The Effect of CEO Narcissism on Audit Risk, and the Moderating
Role of External Monitoring
Cross-Country Analysis between the UK and the Netherlands
ABSTRACT: Prior research on CEO narcissism has primarily focused on organizational outcomes, this research
extends the literature by examining the relation between CEO narcissism and the audit risk perceived by the auditor. In addition, this study investigates the moderating role of external monitoring on the relation between CEO narcissism and audit risk. Using a panel dataset of premium listed firms from the UK and listed firms from the Netherlands over the period 2013-2016, this study finds that CEO narcissism has a significant positive effect on audit risk, indicating that auditors identify more risks of material misstatement when there is a narcissistic CEO. Furthermore, there is no evidence that external monitoring moderates the relation between CEO narcissism and audit risk, and that this moderating role differs between the UK and the Netherlands. Additional analyses show no evidence for the moderating role of board independence (i.e. board independence and non-CEO duality) in both the UK and the Netherlands. This suggests that auditors do not rely on board independence, since it is likely that narcissistic CEOs adversely affected the board members. Furthermore, this study finds that there is a significant difference between the UK and the Netherlands with respect to the moderating role of institutional ownership. The results demonstrate that institutional ownership in the UK strengthens the relation between CEO narcissism and audit risk, whereas it weakens the relation between CEO narcissism and audit risk in the Netherlands. These findings can be explained by the differences in investor protection. To the best of my knowledge, this the first study that researched the influence of CEO narcissism on audit risk by making use of the long form audit reports. The risks of material misstatement presented in the long form audit reports is a more accurate measure for audit risk. The findings can be useful for public accounting firms, professional organizations and regulators who are interested in improving the audit practice. Finally, the results can be useful for public-policy makers who are concerned with structuring the governance regime of their country.
Keywords: CEO narcissism, audit risk, long form audit report, external monitoring, board independence,
non-CEO duality, institutional ownership, investor protection, legal origins, UK, Netherlands.
The Effect of CEO Narcissism on Audit Risk, and the Moderating
Role of External Monitoring
Cross-Country Analysis between the UK and the Netherlands
MSc Accountancy University of Groningen Faculty of Economics and Business
June 25, 2018
Kimberly Lensing
Student number: S2701332 Tel.: (06) 244 11 954
E-mail: kimberlylensing@hotmail.com
Supervisor: dr. Y. (Yasmin) Karaibrahimoglu Assessor: prof. dr. J.A. (Jim) Emanuels Co-assessor: G.C. (Gerard) Helminck MSc EMA RA
TABLE OF CONTENTS
I. INTRODUCTION ... 4
II. LITERATURE REVIEW AND HYPOTHESES DEVELOPMENT ... 7
CEO Narcissism ... 7
Background of Narcissism in Leadership ... 7
CEO Narcissism and Audit Risk ... 7
External Monitoring ... 9
Inherent Risk ... 9
Control Risk ... 10
Cross Country Analysis between the UK and the Netherlands ... 10
Board Structure ... 10
Investor Protection ... 11
III. DATA AND METHODS ... 12
Sample ... 12
Measurement of Variables ... 13
Dependent: Audit Risk ... 13
Independent: CEO Narcissism ... 13
Independent: External Monitoring. ... 14
Control Variables ... 15
Statistical Model... 16
IV. RESULTS ... 17
Descriptive Statistics ... 17
Descriptive statistics for the UK and the Netherlands ... 19
Pearson Correlation Matrix ... 21
Primary Test Results ... 23
Additional Analysis: Robustness Checks ... 25
Clustered Regression ... 25
Audit Partner Fixed Effects ... 25
Additional Control Variables: CEO Characteristics ... 25
Narcissism Score including Relative Pay ... 25
Additional Analysis: Inverse Mills Ratio ... 25
Additional Analysis: Components of External Monitoring as Moderator ... 26
REFERENCES ... 32
APPENDIX A: Measurement of Independent Variables ... 38
APPENDIX B: Robustness Checks H2 ... 40
4
I. INTRODUCTION
Over the last 60 years, CEOs have become more important within organizations (Quigley & Hambrick, 2014). Nowadays, they are seen as the leading figures for their organizations, and often they are treated as brands with accompanying equity (Bendisch et al., 2013). The CEO, as the leader of the organization, has the most influence on the firm’s strategic actions and therefore on the firm’s future performance (Wang et al., 2016).
Due to the increasing influence of CEOs within firms, there have been a lot of studies focusing on the question “how may CEOs influence organizational outcomes?" (Wang et al., 2016; Quigley & Hambrick, 2014). The upper echelon theory is an important theoretical framework within these studies, which argues that organizational outcomes – strategic choices and performance levels – are affected by managers’ experiences, values and personalities (Hambrick and Mason, 1984). These studies mainly focused on observable characteristics, such as gender, age, tenure, education and experience (Hambrick, 2007), while it is also meaningful to investigate more psychological characteristics, such as personality traits.
CEO narcissism is a personality trait which has gained attention in the last few years. Researchers have begun to investigate the influence of CEO narcissism on firm outcomes. They found that narcissistic CEOs are more likely to engage in complex and risky business practices, such as corporate tax avoidance (Olsen & Stekelberg, 2015; Kubick & Lockhart, 2017), overinvestment (Malmendier & Tate, 2005; Malmendier & Tate, 2008; Ham et al., 2013), merger and acquisition deals (Chatterjee & Hambrick, 2007; Malmendier & Tate, 2008; Aktas et al., 2016; Ham et al., 2013) and business activities abroad (Oesterle et al., 2016). In addition to this, they found that CEO narcissism is positively related with earnings management, financial misreporting and fraud (Hales et al. 2011; Schrand & Zechman, 2012; Rijsenbilt, 2013; Buchholz et al., 2014; Olsen et al., 2014; Ham et al., 2017; Capalbo et al.,2017). Firms with narcissistic CEOs are, thus, likely to pose higher inherent risk. Furthermore, a recent study found evidence that firms with narcissistic CEOs are more likely to have internal control weaknesses due to narcissists’ self-serving behavior and their feeling that they are above the law (Judd et al., 2017). This implies that firms with narcissistic CEOs also pose greater control risk. While all these studies imply that firms with narcissistic CEOs pose increased inherent risk and control risk, the literature from auditors side is still open to explore.
If firms with narcissistic CEOs pose greater inherent risk and control risk, it would be likely that auditors identify higher audit risk when doing risk assessments. However, very little is known about auditors’ response to clients with narcissistic CEOs, there are only two papers who have examined this. The first paper, from Johnson et al. (2013), found evidence that narcissistic client behavior and fraud motivation are positively related to auditors’ overall fraud risk assessments. However, their research was a 2x2 experiment limited to the US, whereas this study uses data from firms listed in the UK and the Netherlands. And the other paper, from Judd et al. (2017), examined that CEO narcissism has a positive effect on external audit fees, indicating that auditors should perform more work and/or charge a risk premium when their client’s CEO is relatively more narcissistic. Although these two papers already address the relationship between CEO narcissism and some type of response by the auditor, to the best of my knowledge, no study directly examines the effects of CEO narcissism on audit risk. This study intends to fill a portion of this gap in the literature, by investigating the relationship between CEO narcissism and audit risk. In order to test this relationship, this study examines whether auditors identify more risks of material misstatement in their audit reports as a response to client CEO narcissism.
