𝑟𝑒𝑡𝑖𝑜𝑛𝑎𝑟𝑦 𝐴𝑐𝑐𝑟𝑢𝑎𝑙𝑠
𝑠𝐶𝐸𝑂𝐶𝑜𝐵 = 𝑠𝑒𝑝𝑎𝑟𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝐶𝐸𝑂 𝑎𝑛𝑑 𝐶𝑎𝑖𝑟𝑚𝑎𝑛: 𝑑𝑢𝑚𝑚𝑦 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 'one' if separated, otherwise 'zero' 𝐼𝑛𝑑𝐶𝑜𝐵 = 𝐼𝑛𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒 𝑜𝑓 𝐶𝑎𝑖𝑟𝑚𝑎𝑛: 𝑑𝑢𝑚𝑚𝑦 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 'one' if independent, otherwise 'zero'
𝑠𝑁𝐸𝐷 = 𝐴𝑝𝑝𝑜𝑖𝑛𝑡𝑚𝑒𝑛𝑡 𝑜𝑓 𝑆𝑒𝑛𝑖𝑜𝑟 𝑁𝑜𝑛 − 𝐸𝑥𝑒𝑐𝑢𝑡𝑖𝑣𝑒 𝐷𝑖𝑟𝑒𝑐𝑡𝑜𝑟: 𝑑𝑢𝑚𝑚𝑦 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 'one' if compliance, otherwise 'zero' 𝑟𝐶𝑜𝐵 = 𝑅𝑜𝑙𝑒 𝑜𝑓 𝑡𝑒 𝐶𝑎𝑖𝑟𝑚𝑎𝑛: 𝑑𝑢𝑚𝑚𝑦 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 'one' if compliance, otherwise 'zero'
𝑁𝐸𝐷𝑠 = 𝑃𝑟𝑜𝑝𝑜𝑟𝑡𝑖𝑜𝑛 𝑜𝑓 𝑁𝑜𝑛 − 𝐸𝑥𝑒𝑐𝑢𝑡𝑖𝑣𝑒 𝐷𝑖𝑟𝑒𝑐𝑡𝑜𝑟𝑠: 𝑑𝑢𝑚𝑚𝑦 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 'one' if half of the board, otherwise 'zero' 𝑁𝑜𝑚 = 𝑁𝑜𝑚𝑖𝑛𝑎𝑡𝑖𝑜𝑛 𝐶𝑜𝑚𝑚𝑖𝑡𝑡𝑒𝑒: 𝑑𝑢𝑚𝑚𝑦 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 'one' if properly in place, otherwise 'zero'
𝐸𝑣𝑎 = 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 𝐸𝑣𝑎𝑙𝑢𝑎𝑡𝑖𝑜𝑛: 𝑑𝑢𝑚𝑚𝑦 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 'one' if properly in place, otherwise 'zero' 𝐴𝑢𝑑𝑖𝑡 = 𝐴𝑢𝑑𝑖𝑡 𝐶𝑜𝑚𝑚𝑖𝑡𝑡𝑒𝑒: 𝑑𝑢𝑚𝑚𝑦 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 'one' if properly in place, otherwise 'zero' 𝑅𝑒𝑚. = 𝑅𝑒𝑚𝑢𝑛𝑒𝑟𝑎𝑡𝑖𝑜𝑛 𝐶𝑜𝑚𝑚𝑖𝑡𝑡𝑒𝑒: 𝑑𝑢𝑚𝑚𝑦 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 'one' if properly in place, otherwise 'zero' 𝑆𝑒𝐶𝑜 = 𝑆𝑒𝑟𝑣𝑖𝑐𝑒 𝐶𝑜𝑛𝑡𝑟𝑎𝑐𝑡𝑠: 𝑑𝑢𝑚𝑚𝑦 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 'one' if properly in place, otherwise 'zero' 𝑃𝑠𝑡𝑎𝑡 = 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 𝑆𝑡𝑎𝑡𝑒𝑚𝑒𝑛𝑡: 𝑑𝑢𝑚𝑚𝑦 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 'one' if properly in place, otherwise 'zero' 𝑅𝑂𝐴 = 𝑅𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝐴𝑠𝑠𝑒𝑡𝑠: 𝑁𝑒𝑡 𝐼𝑛𝑐𝑜𝑚𝑒 𝑑𝑖𝑣𝑖𝑑𝑒𝑑 𝑏𝑦 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠 𝐿𝐸𝑉 = 𝑇𝑜𝑡𝑎𝑙 𝐷𝑒𝑏𝑡 𝑑𝑖𝑣𝑖𝑑𝑒𝑑 𝑏𝑦 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠 𝐸𝐸𝑃 = 𝐸𝑥𝑡𝑟𝑒𝑚𝑒 𝑒𝑎𝑟𝑛𝑖𝑛𝑔𝑠 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒: 𝑐𝑎𝑛𝑔𝑒 𝑖𝑛 𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠 𝑐𝑜𝑚𝑝𝑎𝑟𝑒𝑑 𝑡𝑜 𝑝𝑟𝑒𝑣𝑖𝑜𝑢𝑠 𝑦𝑒𝑎𝑟 𝑑𝑖𝑣𝑖𝑑𝑒𝑑 𝑏𝑦 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠 𝐼𝑁𝑉 = 𝐼𝑛𝑣𝑒𝑛𝑡𝑜𝑟𝑦 𝑑𝑖𝑣𝑖𝑑𝑒𝑑 𝑏𝑦 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠 𝐶𝐹𝑂 = 𝑁𝑒𝑡 𝐶𝑎𝑠 𝐹𝑙𝑜𝑤 𝑓𝑟𝑜𝑚 𝑂𝑝𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑠 𝑑𝑖𝑣𝑖𝑑𝑒𝑑 𝑏𝑦 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠 𝑆𝐼𝑍𝐸 = 𝑁𝑎𝑡𝑢𝑟𝑎𝑙 𝑙𝑜𝑔𝑎𝑟𝑖𝑡𝑚 𝑜𝑓 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠 𝐶𝑈𝑅𝑅 = 𝐶𝑢𝑟𝑟𝑒𝑛𝑡 𝑅𝑎𝑡𝑖𝑜: 𝐶𝑢𝑟𝑟𝑒𝑛𝑡 𝐴𝑠𝑠𝑒𝑡𝑠 𝑑𝑖𝑣𝑖𝑑𝑒𝑑 𝑏𝑦 𝐶𝑢𝑟𝑟𝑒𝑛𝑡 𝐿𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑖𝑒𝑠 𝑅𝐸𝐶 = 𝑅𝑒𝑐𝑒𝑖𝑣𝑎𝑏𝑙𝑒𝑠 𝑑𝑖𝑣𝑖𝑑𝑒𝑑 𝑏𝑦 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠
3.3 Corporate Governance Quality
