The mediating effect of internationalization on the relationship between CEO narcissism and financial reporting risk

The mediating effect of internationalization on the relationship

between CEO narcissism and financial reporting risk

By Melissa Willemsen1

First supervisor: dr. Y. Karaibrahimoglu Second supervisor: prof. dr. J.A. Emanuels Third supervisor: G.C. Helminck RA MSc EMA

Date: June 25, 2018

Abstract

This research examines the mediating role of internationalization on the relationship between CEO narcissism and financial reporting risk. Understanding how CEO narcissism and internationalization can affect the financial reporting risk contributes to the existing literature and can provide valuable insights for auditors, shareholders and other stakeholders. Using panel data from 2010-2016 for companies in the Netherlands and United Kingdom, this study provides evidence that internationalization, proxied as the ratio of foreign sales to total sales, mediates the relationship between CEO narcissism and financial reporting risk. In addition, it provides evidence that CEO narcissism and internationalization directly influence financial reporting risk and that CEO narcissism directly influences internationalization.

Keywords: financial reporting risk, internationalization, CEO narcissism Word count: 8679 (excluding tables and references)

1 _{E-mail: m.willemsen.5@student.rug.nl, student number: S2729334, course: Master’s Thesis Accountancy,}

1 Table of Contents Abstract Table of Contents 1 Introduction 2 Theoretical Framework 6 CEO Narcissism 6 Internationalization 7

Financial Reporting Risk 7

Conceptual Model 8

H1: The Impact of CEO Narcissism on Financial Reporting Risk 9

H2: The Impact of CEO Narcissism on Internationalization 10

H3: The Impact of Internationalization on Financial Reporting Risk 11

H4: The Mediating Effect of Internationalization 12

Methodology 13

Data Collection and Sample 13

Measurement of the Dependent Variable 14

Measurement of the Independent Variables 14

Measurement of the Control Variables 16

Statistical Models 17 Results 19 Descriptive Statistics 19 Main Analysis 21 Robustness Checks 22 Regression Analysis 23

Another Proxy for CEO Narcissism 25

Propensity Score Matching 27

Discussion and Conclusions 29

Conclusions 29

Implications 30

Limitations and Future Research 31

References 32

2

Introduction

Throughout history there have been many narcissistic business leaders, such as Henry Ford of Ford Motor Company, Steve Jobs of Apple Incorporation and Michael Eisner of Walt Disney company (Rijsenbilt, 2011). For companies, a narcissistic CEO has been found favourable in some situations and unfavourable in others. First, there are some studies that argue that narcissism has a positive impact on companies. Olsen et al. (2014) have shown that companies with narcissistic CEOs have a higher share price and higher earnings-per-share (EPS) than companies that do not have narcissistic CEOs. They suggest that narcissistic personality characteristics of CEOs have influence on the financial performance measures through their influence over the companies’ operational activities, rather than through accrual and accounting decisions. Moreover, narcissism indirectly has a positive effect on innovation through adaptability, which is important both within and outside an organization (Smith and Webster, 2018). Second, there are also several studies that found a negative impact of narcissism on companies. For example, it creates a weaker internal control quality, a higher probability of restatements (Ham et al., 2017), a lower probability of deal completion by acquisitions (Aktas et al., 2016), extreme fluctuations in the performance of a company (Chatterjee and Hambrick, 2007) and an aggressive use of technological discontinuities (Gerstner et al., 2013).

In addition to the mentioned examples, narcissistic executives can affect a company’s financial reporting risk. There are many reasons for executives to conduct dishonest or unethical behaviour in a company’s financial report. Reasons to overstate the company’s performance in financial reports are for example financial trade-off (Hannan et al., 2006), increasing of monetary payoffs (Maas and van Rinsum, 2013) and presence of opportunities for social comparison (Hales et al., 2012). Especially narcissistic executives are more often associated with unethical behaviour because they lack a moral identity (Duchon and Drake, 2009). Therefore, the objective of this paper is to determine how the presence of CEO narcissism affects the risk that there are material mistakes in financial reporting.

Furthermore, internationalization decisions have become a more essential part of the strategic challenges a firm faces due to the globalization of the world economy (Schotter and Beamish, 2013). This leads to changes in financial reporting rules, as illustrated by the European Union’s adoption of International Accounting Standards (IAS) and International Financial

3

Reporting Standards (IFRS) from 2005. According to IFRS, internationalization decisions are important for companies’ financial reporting risks. Furthermore, these internationalization decisions reflect a prime example for situations in which a CEO is able to achieve personal goals (Oesterle et al., 2016). This is because narcissistic CEOs pursue in self-serving actions (Brown et al., 2010) and choose often for outstanding, risky and spectacular projects (de Vries and Miller, 1985). Therefore, the influence of internationalization decisions on the relationship between CEO narcissism and financial reporting risk is examined in this study. To demonstrate this relationship, the direct relations between CEO narcissism and internationalization and internationalization and financial reporting risks are examined as well.