Stakeholders rely on auditors because they ensure the reliability of the financial statements (Aghazadeh and Joe, 2017). However, ongoing corporate scandals and audit failures raise serious concerns about the audit profession. Auditors received much attention from the press, but also from regulators (Brockman et al., 2017; Jones, 2013). There is a major discussion going on about the role of the auditor and how to improve the audit practice in order to prevent such major scandals in the future. In order to provide answers on such questions, it is important to understand the underlying factors associated with audit risk. Prior research suggests that the observed attitudes and behavior of executives involved in corporate scandals are consistent with narcissism (Cohen et al. 2010). Therefore, it is important to know whether auditors identify narcissistic characteristics and whether they incorporate this information in their risk assessments.
Moreover, this paper extends the research about the relationship between CEO narcissism and the auditor’s response by exploring how the relation between CEO narcissism and audit risk varies with the extent of external monitoring. Narcissists are self-serving in making decisions due to their sense of self-importance, entitlement and
5
exploitativeness (Campbell et al., 2000). They are often more focused on achieving their own goals, such as enhancing their public image, instead of achieving group or organizational goals (Resick et al., 2009; Young et al., 2016). Furthermore, narcissistic CEOs are more likely to engage in risky business practices and even reflect an increased risk of financial misreporting and fraud, which negatively impact shareholder wealth (Chatterjee & Hambrick, 2007; Hales et al. 2011; Rijsenbilt & Commandeur, 2013; Schrand & Zechman, 2012; Aghazadeh & Joe, 2017). This suggests that narcissistic CEOs are less likely to act in the best interests of the shareholders. In other words, CEO narcissism increases the agency problem, which exists when there is separation of ownership and information asymmetry between owners and management (Jensen & Meckling, 1976). It is well recognized that good corporate governance can mitigate this problem, and that directors in a well monitored company are more likely to act in the interests of shareholders. Therefore, the variable ‘external monitoring’ is included into this research.
Prior studies already proved, consistent with the agency theory, that external monitoring influences the CEO’s behavior. They found that external monitoring restrains corporate risk-taking and mitigates the CEO’s power over the board (Tang et al., 2011; Kerr et al., 2016; Buyl et al., 2017; Mathew et al., 2018; Bird et al., 2018). Besides, research suggests that it decreases the CEO’s opportunity to manage earnings and to engage in fraudulent activitities (Zhu Liu, 2014; Ndofor, 2015; Zhong et al., 2017). However, up to now, there is no study that examines auditors’ perspective on the role of corporate governance within firms with narcissistic CEOs. Therefore, this paper intends to fill a portion of this gap, by investigating the moderating role of external monitoring on the relation between CEO narcissism and audit risk.
Finally, this study examines the differences between the UK and the Netherlands with respect to the moderating role of external monitoring by doing a cross-country analysis. The main differences between the UK and the Netherlands arise from the origin of legal rules. The UK is a so called common law country, whereas the Netherlands follows the French civil law (La Porta et al., 1998). Common law countries generally have better investor protection rights than civil law countries (La Porta et al., 1998). Consequently, common law countries its corporate governance rules define an one-tier board structure, while civil law countries, such as the Netherlands, usually adopt a two-tier board system (Jungmann, 2006; van Veen & Elbertsen, 2008). There is a very significant discussion going on concerning the convergence of these two systems (Jungmann, 2006). There are several studies that compared the two systems, however, scholars are inconclusive about favoring one model over the other and they have based their arguments mostly on theoretical approaches (van Veen & Elbertsen, 2008; Chen et al., 2017; Choudhuri, 2017). Yet, no study examines the effectiveness of these governance regimes from auditors’ perspective, therefore, this study aims to contribute in the ongoing discussion, by comparing the results of this study between the UK and the Netherlands.
In conclusion, the goal of this study is to contribute to the accounting literature by examining the effect of CEO narcissism (independent variable) on audit risk (dependent variable), and the moderating effect of external monitoring in both the UK and the Netherlands. The findings can be useful for public accounting firms, professional organizations and regulators who are interested in improving the audit practice. The results suggest that auditors are aware of the link between client narcissism and the increased risks with regard to the audit. Besides, it suggests that auditors do take into consideration the firm’s “tone at the top” when assessing the corporate governance structure. Narcissistic CEOs restrain board independence and surround themselves with directors that support the tone at the top created by them (Zhu & Chen, 2014; Judd et al., 2017). In response, auditors do not rely on the independence of the board (i.e. board independence and non-CEO duality), since it is likely that narcissistic CEOs adversely affected the board members. Nevertheless, narcissistic CEOs are not able to influence the monitoring role of institutional investors. Consequently, the findings show that auditors do take into consideration the impact of institutional ownership. Finally, the findings can be useful for public-policy makers who are concerned with structuring the governance regime of their country. The results of this study provide evidence that auditors do not experience differences with respect to the board structure, in other words, whether a company adopted the one-tier or two-tier system. However, the investor protection within a country does have significant impact on the audit risk perceived by the auditor.
The research question in this paper is:
Does CEO narcissism affect (the perceived) audit risk and to what extent moderates external monitoring this association in the UK and the Netherlands?
6
Figure 1: the conceptual modelThe remainder of this paper is structured as follows: the next section provides the theoretical background of the subject and presents the hypotheses. The third section describes the sample, data and methods used, and the fourth section discusses the results. Finally, the overall conclusion of the study is given in the last section.
CEO
Narcissism
Audit risk
External
monitoring
_
+
7
II. LITERATURE REVIEW AND HYPOTHESES DEVELOPMENT
CEO Narcissism
In order to discuss the expected relation between CEO narcissism and the audit risk perceived by the auditor, it is important to understand narcissism and the role of narcissism in leadership.
Narcissism is a personality trait that can be defined as “a pervasive pattern of grandiosity, self-focus, and self-importance. Narcissists are preoccupied with dreams of success, power, beauty, and brilliance. They live on an interpersonal stage with exhibitionistic behavior and demands for attention and admiration but respond to threats to self-esteem with feelings of rage, defiance, shame, and humiliation. In addition, they display a sense of entitlement and the expectation of special treatment. They are unwilling to reciprocate the favors of others and are unempathetic and interpersonally exploitative.” (Morf & Rhodewalt, 2001, p. 177).
Background of Narcissism in Leadership Narcissistic CEOs often have desirable leadership traits such
as being charismatic, confident and achievement driven (Judge et al., 2006; Rosenthal & Pittinsky, 2006; Resick et al., 2009). Confident leaders are often more risk taking and therefore better able to approach problems and challenges (Luthans et al., 2001). Narcissistic leaders have grandiose belief systems and leadership styles, they are able to attract followers and convince them of their great vision (Maccoby, 2000; Rosenthal & Pittinsky, 2006). They are motivated by their need for power, admiration and being successful, therefore, they set more difficult goals and put forth considerable effort to reach these goals (Luthans et al., 2001; Rosenthal & Pittinsky, 2006; Young et al., 2016). Therefore, a certain level of narcissism can be beneficial for organizations.
However, excessive forms of narcissism can lead to ineffectual leadership (Rosenthal & Pittinsky, 2006; Maccoby, 2004). Narcissists are self-serving in making decisions due to their sense of self-importance, entitlement and exploitativeness (Campbell et al., 2000). They are often more focused on achieving their own goals instead of achieving group or organizational goals (Resick et al., 2009; Young et al., 2016). Furthermore, narcissists were found to have greater overconfidence and willingness to take risks than non-narcissists, and this overconfidence leads to deficits in decision making (Campbell et al., 2004; Foster et al., 2009). Overconfident CEOs are more likely to overestimate their performance, they believe that they are in control even when they are not (Judge et al., 2006; Campbell et al., 2014). Moreover, highly narcissistic CEOs tend to initiate dynamic and strategic change in order to create more success and power, they want to become the center of attention (Chatterjee & Hambrick, 2007; Wang et al., 2016). Finally, narcissistic leaders are sensitive to criticism, they are considered as poor listeners, they lack empathy and have an intense desire to compete with others (Maccoby, 2004). Due to this, narcissists have a higher willingness to cheat, deceive, and exploit certain people or situations (Young et al., 2016). Consistent with this notion, research has shown that extreme narcissism is associated with unethical conduct (Duchon & Drake, 2009).