Prior research has often addressed the effectiveness or consequences of the corporate governance codes in general or particular aspects. It shows that there is no single right measurement to determine the effectiveness or consequence of the good corporate
governance. Corporate governance quality is therefore an abstract concept with no common variable that measures corporate governance quality. Prior studies have used different proxies for corporate governance effectiveness.
Some studies, such as Abbott et al. (2004) use financial restatement as their
measurement since financial restatement have exhibited a emerging increase in magnitude, as well as occurrence and it allows for insights into the audit committee‟s ability to influence both internal and external audit effectiveness. Financial restatements may imply an ineffective internal control system and/or external auditor. However, no control system or external
auditor is capable of preventing or detecting 100 percent of the financial misstatements. As a consequence, if material financial misstatements are discovered and communicated to audit committees in subsequent annual audits, an effective audit committee could be more likely to demand a financial restatement (Abbott et al., 2004). In this context financial restatement are not necessarily a sign for bad corporate governance.
Other studies, such as Carcello & Neal (2000) use going-concern opinions issued as a measurement for effective corporate governance. They expect that an independent audit committee is more likely to mitigate any management pressure on auditors to issue a clean opinion when a going-concern opinion is warranted. In other words, the greater the
independency of the audit committee the more likely it is that the external auditor issues a going-concern opinion. The limitation for using going-concern opinions as a measurement is that, up and till now, only association instead of causation is documented between the audit committee and going-concern opinions. The question remains if audit committees actually affect the issue of a going-concern opinion.
Another measurement often used is discretionary accruals. Discretionary accruals is a form of earnings management and is defined as the practice of distorting the true financial performance of the company (SEC). Klein (2006) uses this measurement to examine whether audit committee and board characteristics are related to earnings management. The motivation behind this measurement is the implicit allegation by the SEC, the NYSE and the NASDAQ that discretionary accruals and poor corporate governance mechanisms are positively related. The limitations of using discretionary accruals as a measurement lays in the fact that earnings
17
management is an unobservable process which needs proxies to make any inferences. This decreases the reliability of the outcomes.