Given this, the main research question of this study is “What is the mediating effect of

internationalization on the relationship between CEO narcissism and financial reporting risk?” This research question can be divided into three sub-questions. The first sub-question

addresses CEO narcissism and financial reporting risk: “What is the relation between CEO

narcissism and financial reporting risk?”. The second sub-question describes CEO

narcissism and internationalization: “What is the relation between CEO narcissism and

internationalization?”. The third sub-question describes internationalization and financial

reporting risk: “What is the relation between internationalization and financial reporting

risk?”.

By answering these questions, this research contributes to the existing literature on CEO narcissism. Few researchers have addressed the relation between CEO narcissism and financial reporting risk. Most studies have focused on other reasons why executives conduct dishonest or unethical behaviour in company’s financial reports (Hannan et al., 2006; Maas and van Rinsum, 2013). Only Hales et al. (2012) have focused on how social status and narcissism affect the willingness of managers to report honestly, by using an experimental design. They found that increasing the salience of social status will increase over-reporting. They also found that narcissistic managers will over-report more, particularly when performance is viewed important. Understanding how CEO narcissism can affect the financial reporting risk contributes to the existing literature and can provide valuable insights for auditors, shareholders and other stakeholders.

4

In addition, this study is also relevant because it investigates the mediating role of internationalization. Oesterle et al. (2016) were the first study that investigates CEO narcissism in an international business context. They only investigated the 31 largest German manufacturing companies for the years 2004-2013. If the assumed relation – the more narcissistic the CEO, the higher the growth of a company’s degree of internationalization – can be identified for companies in the Netherlands and United Kingdom, this study might verify the insights of their research. Furthermore, they did not make a connection with financial reporting risk. Because international operations are connected to benefits but also to some threats, it is relevant for stakeholders and the company itself to understand if narcissism has influence on these internationalization decisions.

Regarding the relation between internationalization and financial reporting risk, there is not a vast amount of research. Therefore, this study contributes to this area in the literature. However, Glassman (2006), commissioner for the SEC, has expressed concerns about the complexity of international accounting standards that can lead to more financial reporting risks. In this context, this study is relevant for international shareholders, creditors and other stakeholders because they need to understand the impact of internationalization on the financial reports they receive.

Concluding, this study contributes to the existing literature by studying companies from the United Kingdom and the Netherlands for the years 2010-2016. Therefore the study, and mainly the internationalization decisions, is not biased by the financial crisis. As a result, this research provides much recent and unique information about narcissism and the consequences.

The results show a significant positive effect of the relation between CEO narcissism and the company’s financial reporting risks, which means that when companies have a narcissistic CEO, the companies’ financial reporting risks are higher than when companies have no narcissistic CEO. In addition, the findings indicate a significant positive relationship between CEO narcissism and internationalization. This means that companies have more international activities when they have a narcissistic CEO. However, this relationship is not found in the robustness tests. The relationship between internationalization and financial reporting risk is significant and positive, which means that when a company performs in international activities, the number of companies’ financial reporting risks is higher. Regarding the

5

mediation, the results show a mediating effect of internationalization on the relationship between CEO narcissism and financial reporting risk.

The remainder of this paper is structured in the following way. The next section reviews prior literature and develops the research hypotheses. The third section describes the methodology, including the sample, measurement of variables and statistical models. In the fourth section the results are presented. The fifth section shows three different robustness checks. The sixth section gives a discussion, conclusions, limitations and recommendations for future research.

6

Theoretical Framework

In this chapter, first the different concepts used in this study will be explained. Thereafter, the conceptual model is presented. This chapter will end with a development of the hypotheses.

CEO Narcissism

Since CEOs have much influence within a company, researchers have been interested in their personal characteristics and how those characteristics are expressed in the results of an organization over a long period of time (Chatterjee and Hambrick, 2007). An important personal characteristic is narcissism because of its ability to manipulate other people, to exercise power and the drive to achieve prestige (Lubit, 2002). Narcissism is a personality disorder that has been investigated for different purposes in the last decade. It can be derived from a combination of genetic factors and early parental relations (Livesley et al., 1993). This paper focuses on CEOs who are narcissistic by the definition of The American Psychiatric Association (2013), which is defined in The Diagnostic and Statistical Manual of Mental Disorders (DMS). They define narcissism as “a pervasive pattern of grandiosity (in fantasy or behaviour), need for admiration and a lack of empathy, beginning by early adulthood and present in a variety of contexts.”