CEO Narcissism and Audit Risk Concluding from the above discussion, narcissistic CEOs have a
different type of intrinsic motivation than non-narcissistic CEOs. As a result, narcissistic CEOs manage firms differently (Young et al., 2016). The way in which the CEO enacts leadership has a huge impact on the firm’s “tone at the top” (Patelli & Pedrini, 2015). According to the upper echelon theory, the “tone at the top” is a reflection of top management’s attitudes, values and individual characteristics, and has an significant influence on a firm’s strategic choices and organizational outcomes (Hambrick & Mason, 1984; COSO 2013; Patelli & Pedrini, 2015). Due to its pervasive impact, auditors should take a CEO’s personality, and the influence on a firm’s “tone at the top”, into consideration when performing risk assessments (ISA 315; Cohen, 2002; NBA, 2012; Schmidt, 2014; Lauck et al., 2016).
According to auditing standards, auditors must make risk assessments during the audit (ISA 315). The audit risk model provides guidance in these risk assessments and describes how audit risk can be managed. This model is
8
stated as follows: Audit Risk1 = Inherent Risk (IR)2 x Control Risk (CR)3 x Detection Risk (DR)4. The auditor’s
overall objective is to obtain reasonable assurance about whether the financial statements as a whole are free from material misstatements (ISA 200). The risk that the financial statements are materially misstated, prior to the audit, consists of two components: inherent risk and control risk (ISA 200). If the client poses a high level of inherent risk and/or control risk, auditors should reduce detection risk by performing additional testing, in order to mitigate audit risk to an acceptable low level (Hogan & Wilkins, 2008). Firms with narcissistic CEOs are likely to pose higher inherent risk and higher control risk. As shown in the audit risk model, higher levels of inherent risk and control risk lead to an increase in audit risk.
Inherent Risk. In general, narcissistic CEOs are strongly motivated to achieve desirable outcomes, but at the same time, they are weakly motivated to avoid negative outcomes. They are often convinced of their own abilities and they think that they are inviolable. O’Reilly et al. (2017) found evidence that narcissistic CEOs expose their firms to undue legal risk because they are overconfident about their ability to win and they do not take the costs of such litigation into account. Consistent with these behavioral tendencies, several studies document that CEO narcissism is associated with corporate tax avoidance (Olsen & Stekelberg, 2015; Kubick & Lockhart, 2017). Furthermore, narcissistic CEOs overestimate their ability to generate returns with respect to investment projects. As a result of this, they overinvest, especially when they have access to internal funds (Malmendier & Tate, 2005; Malmendier & Tate, 2008; Ham et al., 2013). Several studies found that narcissistic CEOs tend to invest aggressively in innovation, such as research and development and new technologies (Ham et al., 2013; Gerstner et al., 2013). Besides, narcissistic CEOs favor bold actions in order to attract attention and obtain admiration. They are, for example, more likely to initiate merger- and acquisition deals (Chatterjee & Hambrick, 2007; Malmendier & Tate, 2008; Ham et al., 2013; Aktas et al., 2016). Other research has shown that narcissistic CEOs are more likely to intensify their business activities abroad (Oesterle et al., 2016). Moreover, Chatterjee and Hambrick (2007) found that CEO narcissism is positively related to strategic dynamism and grandiosity, which results in performance fluctuations and extremeness (Chatterjee & Hambrick, 2007; Tang et al., 2011; Wales et al., 2013). All these findings imply that narcissistic CEOs are more likely to engage in complex and risky business practices. Furthermore, narcissistic CEOs make decisions in order to enhance firm performance, and inherently their personal performance, with the intention of achieving self-enhancement (Campbell et al., 2000). Research have found evidence that highly narcissistic CEOs are associated with both accrual based and real earnings management in order to polish up their personal track record of their tenure in following years (Buchholz et al., 2014; Olsen et al., 2014; Ham et al., 2017; Capalbo et al., 2017). Relatedly, Hales et al. (2011) found that narcissistic CEOs inflate their reported performance in order to obtain a higher social status. Further, due to their overconfidence, narcissistic CEOs exhibit optimistic bias and, as a result of this, they are more likely to start down a slippery slope of growing intentional misstatements (Schrand & Zechman, 2012). In line with this, several studies found that CEO narcissism is positively related with financial misreporting and fraud (Hales et al. 2011; Schrand & Zechman, 2012; Rijsenbilt & Commandeur, 2013). In conclusion, all these findings have shown that firms with narcissistic CEOs pose higher financial reporting risk, in other words, these firms pose increased inherent risk.
Definitions of audit risk model according to the International Federation of Accountants (ISA 200):
1 Audit Risk (AR) is the risk that the auditor expresses an inappropriate audit opinion when the financial statements
are materially misstated. Audit risk is a function of the risks of material misstatement (inherent risk and control risk) and detection risk
2 Inherent Risk (IR) is the susceptibility of an assertion about a class of transaction, account balance or disclosure
to a misstatement that could be material, either individually or when aggregated with other misstatements, before consideration of any related controls.
3 Control Risk (CR) is the risk The risk that a misstatement that could occur in an assertion about a class of
transaction, account balance or disclosure and that could be material, either individually or when aggregated with other misstatements, will not be prevented, or detected and corrected, on a timely basis by the entity’s internal control.
4 Detection Risk (DR) is the risk that the procedures performed by the auditor to reduce audit risk to an acceptably
low level will not detect a misstatement that exists and that could be material, either individually or when aggregated with other misstatements.
9
Control Risk. In addition to higher inherent risk, the behavior of narcissistic CEOs suggests that their firms will also pose a higher level of control risk. They make decisions in order to improve perceptions of themselves, while their firms exhibit poor organizational decision making (Campbell et al., 2000; Chatterjee & Hambrick, 2007; Amernic & Craig, 2010; Olsen et al., 2014). They are, for example, more likely to engage in operational activities to improve short-term earnings, despite potential long-term negative consequences to the firm (Olsen et al. 2014). This implies that they are more focused on achieving their own goals, instead of the organizational goals (Resick et al., 2009; Young et al., 2016). Furthermore, narcissists are feeling that they are above the law, they are less likely to comply with rules and regulations (Olsen & Stekelberg, 2015; Judd et al., 2017). CEOs are legally mandated to fairly present their firms’ financial statements, and their internal control should contribute to this (IAS, 2011; COSO 2013). However, as discussed in the previous paragraph, several studies found evidence that CEO narcissism is positively related with financial misreporting and fraud (Hales et al. 2011; Rijsenbilt & Commandeur, 2013; Schrand & Zechman, 2012). In conclusion, it is less likely that narcissistic CEOs will follow rules and regulations, and it is more likely that they allow improper internal control5in order to get the freedom to act in their own interests. Consistent with these tendencies, Judd et al. (2017) found evidence that CEO narcissism is positively related with internal control weaknesses. This implies that CEO narcissism also increases control risk.