The proven causality between discretionary accruals and corporate governance
together with the encountered difficulty in collecting data regarding the financial restatements and going-concern opinions for U.K. firms led to the selection of discretionary accruals as the dependent variable in this paper.
3.4 Earnings Management
This paper uses earnings management as a proxy for corporate governance quality. The Modified Jones model, as used by Dechow et al. (1995), is used in order to measure the magnitude of earnings management. This model is based on accrual based accounting. Accrual based accounting arises from the notion that there is a difference between cost and expenditures versus benefits and revenue (van Praag, 2001). Because of this difference, net income can be perceived as the adjustment of the operational cash flow for provisionally components resulting in the net income from operations. These adjustments are called accruals and there are a number of subjective decisions involved in the composition of accruals. In the earnings management literature it is assumed that these accruals are open to more discretion than cash flows and that discretionary accruals are often used to manage earnings (van Praag, 2001). However, it is very difficult to separate total accruals into discretionary accruals and non-discretionary accruals (Dechow et al., 1995).
Dechow et al. (1995) tested different accruals models to determine which model was the most powerful model to detect earnings management. They found that the most powerful model in detecting earnings management was the Modified Jones model since this model includes the changes in accounts receivable while determining the change in revenue. The Modified Jones model assumes that all changes in credit sales are the result of earnings
management. This assumption originates from the argumentation that earnings management is easier to realize with revenue recognition from credit sales than with revenue recognition from cash sales (Dechow et al., 1995).
Before the Modified Jones model can be used, it is necessary to determine the total amount of accruals. The total amount of accruals can be calculated by diminishing the Income Before Extraordinary Items year t with the Cash Flows from Operation year t.
𝑇𝐴𝑡 = 𝐸𝑋𝐵𝐼𝑡− 𝐶𝐹𝑂𝑡 (3) Where: 𝑇𝐴𝑡 = 𝑇𝑜𝑡𝑎𝑙 𝐴𝑐𝑐𝑟𝑢𝑎𝑙𝑠 𝑖𝑛 𝑦𝑒𝑎𝑟 𝑡 𝐸𝑋𝐵𝐼𝑡 = 𝐼𝑛𝑐𝑜𝑚𝑒 𝐵𝑒𝑓𝑜𝑟𝑒 𝐸𝑥𝑡𝑟𝑎𝑜𝑟𝑑𝑖𝑛𝑎𝑟𝑦 𝐼𝑡𝑒𝑚𝑠 (𝑖𝑛 𝑦𝑒𝑎𝑟 𝑡) 𝐶𝐹𝑂𝑡 = 𝐶𝑎𝑠 𝐹𝑙𝑜𝑤𝑠 𝑓𝑟𝑜𝑚 𝑂𝑝𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑠 (𝑖𝑛 𝑦𝑒𝑎𝑟 𝑡)