As more researchers did before (e.g. Oesterle et al., 2016), I draw this study on upper echelons theory. The upper echelons theory stated that organizational outcomes are affected by managers’ experiences, values, and personalities (Hambrick and Mason, 1984). This means that different individuals in identical situations can make different, and even opposing decisions. According to this theory, the perceptions, cognitions and values of people manifest themselves in the decision making process. Upper echelons theory is important for this research because narcissism is a personality characteristic and, according to this theory, therefore it can influence the CEO’s decisions and ultimately the organizational outcomes. In addition to upper echelons theory, the agency theory is also used. Jensen and Meckling (1976) have defined the agency theory as “a contract under which one or more persons (the principals) engage another person (the agent) to perform some service on their behalf which involves delegating some decision making authority to the agent” (p. 308). An agency problem arises when the principals do not know what the agent is doing or when the agent does not perform in the interest of the principals because of his self-interest (Eisenhardt,

7

1989). This leads to information asymmetry between the agent and the principals. The agency theory can be applied in this study because narcissistic CEOs tend to act in their own self-interest, as the agent (Rijsenbilt, 2011). Furthermore, the information asymmetry will make it less likely for principals to reveal fraud in the financial statements of a company (Jensen and Meckling, 1976).

Internationalization

International activities are defined as “the extension of business activities across the borders of the domestic country” (Hitt et al., 1994, p. 298). There are different reasons for companies to perform in international activities. Dunning (1973) has shown that lower costs of production or higher labour productivity can be one of the reasons. He also stated that the attitude of the foreign government, political stability, and the prospect for market growth are important aspects. In addition, Kogut and Zander (1993) discovered that the transfer of organizational knowledge across the border is another reason for internationalization.

However, in addition to the benefits of international activities, there are certain risks for performing in international activities. To begin with, international activities provide more coordination costs, information costs and transaction costs (Fisch, 2012). This is the result of the unfamiliar environments since there are economic, cultural and political differences (Oesterle et al., 2016). The same study concludes that the geographical distance between the foreign market and the home market makes coordination between these entities difficult. Another risk that arises from internationalization activities and is often highlighted in the literature is the foreign currency risk (Bartram and Bodnar, 2007). This is the risk that appears when new accounts are introduced in the companies’ financial reports.

Financial Reporting Risk

In market-based economies, there are two important functions of financial reporting. First, it reduces information asymmetry and gives capital providers the opportunity to value companies. Hence, providing the transparency needed for capital markets to function efficiently. Second, financial reports allow external capital providers to monitor the performance of the agreement (Pinnuck, 2012). Therefore, it is important that financial reports include as few risks of material misstatement as possible. In addition, transparency about possible mistakes in the financial reports is important.

8

Prior literature shows that there are many reasons for executives to conduct dishonest or unethical behaviour in companies’ financial reports. According to the study of Hannan et al. (2006), executives trade off the financial advantages of misrepresentation with the intangible utility they deduct from appearing honest. In a later research, Maas and van Rinsum (2013) show that executives report less truthfully when their reported performance increases others’ monetary payoffs than when it decreases others’ monetary payoffs. In addition to the monetary reasons, there are more reasons that can influence the behaviour of the executives. For example, Hales et al. (2012) determined how the presence of opportunities for social comparison affects the honesty of managers in financial reports.

In 2001, the IASB announced a project that introduces accounting standards that further improves the transparency of companies’ financial reports (Mayorga and Sidhu, 2012). This project rectifies IAS 1 and requires companies to disclose major sources of estimation uncertainty. Nowadays, executives of a company indicate in their annual report their estimates regarding the value of items, which can incorporate risks of material misstatements. These uncertainties can arise from several sources, which managers derive based on their assessment of the present and future benefits and obligations associated with assets and liabilities (Mayorga and Sidhu, 2012). It is important that investors understand the company’s important accounting estimates that reflect management judgement and uncertainty. Hence, they can better understand how these estimates impact the company’s financial position, results of operations and changes in financial position (Mayorga and Sidhu, 2012).

Conceptual Model

In Figure 1, this study’s conceptual model is presented. The related hypotheses are described in the next paragraphs. To give a short explanation of the model, CEO narcissism is expected to be positively related to financial reporting risk and internationalization. Internationalization is expected to be positvely related to financial reporting risk. In addition, internationalization is expected to mediate the relationship between CEO narcissism and financial reporting risk.