In conclusion, a client’s “tone at the top” reflects the CEO’s personality and has a significant influence on a firm’s strategic choices and organizational outcomes. Firms with narcissistic CEOs are more likely to engage in risky business practices, financial misreporting and fraud. In addition, firms with narcissistic CEOs show more internal control weaknesses. Due to its pervasisve impact on the firm’s financial reporting practices, and auditors’ ability to rely on the internal controls, it is expected that auditors identify more risks of material misstatements in their audit reports. This indicates that they perceive a higher audit risk when performing risk assessments. Therefore, the first hypothesis in this study is:
H1: CEO narcissism is associated with a higher audit risk perceived by the auditor. External Monitoring
As discussed earlier, auditors should identify and assess the risks of material misstatements. They do this by obtaining an understanding of the entity and its environment, including the ownership and governance structure (ISA 315). This is an important aspect of the risk assessments. Prior research already found that ownership and governance structures have a significant impact on a firm’s inherent risk and control risk. The underlying theory within these studies is the agency theory. The agency theory explains the mechanism through which shareholders and managers interact when there is separation of ownership. This theory suggests that, due to the separation of ownership and the information asymmetry between owners and management, the company’s manager does not always act in the best interests of the shareholders. In order to mitigate this problem, corporate governance can act as a monitoring and controlling mechanism (Jensen & Meckling, 1976).
Inherent Risk Several studies found evidence that good corporate governance is negatively related with
firm risk (Mathew et al., 2018). Bird et al. (2018) conclude in their paper that increased board independence mitigates the CEO’s power over the board and restrains corporate risk-taking. So, external monitoring leads to less extreme decisions and therefore less variability in firm performance (Bird et al., 2018). Accordingly, Tang et al. (2011) found evidence that powerful boards weaken the tendency of dominant CEOs towards extremeness, and Buyl et al. (2017) show that board monitoring discourages narcissistic CEOs to more risk-taking behavior. For example, firms with strong governance systems are less likely to engage in corporate tax avoidance (Kerr et al., 2016). Besides, good governance is also able to prevent potential overinvestment (Chan et al., 2015). In conclusion, external monitoring, especially by an independent board, mitigates a firm’s inherent business risk.
Furthermore, stronger corporate governance is found to be positively related with the financial reporting quality (Karamanou & Vafeas, 2005; Cohen et al., 2013). Karamanou & Vafeas (2005) found that more effective board monitoring leads to greater forecast accuracy. Additionally, managers under more effective monitoring by
5 Internal control is a process, effected by an entity’s board of directors, management, and other personnel, designed
to provide reasonable assurance regarding the achievement of objectives relating to operations, reporting, and compliance (COSO, 2013).
10
institutional investors are more likely to manage expectations downward instead than to manage earnings upward, so institutional ownership increases the firm’s earnings quality (Zhu Liu, 2014; Zhong et al., 2017). Consistent with this notion, Elshandidy & Hassanein (2014) found that board independence is positively associated with accounting conservatism. Several studies even suggest that good corporate governance decreases the CEO’s opportunity to engage in fraudulent activities (Ndofor et al., 2015; Mangala & Kumari, 2017). All these findings imply that external monitoring mitigates a firm’s inherent risk, due to decreased business risk and financial reporting risk.
Control Risk In addition to lower inherent risk, research suggests that external monitoring also mitigates
control risk. The CEO is accountable to the board for discharging the duties and responsibilities entrusted to him or her (FRC, 2016; MCCG, 2016). Furthermore, the monitoring role of the board, but also of institutional investors, create an environment where the CEO is not able to act in his or her own interests. The CEO has to comply with rules and regulations, which also includes a fairly representation of the financial statements. As discussed in the previous section, monitoring by the board, but also by institutional investors, increases the financial reporting quality (Karamanou & Vafeas, 2005; Cohen et al., 2013; Elshandidy & Hassanein, 2014; Zhu Liu, 2014; Ndofor et al., 2015; Mangala & Kumari, 2017; Zhong et al., 2017). This indicates that external monitoring improves the internal control environment, and as a result mitigates control risk. Consistent with this notion, a recent study found evidence that there is a negative relation between board independence and internal control weaknesses (Chen et al., 2017).
In conclusion, prior research have proved that external monitoring mitigates inherent risk and control risk, therefore, it would be likely that the perception of the auditor on audit risk changes if the extent of external monitoring varies. Previous studies have already shown that auditors identify lower control environment risk when corporate governance is strong within a firm. As a result of this, they make more favorable client acceptance recommendations, place greater reliance on internal controls and perform less extensive substantive tests (Bedard & Johnstone, 2004; Sharma et al., 2008). This suggests that auditors perceive lower audit risk when a firm has strong corporate governance. Therefore, the second hypothesis in this study is:
H2: The relationship between CEO narcissism and (the perceived) audit risk will be weakened by the extent of external monitoring.
Cross Country Analysis between the UK and the Netherlands
The third hypothesis recognizes the differences in corporate governance principles between countries. The main differences between the UK and the Netherlands arise from the origin of legal rules. The UK is a so called common law country, whereas the Netherlands follows the French civil law (La Porta et al., 1998). The most important differences between these two legal origins, concerning external monitoring, are with respect to board structure and investor protection. Common law countries generally have better investor protection rights than civil law countries (La Porta et al., 1998). Consequently, common law countries widely adopt the one-tier structure, whereas civil law countries tend to design boards according the two-tier structure (La Porta et al., 1998).
Board Structure The UK is a common law country, they adopted the one-tier board system, which
consists of both executive and non-executive directors. In a one-tier board the executives are responsible for the day-to-day management and the non-executive directors assist in strategy development and monitor the executive directors (Jungmann, 2006; van Veen & Elbertsen, 2008; Choudhuri, 2017). The Netherlands follows the French civil law principles and adopted the two-tier board system, which consists of a management board and a separate supervisory board (Jungmann, 2006; van Veen & Elbertsen, 2008; Choudhuri, 2017). Both systems, the one-tier and the two-tier system, have their strenghts and weaknesses. In a one-tier board system CEO duality or unitary leadership is permitted, this means that the CEO serves as both chairman of the board and managing director (Jungmann, 2006). In general, CEOs in a one-tier board structure, especially in the case of unitary leadership, hold greater influence over corporate decision making (Chen et al., 2017). This does not necessarily mean it leads to bad corporate performance, or decreased operating efficiency and effectiveness (Chen et al., 2017). However, as discussed, narcissistic CEO’s are strongly motivated to achieve desirable outcomes, but at the same time, they are weakly motivated to avoid negative outcomes (Olsen & Stekelberg, 2016). Therefore, I expect that in the case of CEO narcissism there is an increased risk with respect to one-tier board structures, and especially unitary leadership. Furthermore, prior research suggests that one-tier board systems restrain board independence, while
11
this is considered to be one of the most important aspects of effective board monitoring (Krivogorsky, 2006; Bliss, 2011). Two-tier boards, on the other hand, clearly distinguish executive and supervisory boards, therefore, these boards are perceived to be more independent. Consequently, increased board independence mitigates the CEO’s power and limits corporate risk-taking (Bird et al., 2018). Following these arguments, it is expected that auditors will perceive lower audit risk when there is a two-tier board system.