The following equations (Equations 4, 5 & 6) are used to estimate the Discretionary Accruals. At first the parameters of equation 4 are estimated.
𝑇𝐴𝑡 = 𝛽1,𝑡 1 𝐴𝑡−1 + 𝛽2,𝑡 ∆𝑅𝐸𝑉𝑡 𝐴𝑡−1 + 𝛽3,𝑡 ∆𝑃𝑃𝐸𝑡 𝐴𝑡−1 + 𝜀 (4) Where: 𝐴𝑡−1= 𝐴𝑠𝑠𝑒𝑡𝑠 𝑖𝑛 𝑦𝑒𝑎𝑟 𝑡 − 1 ∆𝑅𝐸𝑉𝑡 = 𝐶𝑎𝑛𝑔𝑒 𝑖𝑛 𝑅𝑒𝑣𝑒𝑛𝑢𝑒 (𝑦𝑒𝑎𝑟 𝑡 − 1 𝑡𝑜 𝑦𝑒𝑎𝑟 𝑡) ∆𝑃𝑃𝐸𝑡 = 𝐶𝑎𝑛𝑔𝑒 𝑔𝑟𝑜𝑠𝑠 𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑦, 𝑝𝑙𝑎𝑛𝑡 𝑎𝑛𝑑 𝑒𝑞𝑢𝑖𝑝𝑚𝑒𝑛𝑡 (𝑦𝑒𝑎𝑟 𝑡 − 1 𝑡𝑜 𝑦𝑒𝑎𝑟 𝑡) 𝛽𝑡 = 𝑃𝑎𝑟𝑎𝑚𝑒𝑡𝑒𝑟𝑠 𝜀 = 𝐸𝑟𝑟𝑜𝑟 𝑜𝑓 𝐸𝑠𝑡𝑖𝑚𝑎𝑡𝑒
Then the outcomes from equation 4 are used to find any Non-Discretionary Accruals, see equation 5. 𝑁𝐴𝑡 = 𝛽1,𝑡 1 𝐴𝑡−1 + 𝛽2,𝑡 ∆𝑅𝐸𝑉𝑡− ∆𝑅𝐸𝐶𝑡 𝐴𝑡−1 + 𝛽3,𝑡 ∆𝑃𝑃𝐸𝑡 𝐴𝑡−1 + 𝜀 (5) Where: ∆𝑅𝐸𝐶𝑡 = 𝐶𝑎𝑛𝑔𝑒 𝑖𝑛 𝑅𝑒𝑐𝑒𝑖𝑣𝑎𝑏𝑙𝑒𝑠 (𝑦𝑒𝑎𝑟 𝑡 − 1 𝑡𝑜 𝑦𝑒𝑎𝑟 𝑡)
Eventually, the Discretionary Accruals are found by subtracting the Non-Discretionary Accruals from the Total Accruals (Equation 6).
𝐷𝐴𝑡 = 𝑇𝐴𝑡− 𝑁𝐴𝑡
(6) Where:
𝐷𝐴𝑡 = 𝐷𝑖𝑠𝑐𝑟𝑒𝑡𝑖𝑜𝑛𝑎𝑟𝑦 𝐴𝑐𝑐𝑟𝑢𝑎𝑙𝑠 (𝑖𝑛 𝑦𝑒𝑎𝑟 𝑡) 𝑁𝐴𝑡 = 𝑁𝑜𝑛 𝑑𝑖𝑠𝑐𝑟𝑒𝑡𝑖𝑜𝑛𝑎𝑟𝑦 𝐴𝑐𝑐𝑟𝑢𝑎𝑙𝑠 (𝑖𝑛 𝑦𝑒𝑎𝑟 𝑡)
19 4. Results
4.1 Descriptive Statistics
The descriptive statistics stated in table 1 provide the following insights. The dependent variable, discretionary accruals, have a mean of -0.0003 and a standard deviation of 0.051. It could be concluded that proportion of companies applying upward earnings management and companies applying downward earnings management is roughly the same in this sample. It must be noted that positive discretionary accruals are mainly related to “upward earnings management”, where negative discretionary accruals are partly related to “downward earnings management” and “reversals” (Epps & Guthrie, 2010). The mean of the full compliance dummy variable is 0.4778 which indicates that this sample consists of slightly more companies that are not in full compliance with the U.K. Corporate Governance Code. The means of the dummy variables regarding the compliance with separate provisions are all above 0.8167, except for independent chairman which is 0.6500, indicating that most
companies are in compliance with these separate provisions. The mean of the full compliance dummy variable is far below the means of the separate provision dummy variables is due to the fact that full compliance is achieved when the requirements for all the separate provision dummy variables are met. Since many companies lack compliance with only one or two of the separate provision dummy variables, the achievement of full compliance drops substantially faster than the achievement of the separate provisions.