9

Figure 1: Conceptual Model

H1: The Impact of CEO Narcissism on Financial Reporting Risk

Drawing on preliminary work, I expect that there is a positive relationship between narcissistic CEOs and the company’s financial reporting risk. Prior research expects this positive relation for the following reasons.

First, as mentioned earlier, there is a vast amount of literature determining that people are willing to be dishonest for a monetary return. According to Maas and van Rinsum (2013), the actual level of managers’ overstatement of their performance depends on how individual performance reports will impact on other managers at the same hierarchical level. Overstatement of performance can result in conducting unethical or dishonest behaviour in company’s financial reports.

Second, different studies have shown that narcissists take more risks than others. These studies often examined specific risky activities, such as aggressive driving (Britt and Garrity, 2006) and gambling (Lakey et al., 2008). However, Foster et al. (2009) investigated why narcissists engage in risky behaviours. They concluded that “narcissists engage in dysfunctional behaviours because of a surplus of eagerness, not a deficit of inhibition” (p. 889). In addition, Zhu and Chen (2015) stated that narcissistic CEOs tend to take risky actions to gather admiration and attention.

10

Third, Duchon and Drake (2009) suggested that extreme narcissistic CEOs are more often associated with unethical business practices because they lack a moral identity. According to their study, narcissists do not necessarily act intentionally unethical, but they become self-obsessed and use sense of entitlement, denial and rationalizations to justify the things they do. Amernic and Craig (2010) added texture to this conclusion. They argue that the special features possessed by financial accounting facilitate this behaviour.

Last, Hales et al. (2012) concluded in their research that narcissism leads to a higher probability of misstatements. In addition, Schrand and Zechman (2012) argue that narcissistic CEOs are more likely to exhibit an optimistic bias and thus are more likely to start down a slippery slope of growing misstatements.

In line with the above arguments, the first hypothesis is:

H1. CEO narcissism is positively associated with financial reporting risk.

H2: The Impact of CEO Narcissism on Internationalization

I expect that there is a positive relationship between CEO narcissism and internationalization. The arguments in the literature regarding international activities can be divided into two groups. First, there are companies that internationalize through cross-border sales. Second, there are companies that internationalize through cross-border acquisitions. The positive relationship is expected because of the following reasons.

The study of Oesterle et al. (2016) was the first study that investigated the direct impact of CEO narcissism on internationalization decisions. They found that the more narcissistic a CEO is, the higher the growth of a company’s foreign sales relative to their total sales is. According to Oesterle et al. (2016), a possible explanation can be found in altering narcissistic tendencies of CEOs, which lead to hubris and shortcomings in decision-making when entering international markets.

According to the study of Zhu and Chen (2015), CEOs with a relatively low narcissistic tendency place emphasis on stability of the company. This could mean that CEOs with a relatively low narcissistic tendency are likely to perform in less cross-border acquisitions because this has influence on the stability of the company.

11

De Vries and Miller (1985) concluded that narcissists often choose outstanding, spectacular and risky projects. This can result in overconfidence in performing international acquisitions. In addition, due to the overconfidence of narcissists, executives can make errors in assessing the intrinsic value of cross-border acquisitions (Dutta et al., 2016). However, they acquire the company anyway because they think their assessment of the intrinsic value is correct (Seth et al., 2000). Narcissistic CEOs often overestimate their ability to generate returns from a project, which results in the overpaying for target companies and undertaking of value-destroying mergers (Malmendier and Tate, 2008). In addition, they pay non-justified acquisition premiums (Hayward and Hambrick, 1997).

In line with the above arguments, the second hypothesis is:

H2. CEO narcissism is positively associated with internationalization. H3: The Impact of Internationalization on Financial Reporting Risk

I expect that there is a positive relationship between internationalization and financial reporting risk. This relationship is expected for the following reasons.

Nowadays, there is a world economy in which investments and trades show no regard for national borders. The large growth of the multinationals has been followed by an increasing demand for understandable, comparable and assessable financial reporting by shareholders, investors and creditors all over the world (Ding et al., 2008).

First, internationalization leads to more complex financial reporting. According to Glassman (2006), who is a commissioner for the SEC, international accounting standards have enhanced to give a false picture of a company’s financial position, due to the complexity of the standards. The various exceptions of the principles may facilitate overly aggressive accounting, which encourage companies to take advantage of the complexity and misleading disclosure. Therefore, more complex financial reporting leads presumably to more risks of material misstatements in the financial reports.