However, prior research focused heavily on the theoretical aspects of monitoring, while it could also be meaningful to shed light on this subject from a more practical perspective. In my opinion, the influence of an one-tier board system could be much higher than the influence of a two-one-tier board system. One-one-tier boards are much more involved within the company, they are able to obtain more detailed information about corporate decision making, since they are part of the executive board.Prior research already suggested that one-tier boards facilitate smoother information flow between the board and the managers (Kriovorsky, 2006). Two-tier boards, on the other hand, only meet a few times a year. In these meetings, the CEO informs the supervisory board only with information they want to share with the board, while one-tier boards are more able to obtain insider information, and therefore better able to monitor, advice and influence the corporate decision making. Especially in the case of narcissitic CEO’s, who are quite persuasive and manipulative (Maccoby, 2000; Rosenthal & Pittinsky, 2006), the effectiveness of external monitoring will be higher when there is an one-tier board. Although there is no empirical evidence for this reasoning, based on these arguments I would expect that auditors identify less audit risk when there is an one-tier board system. In summary, either structure has its benefits and drawbacks, therefore it remains unclear which structure would be preferred by the auditor.
Investor Protection Furthermore, there is a difference in investor protection between the two legal
origins. Common law countries generally have the strongest legal protections of investors, wheras French civil law countries have the weakest legal protections of investors (La Porta et al., 1998). These differences in legal protection have an impact on the rights attached to securities. These rights are very important for investors, since these rights give investors the power to extract from managers the returns on their investment (La Porta et al., 1998). When there is low investor protection, shareholders consider it necessary to create control and influence on the firms by themselves (La Porta et al., 1998). As a result, the monitoring role of institutional investors in countries with low investor protection is much stronger than in countries with high investor protection (La Porta et al., 1998). Based on these arguments, I would expect that the relation between CEO narcissism and audit risk will be more weakened in the Netherlands.
However, there are also papers that found evidence that institutional ownership is associated with better earnings quality, and this association is found to be stronger in countries with better investor protection (Francis & Wang, 2008; Zhong et al., 2017). It is argued that CEOs of firms located in strong investor protection countries have less ability to acquire private control benefits, as a result they are less likely to manage earnings (Zhong et al., 2017). As better earnings quality reduces audit risk (Lawson & Wang, 2016), I would expect that the relation between CEO narcissism and audit risk will be more weakened by external monitoring in the UK. However, others argue that strong investor protection regime is associated with higher earnings quality, because auditors are more inclined to perform good audits since the likelihood that auditors get punished when CEOs misreport earnings increases when there is strong investor protection (Francis & Wang, 2008). This reasoning infers that auditors perform more work and identify higher audit risk in countries with strong investor protection. In conclusion, there are only a few studies that investigated the impact of external monitoring in both investor protection regimes, little is known about the actual effectiveness of both systems (Larrain et al., 2017). Besides, there are different reasonings for the evidence found in these studies, therefore, it is unclear which regime is preffered by auditors.
It is evident that there are a lot differences between the UK and the Netherlands, though, there is no clear expectation in which country the moderating role of external monitoring will be stronger. For this reason, the third hypothesis in this study is:
H3: The moderating influence of external monitoring on the relationship between CEO narcissism and (the perceived) audit risk will be different in the UK, compared to the Netherlands.
12
III. DATA AND METHODS
The upcoming section will cover the methodology of this research. It describes the selection of the sample, it provides information about the variables and data collection, and, finally, the analysis procedure is discussed.
Sample
The starting point for this research is a sample of premium-listed firms from the UK and listed firms from the Netherlands over the period 2013-2016. This research uses the risks of material misstatement, provided by the auditor in the long form audit report, as a proxy for audit risk. Therefore, the sample need to contain firms where the auditor issues a long form audit report. Long form audit reports are, compared to traditional audit reports, more extensive and informative (ICAEW, 2017). In 2013, the UK was the first country that implemented long form audit reporting, premium-listed companies were required to include a long form audit report within their annual report (ICAEW, 2017). One year later, the NBA made it mandatory for listed firms in the Netherlands (NBA, 2014). Since the most data about audit risk is available in these countries, I included the UK and the Netherlands within the sample.
The data for both CEO narcissism and audit risk is hand collected from annual reports. In order to gather the data, we were dependent on the availability of the annual reports and the information within these reports. For firms in the UK, the data about audit risk is collected for the years 2013-2016, as the long form audit reports are only available from 2013. The same applies for firms in the Netherlands, the data about audit risk is only available from 2014, therefore the data is collected for the years 2014-2016. The latest year is 2016, because most of the annual reports for 2017 are not available yet at the time of data collection. We collected the data about CEO narcissism initially for the period 2010-2012 (for the UK) and 2011-2013 (for the Netherlands)(Chatterjee & Hambrick, 2007). However, if there was a CEO change during these years, data for the CEO narcissism measure is collected in other years6. We used lagged measurement7 in order to remove any circular or recursive relationship
between CEO narcissism and audit risk (Chatterjee & Hambrick, 2007; Judd et al., 2017).
From this sample, financial firms are excluded, because they have a unique regulatory and reporting environment, which could have different influences on perceived audit risk and the perceived role of external monitoring (Donker et al., 2009; Duellman et al., 2015). Further, all missing observations due to the absence of annual reports or due to an auditor switch eliminated from the sample. Finally, observations with missing data for audit risk and the independent variables8 (CEO narcissism and external monitoring) are excluded. For the control
variables, all the empty cells of the continuous control variables are replaced with the mean of the firm and all the indicator control variables9 with the median. However, there were still some empty cells left, because for certain
firms there was no data available. Therefore, after replacing the empty cells with the mean or median of the firm, I conducted mean and median imputation for the total sample10. After imposing these various data restrictions, a
final sample of 1155 CEO-year observations is obtained for hypothesis 1, representing 399 CEOs at 392 unique
6 In order to measure the association between CEO narcissism and audit risk, we had to take into consideration if
the same CEO was managing the firm when collecting data about narcissism and when collecting data about audit risk. If there is a CEO change in the mandatory period, the narcissism data is collected for only one year, namely the first year of the CEO’s tenure.
7 CEO narcissism is considered as a stable personality trait, therefore, we were able to make use of lagged
measurement (Chatterjee & Hambrick, 2007; Judd et al., 2017).
8 For H1, I only excluded observations if there was missing data for audit risk and CEO narcissism, and for H2
and 3, I excluded observations if there was missing data for audit risk, CEO narcissism and external monitoring.
9 Except for industry fixed effects (INDUSTRY), audit partner fixed effects (AUDITPARTNER) and audit
company (AUDITOR). The missing observations for industry fixed effects and audit partner fixed effects are excluded from the research. Note that audit partner fixed effects is only used in one additional analysis, therefore, the missing observations about audit partners are only excluded for that regression model. For the additional control variable ‘audit company’, all the empty cells are replaced with the latest reported audit company.
10 With respect to mean and median imputation, a distinction is made between the UK sample and the NL sample.
The mean or median of the UK sample is computed and all the empty cells in the UK sample are replaced with this particular mean or median, and for the NL sample the same method is used.
13
firms. For hypothesis 2 and 3, a final sample of 899 CEO-year observations is obtained. This sample represents 360 CEOs at 353 unique firms11.
Measurement of Variables
Dependent: Audit Risk The dependent variable audit risk (AUDITRISK) is measured in earlier studies,
mostly, by using external audit fees (e.g. Brockman et al., 2017; Judd et al., 2017; Lu et al., 2017). Accounting scholars theorize that external audit fees is a function of a client’s audit and business risk (Judd et al., 2017). However, audit fees are also influenced by many other factors. Therefore, this study uses another proxy for audit risk, namely the ‘risks of material misstatement’, which are provided in long form audit reports. Long form audit reports are, compared to traditional audit reports, more extensive and informative. They provide, for example, information about the group materiality, the scope of the audit, the risks of material misstatement identified by the auditor and auditors’ response to these risks (ICAEW, 2017). The ‘risks of material misstatement’ provide more directly information about the audit risk perceived by the auditor than external audit fees. In line with some working papers, audit risk is measured as the number of ‘risks of material misstatement’ (Lennox et al., 2017; Porumb et al., 2018). The more ‘risks of material misstatement’ in the long form audit report, the higher the auditor identifies a firm’s inherent risk and control risk, and, therefore, the higher the audit risk perceived by the auditor.