Table 1: Descriptive Statistics
Variables N mean sd min max Var skewness kurtosis
Discretionary Accruals 180 -0.0003 0.051 -0.157 0.179 0.0026 0.195 4.198
Full Complianceᵃ 180 0.4778 0.501 0.000 1.000 0.2510 0.089 1.008
Separation of CEO and Chairmanᵇ 180 0.9833 0.128 0.000 1.000 0.0165 -7.551 58.02
Independent Chairmanᵇ 180 0.6500 0.478 0.000 1.000 0.2290 -0.629 1.396
Senior Non-Executive Directorᵇ 180 0.9778 0.148 0.000 1.000 0.0218 -6.482 43.02
Role Chairmanᵇ 180 0.9944 0.075 0.000 1.000 0.0056 -13.30 178.0
Proportion Non-Executive Directorsᵇ 180 0.9000 0.301 0.000 1.000 0.0905 -2.667 8.111
Nomination Committeeᵇ 180 0.9611 0.194 0.000 1.000 0.0376 -4.770 23.75 Performance Evaluationᵇ 180 0.8167 0.388 0.000 1.000 0.1510 -1.637 3.679 Audit Committeeᵇ 180 0.9389 0.240 0.000 1.000 0.0577 -3.665 14.43 Remuneration Committeeᵇ 180 0.9111 0.285 0.000 1.000 0.0814 -2.889 9.348 Service Contractsᵇ 180 0.9500 0.219 0.000 1.000 0.0478 -4.129 18.05 Performance Statementsᵇ 180 0.8778 0.328 0.000 1.000 0.1080 -2.307 6.321 Return on Assets 180 0.0657 0.064 -0.133 0.260 0.0041 0.297 4.430
Debt to Assets Ratio 180 0.1751 0.148 0.000 0.562 0.0220 0.677 2.764
Extreme Earnings Performance 180 0.0086 0.037 -0.096 0.116 0.0014 0.019 3.879
Inventory to Assets Ratio 180 0.1049 0.116 0.000 0.439 0.0134 1.447 4.546
Cash Flows to Assets Ratio 180 0.1116 0.069 -0.086 0.294 0.0048 0.148 3.624
Size 180 0.3621 0.034 0.289 0.442 0.0012 0.087 2.531
Current Ratio 180 0.2784 0.579 -1.195 1.660 0.3350 0.028 3.013
Receivables to Assets Ratio 180 0.3415 0.134 0.071 0.722 0.0179 0.234 2.756
Negative Net Incomeᶜ 180 0.1056 0.308 0.000 1.000 0.0949 2.567 7.592
Note: ᵃ dummy variable (1=Full Compliance, 0=Not in Full Compliance), ᵇ dummy variable (1=in Compliance with Provision,
The dependent variable and control variables are checked for normality. This is not done for my independent variables as they are all dummy variables, with either the value “zero” or “one”, which do not need to be normally distributed. The dependent variable and all the control variables, except Negative Net Income which is also a dummy variable, are
checked for normality with the use of a skewness and kurtosis test for normality (sktest). The sktest conducts two normality tests for each variable, one normality test based on skewness and another normality test based on kurtosis. It then combines the results of the two tests into an overall test statistic and sets the null-hypothesis that the variable is normally distributed. The p-value of this test needs to exceed the alpha level, alpha level is set at 0.10, in order to not reject the null-hypothesis. The results of this test (appendix 3) show a p-value of 0.0273 for Discretionary Accruals, 0.0083 for Return on Assets, 0.0040 for Debt to Assets and 0.0000 for Inventory to Assets which are all smaller than the pre-set alpha level of 0.10. It could therefore be concluded that the variables Discretionary Accruals, Return on Assets, Debt to Assets and Inventory to Assets are not normally distributed. The other control variables all show a p-value which exceeds 0.10 and are therefore assumed to be normally distributed. There are no assumptions made regarding the distribution of the dependent and independent variables since the OLS regression is not very sensitive to variables which are not normally distributed. In general, this data set qualifies for an OLS regression.
Table 2 provides the Pearson correlation matrix. This matrix calculates the separate correlation coefficients between the variables used in the OLS regression. It is therefore a univariate analysis. The coefficients stated in bold are significant. Issues regarding the multi-collinearity of variables could arise when the correlation coefficient between two independent variables exceeds 0.8. Multi-collinearity is a disturbance in the data set since one independent variable could be linearly predicted from the other independent variable. This could lead to unreliable inferences based on the data set.