Second, another risk that arises directly from internationalization activities is the foreign currency risk (Bartram and Bodnar, 2007). The increase of a firm’s sales into foreign countries will introduce accounts in the company’s financial statements that are translated from foreign currencies (Lee and Jang, 2010). According to their study, these accounts will

12

appreciate or depreciate in a volatile way with respect to movements in foreign currency values, which is affecting the firm value. In addition, fluctuating exchange rates would increase the risks of operations for companies, which invested across currencies (Hekman, 1981). So, it is a common agreement that exchange rates affect the value of multinational companies (Bartov and Bodnar, 1994).

Therefore the third hypothesis is:

H3. A higher degree of internationalization has a positive effect on the financial reporting risk.

H4: The Mediating Effect of Internationalization

As shown in the first hypothesis, CEO narcissism seems to be positively related to financial reporting risk. Furthermore, the second hypothesis predicts that CEO narcissism is positively related to internationalization. Last, hypothesis 3 implies that internationalization is positively related to financial reporting risk. Hypothesis 4 predicts that hypothesis 2 and 3 are stronger than hypothesis 1. This indicates the mediation of internationalization on the relationship between CEO narcissism and financial reporting risk. This is expected because more CEO narcissism leads to more international activities (Oesterle et al., 2016) and more international activities leads to more complex financial accounting (Glassman, 2006).

Therefore, regarding the main question, the fourth hypothesis is:

H4. Internationalization mediates the relationship between CEO narcissism and financial reporting risk.

13

Methodology

In this chapter, there will be explained how the data of this study has been obtained and how the sample has been composed. Next, the measurement of the dependent variable, the independent variables and the control variables will be explained. Finally, the statistical models are presented.

Data Collection and Sample

Data collection. This research made use of archival data that was obtained from

different databases. First, the CEO data was requested from the BoardEx database, which is available through Wharton Research Data Services (WRDS). The CEO data included name, company, age, gender, tenure, network size and qualifications. Second, firm-level information, such as total sales, total assets, and foreign sales were added to the dataset from the Datastream database. Last, the CEO narcissism data and financial reporting risk data were hand-collected by eight students from the annual reports. This data is all collected for the years 2010-2016, except for the narcissism data. I have chosen not to include the years before 2010, because then the financial crisis was at its peak, which influences the internationalization decisions of a company (Costa et al., 2017). The year 2016 is chosen because it was the last year for which data was available.

Sample. The sample selection process was started by selecting listed companies in the

United Kingdom and the Netherlands, except financial companies. Financial companies were excluded because these companies are commonly examined within industry due to significant differences in their financial statements. The initial sample consisted of 3622 firm-year observations from 536 companies over the 2010-2016 period.

The CEOs who were examined are the ones who lead a listed company in the United Kingdom or the Netherlands. I faced data restrictions for the variable “CEO narcissism” for which I followed the model of Chatterjee and Hambrick (2007). Following their model, I only used two years of tenure to calculate the narcissism score and not the whole tenure because narcissism is a stable personality construct (Chaterjee and Hambrick, 2007). Therefore, I only used narcissism information for the years 2011 and 2012, where it was available. However, due to data limitations, I used other years when this data was not available.

14

During the process, I also eliminated observations with missing information for the narcissism score, foreign sales and financial reporting risks. To reduce data limitation, I used the average for the missing information for the control variables. The final sample consisted of 2141 firm-year observations, of which 82 are Dutch companies (390 firm-firm-year observations) and 322 are UK companies (1751 firm-year observations).

Measurement of the Dependent Variable

Financial reporting risk. The data about financial reporting risk was hand-collected

from the companies’ annual reports. According to IAS 1, paragraph 125, a company needs to disclose information about the assumptions the company makes about the future. They also need to disclose other major sources of estimation uncertainty at the end of the reporting period, that have significant risks of resulting in a material misstatement to the carrying amounts of assets and liabilities within the next financial year. Following a part of the method of Mayorga and Sidhu (2012), the annual reports of each company were reviewed to search if the company has accounts that are likely to be subject to estimation uncertainty. Next, the minimum number of risks for all companies was estimated, which turned out to be 1. For the analyses, the abnormal number of risks for every company was used, because some companies have a number of risks that are always in their financial reports and add no value anymore (for example revenue recognition). The abnormal number of risks is calculated by subtracting the minimum number of risks from the total number of risks disclosed.