This data is hand collected from long form audit reports, which are incorporated in annual reports. The annual reports are obtained from www.annualreports.com or the firm’s website. We collected this data for the years in which long form audit reports are mandatory, so 2013-2016 (for the UK) and 2014-2016 (for the Netherlands). All in all, by using the ‘risks of material misstatement’ as a proxy for audit risk, this study contributes to the accounting literature, and it also contributes to the available data about audit risk. Furthermore, this research is also a response to the call from the ICAEW, to use the rich data in extended audit reports to help shine a light on audit quality (ICAEW, 2017).
Independent: CEO Narcissism The prevailing instrument for measuring narcissism is the Narcissistic Personality Inventory (NPI) (Raskin & Terry, 1988). However, the NPI is a self-reported instrument, therefore, it is difficult to obtain CEOs that are willing to complete these surveys and results could be biased by social desirability (Chatterjee & Hambrick, 2007; Olsen et al., 2014; Judd et al., 2017). Unobtrusive indicators of narcissistic tendencies have been shown to be workable and credible alternatives. Therefore, in this research CEO narcissism is measured using an adapted version of Chatterjee and Hambrick (2007). Their measurement is well used in prior accounting literature (Olsen et al., 2014; Olsen & Stekelberg, 2016; Judd et al., 2017).
Initially12, I created a composite narcissism measure using 4 of the 5 measures from Chatterjee & Hambrick
(2007)13: (1) The prominence of the CEO’s photograph in the annual report (on a 5-point scale, depending on its
size and of whether it showed the CEO alone or together with others); (2) The use of first-person singular pronouns in the CEO’s speech in the annual report; (3) The CEO’s relative cash pay; (4) The CEO’s relative non-cash pay. The Size and Prominence of the CEO’s Photograph. CEOs have a great influence on the content and design of annual reports, they have strong opinions and control on how they want to portray themselves in the annual report (Chatterjee & Hambrick, 2007). As discussed, narcissistic CEOs have an inflated self-concept, a grandiose sense of self-importance and strong desire for recognition (Morf & Rhodewalt, 2001; Campbell et al., 2000). Therefore, we attempt to capture these attributes of narcissism by using the size and prominence of the CEO’s
11 This sample is divided in an UK sample of 738 CEO-year observations (294 CEOs at 287 unique firms), and a
NL sample of 161 CEO-year observations (66 CEOs at 66 unique firms).
12 Due to data limitations with respect to the relative pay, we used in the primary tests a narcissism variable
(NARCISSISM2) which only includes the other two components: size and prominence of the CEO’s photograph and the CEO’s use of first-person singular pronouns. And in the robustness checks, we used the initial narcissism variable (NARCISSISM1) which includes all four components.
13 Chatterjee and Hambrick (2007) included an additional component in their narcissism measure, namely the
prominence of the CEO in the firm’s press releases. Due to data limitations, we excluded this component in our overall measure. Prior research also excluded this component (Olsen et al., 2014; Olsen & Stekelberg, 2016; Judd et al., 2017), and according to us the measure still reflects the diversity of narcissism traits, such as superiority, vanity, exhibitionism, self-absorption and exploitativeness.
14
photograph in the annual reports. Following prior research14 (Chatterjee & Hambrick, 2007; Olsen et al., 2014;
Olsen & Stekelberg, 2016), each photograph of the CEO is rated from one (1) to five (5) on the following scale: (1) The annual report contains a photograph of the CEO.
(2) The CEO was photographed with other executives.
(3) The CEO was photographed alone and the photograph occupies less than half of a page. (4) The CEO was photographed alone and the photograph occupies at least half of a page. (5) The CEO was photographed alone and the photograph occupies the entire page.
To ensure that every rater measures the photographs in the same way, we discussed doubtful conditions in the group and decided together the rating for those ones.
CEO’s use of First-Person Singular Pronouns. The use of first-person singular pronouns reflects superiority and self-absorption, therefore, it is an accurate measure that captures the personality traits of narcissistic CEOs. In this research, we used transcripts of the CEO’s speech in an annual report. A speech is a form of expressive behavior, it therefore reflects the most important personality traits of an individual (Chatterjee & Hambrick, 2007). We also used content analysis for the CEO’s use of first-person singular pronouns in his or her speech in the annual report. We measured this by counting the number that the CEO uses first-person singular pronouns (I, me, mine, my, myself).
Measures of Relative Pay. Relative cash pay and relative non-cash pay are considered to be reasonable unobtrusive indicators of CEO narcissism. CEOs have considerable influence over their own pay, but almost total control over the pay of other executives. Highly narcissistic CEOs believe they are more valuable for the company than any other executives. Therefore, the relative pay of the CEO, compared to other executives in the company, will be higher in the case of CEO narcissism. So, relative pay measures reflect the senses of superiority, self-importance, grandiosity and exploitativeness of narcissistic CEOs (Chatterjee & Hambrick, 2007; Judd et al., 2017). We used two measures to capture relative pay, namely relative cash pay and relative non-cash pay. Relative cash pay is measured as the CEO’s cash compensation (salary and bonus) divided by that of the second-highest-paid executive in the firm. Relative non-cash pay is measured as the CEO’s non-cash compensation (deferred income, stock grants, and stock options) divided by that of the second-highest-paid executive. The CEO’s cash compensation and the CEO’s non-cash compensation are collected from BoardEx.
Next, we performed a factor analysis to confirm that the four components are capturing the same construct. We then used the factor weightings from this analysis along with the standardized values of the four components to create a continuous summary measure of CEO narcissism. We collected the data about CEO narcissism initially for the period 2010-2012 (the UK) and 2011-2013 (the Netherlands). For these years, we are sure that the data is available and because we expect the CEO’s personality not to change over years, we were able to make use of lagged measurement (Chatterjee & Hambrick, 2007; Judd et al., 2017).
Independent: External Monitoring. The moderator external monitoring (MONITORING) is not as easy to observe, because the monitoring process is not transparent and there is no measure that fully captures the overall governance and monitoring structure of the firm. Following several previous studies, this research uses three common monitoring attributes: (1) percentage of independent non-executive directors; (2) Non-CEO duality; and (3) percentage of institutional ownership (Ahmed & Duellman, 2013; Zhu Liu, 2014; Goranova et al., 2017). These measures are included in this research given their prevalence in the accounting literature.
Board Independence. It is widely accepted that board independence mitigates the agency problem, as independent board members are better able to monitor management effectively and objectively (Fama & Jensen, 1983; Uang et al., 2006; Jizi & Nehme, 2018). Independent directors have different incentives than inside directors, they have less personal interests and they are better able to ensure that managers not pursue their self-interests at the expense of the shareholders (Jensen & Meckling, 1976; Fama & Jensen, 1983). Therefore, board independence
14 The classification of the prominence and size of the CEO’s photograph in annual reports is similar, but not
identical to the one used by Chatterjee and Hambrick (2007). We follow Olsen et al. (2014) by creating an additional category for CEO photographs that take up the whole page.
15
is considered as a component of external monitoring. Following prior literature, board independence is measured as the percentage of independent non-executive directors (Ahmed & Duellman, 2013; Goranova et al., 2017).