The correlation matrix (table 2) shows that the return on assets and cash flow to assets ratio have a high correlation. However, this correlation is expected since both variables express a form of profit as a ratio of the total assets. Return on assets also experiences a high correlation with the loss dummy. This correlation is explained by the fact that the main element of these two variables is the net income. It should also be noted that not all control variables are significantly correlated to each other. Finally, the discretionary accruals do not appear to be highly correlated with any of the other variables. Apart from the significant correlations between the variables return on assets (0.3163), cash flow to assets ratio (0.3653),
21
Table 2: Pearson Correlation Matrix
n = 180 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 1. DAC -2. FC -0.0205 -3. sCEOCoB -0.0106 0.1245 * -4. IndCoB 0.0048 0.7019 *** 0.1774 ** -5. sNED -0.0169 0.1442 * -0.0196 0.1264 * -6. rCoB -0.0067 0.0715 -0.0097 0.1019 0.4958 *** -7. NEDs -0.0159 0.3188 *** 0.1013 0.1437 * 0.3266 *** 0.2242 *** -8. Nom -0.0488 0.1924 *** -0.0262 0.2139 *** 0.5545 *** 0.3716 *** 0.5077 *** -9. Eva -0.0919 0.4532 *** -0.0617 0.1340 * 0.3182 *** 0.1578 ** 0.2728 *** 0.3503 *** -10. Audit -0.0238 0.2440 *** -0.0332 0.2991 *** 0.4336 *** 0.2930 *** 0.3788 *** 0.5485 *** 0.3586 *** -11. Rem -0.0703 0.2988 *** -0.0407 0.2619 *** 0.3502 *** 0.2393 *** 0.3514 *** 0.6440 *** 0.3061 *** 0.6538 *** -12. SeCo -0.0247 0.2194 *** -0.0299 0.0989 0.1383 * 0.3258 *** 0.2634 *** 0.3494 *** 0.2866 *** 0.2607 *** 0.2866 *** -13. Pstat -0.0003 0.3569 *** 0.0839 0.1174 0.1739 ** 0.2003 *** 0.1583 ** 0.2759 *** 0.4369 *** 0.2588 *** 0.1815 ** 0.3035 *** -14. ROA 0.3163 *** -0.0161 -0.1594 ** 0.0907 0.0830 0.0489 -0.0107 0.0022 -0.0793 0.0505 0.0005 0.1097 -0.0983 -15. LEV -0.0002 0.0097 0.1296 * -0.0773 -0.0034 0.0885 0.0132 -0.0736 0.0744 -0.0183 0.0150 -0.0830 0.0236 -0.2549 *** -16. EEP -0.0732 -0.1220 0.0749 0.0735 -0.0900 -0.2156 *** -0.1182 -0.0684 -0.2175 *** -0.0583 -0.0466 -0.2492 *** -0.2162 *** 0.3010 *** -0.1809 ** -17. INV 0.1323 * 0.0332 0.0416 0.0747 0.0873 0.0679 0.0743 0.1425 * -0.0834 0.1035 0.1077 0.1665 ** -0.1064 0.1702 ** -0.3500 *** 0.0821 -18. CFO -0.3653 *** -0.0408 -0.1848 ** 0.0652 0.1566 ** 0.2138 *** -0.0003 0.0836 -0.0398 0.1165 0.0758 0.1020 -0.0745 0.6086 *** -0.1058 0.3028 *** -0.1436 * -19. SIZE -0.0611 -0.0973 -0.0555 -0.0113 -0.0243 -0.1757 ** -0.2101 *** -0.0821 -0.2546 *** -0.1610 ** -0.1472 ** -0.0673 -0.2149 *** 0.3224 *** -0.3477 *** 0.2230 *** 0.1638 ** 0.2612 *** -20. CURR 0.0686 0.0713 -0.0550 0.0635 0.1069 0.0262 0.0075 0.0830 0.0039 0.1357 * 0.1047 0.0092 -0.0643 0.0833 -0.2906 *** 0.0883 0.4071 *** -0.0284 0.1929 *** -21. REC -0.1045 0.0891 -0.0363 0.1431 * 0.0292 -0.0788 -0.1061 0.0724 0.1590 ** 0.1175 0.1644 ** 0.0760 0.0837 0.1689 ** -0.1954 *** 0.0679 -0.0514 0.0847 0.2511 *** 0.0736 -22. LOSS -0.2881 *** -0.0028 0.0447 -0.0512 -0.0709 0.0257 -0.0060 -0.0244 -0.0241 -0.0633 0.0438 -0.1701 ** -0.0926 -0.5798 *** 0.1624 ** -0.1120 -0.0882 -0.1398 * -0.1219 0.0719 -0.1746 **
22
Variable VIF 1/VIF
Return on Assets 3.22 0.31088
Nomination Committee 2.69 0.37208
Cash Flows to Assets Ratio 2.52 0.39719
Remuneration Committee 2.40 0.41637
Audit Committee 2.15 0.46529
Senior Non-Executive Director 1.95 0.51386
Negative Net Income 1.93 0.51707
Role Chairman 1.73 0.57758
Inventory to Assets Ratio 1.68 0.59403
Performance Evaluation 1.61 0.62268
Size 1.57 0.63661
Service Contracts 1.56 0.64207
Proportion Non-Executive Directors 1.53 0.65186
Performance Statements 1.45 0.69097
Extreme Earnings Performance 1.44 0.69275
Debt to Assets Ratio 1.43 0.69921
Current Ratio 1.34 0.74764
Receivables to Assets Ratio 1.30 0.76840
Separation of CEO and Chairman 1.20 0.83306
Independent Chairman 1.20 0.83591
Mean VIF 1.79
Table 4: Multicollinearity test Provisions
negative net income (0.2881) and the inventory to assets ratio (0.1323), none of the other correlations appear to be significant or exceed 0.1045. These correlations indicate a relatively weak to negligible correlation between the discretionary accruals and the other variables. This suggests that no single variable explains a large proportion of the discretionary accruals variation.