Measurement of the Independent Variables

CEO narcissism measure. The CEO narcissism score was measured by using the

validated research of Chatterjee and Hambrick (2007). They designed a proxy that is based on modest indicators of CEO’s narcissistic tendencies. The CEO’s narcissism is measured through the following indicators: the prominence of the CEO’s photograph in annual reports, the prominence in the company’s press releases, the CEO’s use of first-person singular pronouns in interviews, and the CEO’s relative cash and non-cash compensation. In this study, the prominence in company’s press releases was not used because of data limitations. The first indicator used in this study is the prominence of the CEO’s photograph in annual reports. According to Chatterjee and Hambrick (2007), in the company’s annual report CEOs have an opportunity to present themselves as the company’s leader. It is expected that highly narcissistic CEOs will search for visibility in their annual report. The indicator will be ranked

15

(1) (2) (3) (4) (5)

Variable N Mean SD Min Max

Picture score 229 2.533 2.738 0.000 5.000

First-person singular pronouns 229 2.738 4.624 0.000 44.000

Relative cash pay 229 9.800 9.590 0.150 81.640

Relative non-cash pay 229 1.580 5.750 0.000 64.790

Narcissism score 229 -0.020 0.351 -1.425 1.717

TABLE 1

Descriptive Statistics - Narcissism

on a scale from 1-5: 1 point when there is no picture, 2 points when the CEO is on the picture with other executives, 3 points when there is a photo of the CEO alone that is less than a half page, 4 points when there is a photo of the CEO alone that is more than a half page and 5 points when there is a photo of the CEO alone that covers the whole page.

The second indicator that was used in this study is the CEO’s use of first-person singular pronouns in interviews. According to Ramsay (1968) and Hogben (1977), a speech is a form of expressive behaviour and reflects the most dominant and consistent personality traits of a person. For this, I used the CEO’s speech in the annual report, which is often called the CEO’s statement. To calculate the score, I counted the number of first-person singular pronouns (I and my) in the speeches, which ranged between 0 and 44, as shown in Table 1. The third indicator is the CEO’s relative cash and non-cash compensation. Chatterjee and Hambrick (2007) stated that CEOs have influence in their own pay and the pay of other executives. The relative cash pay will be measured through the CEO’s cash compensation (salary and dividend) divided by that of the second-highest-paid executive in the company. The relative non-cash pay (deferred income, stock options, stock grants) will be measured in the same way.

These three indicators together compute the narcissism score. In Table 1, the descriptive statistics regarding this narcissism score are presented. The picture score has a mean of 2.533 and a minimum and maximum of 0 and 5, respectively. The first-person singular pronouns are on average 2.738. The average relative cash pay is 9.800 and the average relative non-cash pay is 1.580.

16

Internationalization. According to Oesterle et al. (2016), internationalization is a

promising way to realize growth strategies. Therefore, following their study, I measured the internationalization through “the growth of a company’s degree of internationalization (DOI)”. The DOI is measured by the ratio of foreign sales to total sales (FSTS). According to Aggarwal and Samwick (2003), this is the most commonly used measure for internationalization.

Measurement of the Control Variables

In order to make sure that the dependent variable ‘financial reporting risk’ is explained by the independent variable ‘CEO narcissism’ and mediating variable ‘internationalization’, control variables were included to control for the confounding effects of these variables. These control variables are separated in firm characteristics, CEO characteristics and fixed effects.

Firm characteristics. First, I controlled for firm size. This is important because old

and large companies may face bureaucratic issues (Chatterjee and Hambrick, 2007). In addition, firm size is typically related to extensive international activities, at which larger companies are more likely to engage in international activities (Oesterle et al., 2016). Firm size is measured by the natural logarithm of total assets. Firm performance is another important control variable. This is also interpreted as being related to extensive international activities (Oesterle et al., 2016). Therefore, I controlled for firm performance by using the company’s return on assets (ROA). Additionally, I controlled for closely held shares, which are the shares that are owned by a few dominant insiders. Last, I controlled for financial leverage by using the market-to-book ratio.

CEO characteristics. In addition to the firm characteristics, I controlled for CEO

characteristics. First, I controlled for the CEO’s age and tenure. Prior studies have shown that the tendency to engage in grandiose or dynamic strategies may vary with age or tenure (Chatterjee and Hambrick, 2007). Also for the internationalization decisions, the study of Oesterle et al. (2016) has controlled for CEO’s age because it might be an indicator of decision-making confidence. Second, I controlled for CEO gender through a dummy where a female CEO is 0 and a male CEO is 1. Third, I controlled for CEO’s network size. This is measured as the number of individual’s with whom the CEO overlaps. This can be in education roles, in employment or in other activities at the same company. Last, I controlled

17

for the number of qualifications of the CEO. This is measured as the number of qualifications earned from an institution.