Non-CEO Duality. Non-CEO duality is also recognized as a proxy for board independence. According to agency theorists, CEO duality weakens board independence and thereby reducing the board’s motivation and ability to execute their monitoring and control role (Morck et al., 1989; Tuggle et al., 2008; Bliss, 2011). Prior research found evidence that CEO duality reduces board members’ monitoring behaviors, they do not consistently monitor management in order to protect shareholder value (Morck et al., 1989; Tuggle et al., 2010). Therefore, in order to measure the strength of external monitoring, non-CEO duality is included as a component of external monitoring. A dummy variable is created and set equal to 1 if there is non-CEO duality, and 0 otherwise.
Institutional Ownership. Institutional ownership is also considered as an important external monitoring mechanism (Jensen & Meckling, 1976). Large institutional investors have more incentives to monitor, due to the size of their shareholdings. They are more likely to receive special attention from management, and therefore they are more likely to hold an information advantage (Schnatterly et al., 2008; Goranova et al., 2017). As a result, they are better positioned to monitor the company. Given the aforementioned arguments, I also included institutional ownership in this research. The percentage of a firm’s outstanding shares held by institutional investors is used as a measurement for institutional investors’ monitoring intensity (Ahmed & Duellman, 2013; Zhu Liu, 2014; Goranova et al., 2017).
To create the external monitoring measure, a factor analysis15 is performed of the three components.
Thereafter, the factor weightings from this analysis are used along with the standardized values of the three components to create a continuous summary measure. Data about external monitoring is obtained from Datastream and BoardEx, and it is collected for the same years as audit risk.
Control Variables To disclose that the relationship between CEO narcissism and audit risk is being
influenced by other factors, this research included several control variables (CONTROLS). The control variables are categorized into five groups.
Following prior research about audit risk, the first group of control variables captures client complexity and client business risk (Brockman et al., 2017; Judd et al., 2017; Lu et al., 2017). Clients that are more diverse and widespread are more likely to face incremental risks, and the financial reporting process is therefore more difficult (Hay et al., 2008). As a result, the audit risks identified by the auditor could be higher. In order to control for client complexity, client firm size (CSIZE), the number of operating segments (B_SEGMENT), the amount of foreign sales (FOREIGN), the merger and acquisition activities (MERGER) and the materiality (MATERIALITY)16 are
included in this research. I also control for client business risk. Prior research already found evidence that audit fees are a reflection of clients’ business risk (Lyon & Maher, 2005; Stanley, 2011). This indicates that auditors charge a risk premium and they should perform more work when a client’s business risk is high. Therefore, measures of client business risk, such as financial leverage (LEVERAGE), market to book ratio (MTB), negative net income (LOSS), profitability (ROA), bankruptcy risk (ZSCORE), inherent risk (INHERENT), liquidity risk (LIQUIDITY) and client restatement (RESTATE), are included in the research model. Next, I control for auditor related variables. These variables represent the available resources for performing an audit, and it recognizes that different audit companies may have different risk assessment procedures. In accordance with previous studies, I included the type of business season (BUSY), audit company (BIGN), audit lag (AUDITLAG) and audit opinion (OPINION). Furthermore, in the audit risk literature, there are several studies that suggest that ownership structure has an impact on corporate governance, and therefore influences the auditor's perception (Setia-Atmaja, 2009; Mitra et al., 2017). This especially happens when financial reporting risk is high, which is expected to be the case
15 I chose for factor analysis, instead of creating a dummy as Ahmed and Duellman (2013) did in their research,
because it is a commonly used statistical method which takes the variance of the observed variable for each factor in consideration. Therefore, it is a more accurate method for creating one overall measure.
16 Materiality is not used before as a control variable, since this data was not available before long form audit
reports were initiated. I included this variable because it represents the threshold above which misstatements in the financial statements are considered as material. So, materiality could have an impact on the risks of material misstatements identified by the auditor (i.e. audit risk).
16
with CEO narcissism (Setia-Atmaja, 2009; Mitra et al., 2017). Therefore, this research included also ownership characteristics, namely closely held shares (CLOSELY_HELD) and managerial ownership (MAN_HELD). According to previous studies, the size of the audit committee, the independence and the financial expertise of the members of the audit committee have an influence on audit fees, which is a proxy of audit risk (Krishnan & Eun Lee, 2009; Contessotto & Moroney, 2014; Jizi & Nehme, 2018; Rani, 2018). Audit committees have an independent oversight role for the audit processes and the financial reporting processes. Therefore, I also control for audit committee characteristics such as audit committee size (AC_SIZE), audit committee independence (AC_INDEP) and audit committee expertise (AC_EXPERTISE). Finally, I control for industry (INDUST), country (COUNTRY) and time (YEAR) effects. All variables are further explained in Appendix A.
Statistical Model
The ordinary least square method is used in this research to estimate the following model: 𝑃𝐴𝑈𝐷𝐼𝑇𝑅𝐼𝑆𝐾= 𝐸(𝑓)
= 𝛽1+ 𝛽2 𝑁𝐴𝑅𝐶𝐼𝑆𝑆𝐼𝑆𝑀2 + 𝛽3 𝑀𝑂𝑁𝐼𝑇𝑂𝑅𝐼𝑁𝐺 + 𝛽4(𝑀𝑂𝑁𝐼𝑇𝑂𝑅𝐼𝑁𝐺 ∗ 𝑁𝐴𝑅𝐶𝐼𝑆𝑆𝐼𝑆𝑀2)
17
IV. RESULTS
Descriptive Statistics
Table 1 presents the descriptive statistics for the variables employed in this study. The average risks of material misstatement (AUDITRISK) disclosed in the long form audit reports are 3.69. This average is in line with other studies that make use of long form audit reports (Porumb et al., 2018; Lennox et al., 2017)17. The mean of
CEO narcissism (NARCISSISM2) is -0.0076, and the standard deviation, which indicates the variation of CEO narcissism within our sample, is 0.43. Both are comparable with prior research about CEO narcissism (Gerstner et al., 2013; Buyl et al., 2017)18.
In general, all control variables are in line with prior studies and appear to be reasonable given our sample (Jing et al., 2008; Olsen et al., 2014; Olsen & Stekelberg, 2016; Judd et al., 2017; Porumb et al., 2018). However, the foreign activities (FOREIGN), which is the percentage of foreign income, are low compared to Judd et al. (2017)(62%), but high compared to Olsen & Stekelberg (2016)(36%). In fact, the average of foreign activities in our study is exactly in the middle (51%). Also the average calendar days from fiscal year-end to the signature date of the auditor’s report (i.e. AUDITLAG) is different compared to prior research. The average audit lag in this study is 75 days, while Judd et al. (2017) had an average of 58 days. Additionally, the likelihood of having a financial restatement (RESTATE) in our sample is 29%, this is much higher than the average in Olsen et al. (2014) of 5%. Finally, the most remarkable exception is company size (CSIZE). Company size represents the firm’s total assets, and in other studies the average of total assets is often much higher. Judd et al. (2017) their sample had, for example, an average of 23.3 million dollar, while the average in this research is 8.1 million dollar. An explanation for this could be the countries that are included in the sample. This research focuses on the UK and the Netherlands, and in these countries firms are, in general, smaller as in the US. The working paper from Porumb et al. (2018) focused also on premium listed firms in the UK, and the mean of total assets in their research is more in line with the mean in this research.