The Pearson correlation matrix shows no correlation coefficients which exceeds 0,8. However, it could be noticed that the correlation coefficient between full compliance and the presence of an independent chairman is relatively high: 0.7019. But, since these variables are not examined in the same regression model the possible multi-collinearity between these variables can be neglected. The coefficients between the presence of a remuneration
committee and both the presence of an audit committee, with a coefficient of 0.6538, and the
presence of a nominationcommittee, with a coefficient of 0.6440, are also relatively high.
These variables need to be checked for multi-collinearity since they are examined in the same regression model. A multi-collinearity check is done with the use of a multivariate analysis
which examines relationships among multiple variables at the same time.
A variance inflation factor (VIF) test conducts a multivariate analysis and checks if these high correlation coefficients leads to multi-collinearity in the OLS regression. The VIF test estimates whether an independent variable could be considered as a linear combination of the other independent variables. A VIF value of 10 or higher indicates that this linear
combination is likely in place. Tables 3 and 4 provide the VIF tests for an OLS regression
Variable VIF 1/VIF
Return on Assets 3.07 0.325327
Cash Flows to Assets Ratio 2.09 0.477579
Negative Net Income 1.79 0.558568
Inventory to Assets Ratio 1.49 0.672469
Debt to Assets Ratio 1.35 0.740430
Size 1.33 0.753924
Current Ratio 1.30 0.771140
Extreme Earnings Performance 1.18 0.850848
Receivables to Assets Ratio 1.15 0.867422
Full Compliance 1.05 0.955073
Mean VIF 1.58
23
based on full compliance (table 3) and based on compliance with the separate provisions (table 4). Both tables show that no variable has a VIF value higher than 3.22, which means that multi-collinearity is not an issue in the OLS regressions and that the coefficient estimates are stable and less sensitive to minor changes in the model.
4.2 Hypothesis Tests
In order to effectively perform an OLS regression with my data, the following assumptions must be met: there must be no multi-collinearity amongst the independent variables, the residuals must be normally distributed and the residual variance should be constant
(homoscedastic). The VIF test already showed that there is no multi-collinearity amongst the independent variables. The Shapiro-Wilk test is used to check if the residuals are normally distributed. This test sets the null hypothesis that the residuals are normally distributed. The p-value of this test needs to exceed the alpha level, alpha level is set at 0.05, in order to not reject the null-hypothesis. The results of this test (appendix 4) show a p-value of 0.0005 which is smaller than the pre-set alpha level of 0.05. It could therefore be concluded that the residuals are not normally distributed. Next I used the White‟s general test to check for heteroscedasticity. This test sets the null-hypothesis that the error distribution is
homoscedastic. The p-value of this test needs to exceed the alpha level, alpha level is set at 0.05, in order to not reject the null-hypothesis. The results of this test (appendix 5) show a p-value of 0.012 which is smaller than the pre-set alpha level of 0.05. It could therefore be concluded that the residuals are heteroscedastic. This means that, in the current situation, it is not possible to effectively perform an OLS regression with the data. However, it is possible to correct my data for heteroscedasticity. The data is corrected by use of robust standard errors. Robust standard errors address the problem of errors that are not independent and identically distributed. It does not change the estimated coefficients but it changes the standard errors and their significance tests. The conclusions of this paper are therefore based on the OLS
regression outcomes with robust standard errors.
Using the OLS regression with robust standard errors brings two implications. Firstly, as each observation has its own variance, the sum of the squared errors is no longer
distributed as χ2, and the ratios of the model and the residual sums of the squares are no longer distributed as an F distribution. It is therefore not possible to rely on the Prob > F value of the OLS regression since it does not have a good justification. Secondly, since each
observation contributes a different weight of information, it is not possible anymore to correct the r-square for the prognostic value the way the adjusted r-square does (one observation does