Fixed effects. Last, I controlled for some fixed effects. First, I controlled for

industry-fixed effects, since different industries might attract different CEOs. The distinction between industries is made on two digit SIC codes. The following industries are distinguished: agriculture and forestry, mining, construction, transportation and public utilities, manufacturing, wholesale trade, retail trade, services, insurance and real estate, and public administration. Second, I controlled for year-fixed effects, where the years 2010 until 2016 are distinguished. Last, I controlled for country-fixed effects, where I made a distinction between companies headquartered in the United Kingdom and companies headquartered in the Netherlands. In addition, I added some companies that are listed on the UK or Dutch stock exchange.

Statistical Models

This study tested the relationship between CEO narcissism, internationalization and financial reporting risk. Table 2 provides a summary of the variables including their labels and proxies. Due to the mediator, I made use of four different models.

The first model:

𝐹𝑅𝑅𝐼𝑆𝐾 = 𝛽_!+ 𝛽_!𝑁𝐴𝑅𝐶𝐼𝑆 + 𝛽_!𝐹𝐼𝑅𝑀𝑆𝐼𝑍𝐸 + 𝛽_!𝑃𝐸𝑅𝐹𝑂𝑅𝑀 + 𝛽_!𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸 + 𝛽_!𝐶𝐿𝑂𝑆𝐸𝐻𝐸𝐿𝐷 + 𝛽_!𝐶𝐸𝑂𝐴𝐺𝐸 + 𝛽_!𝐶𝐸𝑂𝑇𝐸𝑁𝑈𝑅𝐸 + 𝛽_!𝐶𝐸𝑂𝑁𝐸𝑇𝑊𝑂𝑅𝐾 + 𝛽_!𝐶𝐸𝑂𝑄𝑈𝐴𝐿

+ 𝛽_!"𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 + 𝛽_!!𝐶𝑂𝑈𝑁𝑇𝑅𝑌 + 𝛽_!"𝑌𝐸𝐴𝑅 + 𝛽_!"𝐺𝐸𝑁𝐷𝐸𝑅 + 𝜀

The second model:

𝐼𝑁𝑇𝐸𝑅 = 𝛽!+ 𝛽!𝑁𝐴𝑅𝐶𝐼𝑆 + 𝛽!𝐹𝐼𝑅𝑀𝑆𝐼𝑍𝐸 + 𝛽!𝑃𝐸𝑅𝐹𝑂𝑅𝑀 + 𝛽!𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸 + 𝛽!𝐶𝐿𝑂𝑆𝐸𝐻𝐸𝐿𝐷

+ 𝛽_!𝐶𝐸𝑂𝐴𝐺𝐸 + 𝛽_!𝐶𝐸𝑂𝑇𝐸𝑁𝑈𝑅𝐸 + 𝛽_!𝐶𝐸𝑂𝑁𝐸𝑇𝑊𝑂𝑅𝐾 + 𝛽_!𝐶𝐸𝑂𝑄𝑈𝐴𝐿 + 𝛽!"𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 + 𝛽!!𝐶𝑂𝑈𝑁𝑇𝑅𝑌 + 𝛽!"𝑌𝐸𝐴𝑅 + 𝛽!"𝐺𝐸𝑁𝐷𝐸𝑅 + 𝜀

The third model:

𝐹𝑅𝑅𝐼𝑆𝐾 = 𝛽_!+ 𝛽_!𝐼𝑁𝑇𝐸𝑅 + 𝛽_!𝐹𝐼𝑅𝑀𝑆𝐼𝑍𝐸 + 𝛽_!𝑃𝐸𝑅𝐹𝑂𝑅𝑀 + 𝛽_!𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸

+ 𝛽₅𝐶𝐿𝑂𝑆𝐸𝐻𝐸𝐿𝐷 + 𝛽₆𝐶𝐸𝑂𝐴𝐺𝐸 + 𝛽₇𝐶𝐸𝑂𝑇𝐸𝑁𝑈𝑅𝐸 + 𝛽₈𝐶𝐸𝑂𝑁𝐸𝑇𝑊𝑂𝑅𝐾 + 𝛽₉𝐶𝐸𝑂𝑄𝑈𝐴𝐿+ 𝛽_!"𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 + 𝛽_!!𝐶𝑂𝑈𝑁𝑇𝑅𝑌 + 𝛽_!"𝑌𝐸𝐴𝑅 + 𝛽₁₃𝐺𝐸𝑁𝐷𝐸𝑅 + 𝜀