Furthermore, it is important to note that nearly in every situation an unqualified audit opinion is provided, therefore, the mean of audit opinion (OPINION) is almost equal to zero and does not vary significantly. As a result, in several regression models, audit opinion is omitted due to the lack of variation.
17 Porumb et al. (2018) reported a mean of 4.146, and Lennox et al. (2017) reported a mean of 3.83.
18 Buyl et al. (2017) reported a mean of 0.00 and a standard deviation of 0.59, and Gerstner et al. (2013) reported
18
Table 2 provides the descriptive statistics and correlations for the components of external monitoring (MONITORING). On average, boards of companies in this research consist for 59% of independent non-executive board members. Additionally, in almost 84% of the cases, the role of the chairman and the role of the CEO are separated. In general, board independence (BOARD_INDEP) and non-CEO duality (NON_DUALITY) are in line with prior research (Weir & Lang, 2003; Jing et al., 2008; Mathew et al., 2018; Jizi & Nehme, 2018; Mathew et al., 2018). However, the percentage of institutional ownership is in our sample much lower as in comparable studies (Zhu Liu, 2014; Mathew et al., 2018). The mean in our sample is around 8%, while other studies reported a mean of 30% (Zhu Liu, 2014; Mathew et al., 2018). This might be an indication that institutional ownership has declined in the last years or that the firms included in this research (premium listed firms from the UK, and listed firms from the Netherlands), in general, have lower institutional shareholdings compared to firms included in other studies. Furthermore, the correlations among the components are provided in Panel B, almost all components are positively and significantly correlated.
Variable n M ean M edian Std. Dev. M in. P25 P75 M ax.
AUDITRISK 1,155 3.685.714 4 1.314.139 1 3 5 7
NARCISSISM 2 1,155 -0.0076384 -0.0126 0.430929 -0.68289 -0.3635435 0.24185 140.915 NARCISSISM 2_n 1,155 0.3227721 0.320399 0.2059853 0 0.1526484 0.44203 1 M ONITORING 899 -8.59E-10 0.117868 0.4357457 -1.652.419 -0.1743685 0.31316 0.884 M ONITORING_n 899 0.6514769 0.697947 0.1717956 -8.26E-08 0.582731 0.77494 1 CSIZE 1,155 8106209 1363715 2.26E+07 12584 379867 4849145 1.57E+08 B_SEGM ENTS 1,155 1.714.525 1.703.168 8.024.448 1 13 21 42 FOREIGN 1,155 0.5102256 0.5560 0.3755858 0 0.0739 0.8734 10.032 M ERGER 1,155 0.5064935 0.75 0.4982215 0 0 1 1 M ATERIALITY 1,155 6.294.384 4.897.005 4.745.546 1.282.249 3.141.722 7.807.244 281.163 LEVERAGE 1,155 0.5753649 0.5808661 0.2040907 0.084636 0.4384027 0.7141 114.334 M TB 1,155 3.543.028 2.35 4.631.345 -8.03 1.37 3.98 26.44 LOSS 1,155 0.1722944 0 0.3777997 0 0 0 1 ROA 1,155 5.420.966 6 9.927.005 -38.68 2.48 9.66 30.65 ZSCORE 1,155 0.7614719 1 0.8144788 0 0 1 2 INHERENT 1,155 0.2755483 0.2552914 0.190271 0.009043 0.1286931 0.37719 0.91046 LIQUIDITY 1,155 1.783.839 1.409.826 1.534.634 0.315109 1.024.227 194.319 12.306 RESTATE 1,155 0.2900433 0 0.4539787 0 0 1 1 BUSY 1,155 0.6588745 1 0.4742928 0 0 1 1 BIGN 1,155 0.9142857 1 0,280063 0 1 1 1 AUDITLAG 1,155 7.581.508 64 5.248.955 26 55 77 418 OPINION 1,155 0.0008658 0 0.0294245 0 0 0 1 CLOSELY_HELD 1,155 0.2190333 0.1402 0.2375866 0 0.0226 0.3479 0.8692 M AN_HELD 1,155 0.0900943 0 0.1713153 0 0 0.1 0.73 AC_SIZE 1,155 3.484.052 3 0.9463599 1.5 3 4 7 AC_INDEP 1,155 0.9487901 1 0.1453914 0 1 1 1 AC_EXPERTISE 1,155 0.252346 0.25 0.2041974 0 0 0.33333 0.77778 TABLE 1 Descriptive S tatistics
NARCISSISM2_n and MONITORING_n represent NARCISSISM and MONITORING in normalized values. All variables are defined in Appendix A.
Note: n = number of observations. Std. Dev. = standard deviation, indicates the variation of a variable. Min. = minimum value. Max. = maximum value.
This table presents the descriptive statistics for the variables employed in this study. All continuous variables are winsorized at the 1 and 99 percent levels.
19
Descriptive statistics for the UK and the NetherlandsTable 3 divides the full sample into two different samples representing the UK (COUNTRYSAMPLE=1) and the Netherlands (COUNTRYSAMPLE=0), and presents the means and standard deviations for each variable employed in this study. Furthermore, the table presents the differences in means and whether this difference is statistically significant.
The t-tests provide evidence that a lot variables between the UK sample and the NL sample are significantly different from each other. For example, the proportion of foreign income (FOREIGN) is much higher in the Netherlands, but also leverage (LEVERAGE) is on average higher in the Netherlands. In addition, the net negative income (LOSS), the bankruptcy risk (ZSCORE), inherent risk (INHERENT) and liquidity risk (LIQUIDITY) are all found to be significantly higher in the Netherlands compared to the UK. This indicates that the inherent client risk is higher in the Netherlands. Another noticeable difference is the higher percentage of inside held shares (CLOSELY_HELD) and managerial ownership (MAN_HELD). This indicates that the ownership structure in the Netherlands differs from the UK. On the other hand, the merger and acquisition activity (MERGER) is higher in the UK, just as the market to book ratio (MTB) and the return on assets (ROA). Further, there are some remarkable differences with respect to audit committee characteristics. The audit committee size (AC_SIZE) and independence (AC_INDEP) are significantly higher in the UK, while the average financial expertise of audit committees (AC_EXPERTISE) is higher in the Netherlands.
n M ean M edian Std. Dev. M in. P25 P75 M ax. 899 -2.12e-09 0.1178678 0.4357457 -1.652.419 -0.1743685 0.3131628 0.8840008 899 0.6514769 0.6979471 0.1717956 -8.26e-08 0.582731 0.779471 1 899 0.5862279 0.625 0.163604 0 0.5 0.7 0.9 899 0.839822 1 0.0944643 0 1 1 1 899 0.0799889 0.06 0.3669752 0 0 0.14 0.42 1. 2. 3. 4. 1. 1 2. 0.6913*** 1 3. 0.7709*** 0.1700*** 1 4. 0.3536*** 0.0173 0.0854** 1 TABLE 2
Descriptive S tatistics and Correlation of External Monitoring Panel A: Descriptive S tatistics of External Monitoring
Panel B: Correlations of External Monitoring
Panel A presents the descriptive statistics for the moderator and its components. All continuous variables are winsorized at the 1 and 99 percent levels.
Variable
This table presents Pearson correlations among the individual components of external monitoring (MONITORING).
*, **, *** Denote statistical significance (two-tailed) at the 10 percent, 5 percent, and 1 percent levels, respectively.
All continuous variables are standardized and winsorized at the 1 and 99 percent levels. M ONITORING BOARD_INDEP_w_std NON_DUALITY INSTIT_OWN_w_std M ONITORING M ONITORING_n BOARD_INDEP NON_DUALITY INSTIT_OWN