18

The fourth model:

𝐹𝑅𝑅𝐼𝑆𝐾 = 𝛽_!+ 𝛽_!𝑁𝐴𝑅𝐶𝐼𝑆 + 𝛽_!𝐼𝑁𝑇𝐸𝑅 + 𝛽_!𝐹𝐼𝑅𝑀𝑆𝐼𝑍𝐸 + 𝛽_!𝑃𝐸𝑅𝐹𝑂𝑅𝑀 + 𝛽_!𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸 + 𝛽!𝐶𝐿𝑂𝑆𝐸𝐻𝐸𝐿𝐷 + 𝛽!𝐶𝐸𝑂𝐴𝐺𝐸 + 𝛽!𝐶𝐸𝑂𝑇𝐸𝑁𝑈𝑅𝐸 + 𝛽!𝐶𝐸𝑂𝑁𝐸𝑇𝑊𝑂𝑅𝐾

+ 𝛽_!"𝐶𝐸𝑂𝑄𝑈𝐴𝐿 + 𝛽_!!𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌 + 𝛽_!"𝐶𝑂𝑈𝑁𝑇𝑅𝑌 + 𝛽_!"𝑌𝐸𝐴𝑅 + 𝛽_!"𝐺𝐸𝑁𝐷𝐸𝑅 + 𝜀

Variable Label Proxy

Financial Reporting Risk FRRISK Abnormal risks disclosed CEO Narcissism NARCIS Computed CEO narcissism score Internationalization INTER Ratio of foreign sales to total sales Firm Size FIRMSIZE Natural logarithm of total assets Firm Performance PERFORM Return on assets (ROA) Financial Leverage LEVERAGE Market-to-book ratio Closely Held Shares CLOSEHELD Shares held by insiders

CEO Age CEOAGE Age of CEO in years

CEO Tenure CEOTENURE Tenure of CEO in years

CEO Network CEONETWORK Size of the CEO network

CEO Qualifications CEOQUAL Number of qualifications earned by a CEO

Industry INDUSTRY Industry dummies

Country COUNTRY Country dummies

Year YEAR Year dummies

CEO Gender GENDER Gender dummies (male/female)

TABLE 2 Description of Variables

19

(1) (2) (3) (4) (5)

Variable N Mean SD Min Max

CEO Narcissism 1842 -0.005 0.400 -0.683 2.736

Internationalization 1842 50.577 37.726 0.000 100.000

Financial Reporting Risk 1842 4.246 2.252 -1.000 19.000

Firm Size 1842 6.081 0.852 2.350 8.633

Financial Leverage 1842 3.038 13.354 -351.980 297.860

Firm Performance 1842 6.704 9.703 -58.300 109.710

Closely Held Shares 1842 20.655 21.106 0.000 91.060

CEO Age 1842 52.066 5.786 31.000 76.000

CEO Tenure 1842 6.153 4.769 0.000 40.400

CEO Network Size 1842 676.318 846.270 0.000 4535.000

CEO Qualifications 1842 1.702 1.064 0.000 6.000

TABLE 3 Descriptive Statistics

Results

In this chapter, the results will be discussed. First, the descriptive statistics are presented, including the correlations. Thereafter, the results of the main analysis are explained.

Descriptive Statistics

Table 3 provides the descriptive statistics of the final sample over the 7-year sample period of 2010-2016. The table shows the means, medians, standard deviations, minimum and maximum values for the variables. The average CEO narcissism score is -0.005, with a standard deviation of 0.400. The minimum and maximum value of the variable are -0.683 and 2.736, respectively. Internationalization has an average foreign sales to total sales ratio of 50,577% and a large standard deviation of 37.726. The minimum is a ratio of 0% and the maximum is a ratio of 100%, which means no sales cross-border and all sales cross-border, respectively. The financial reports disclose an average number of abnormal risks of 4.246, with a standard deviation of 2.252. The minimum number of abnormal risks disclosed is -1 and the maximum number of abnormal risks disclosed is 19. The average natural logarithm of total assets, which is the proxy for firm size, is 6.081. The average market-to-book ratio is 3.038 and the average return on assets is 6.704, which are the proxies for financial leverage and firm performance, respectively. The insiders hold an average of 20.655% of the shares. The maximum shares hold by insiders is 91.060%. Regarding the CEO characteristics, the average age of the CEO is 52.066, with a minimum age of 31 and a maximum age of 76. The average tenure of the CEO is 6.153. The CEO’s have an average network size of 676.318 with a large standard deviation of 846.270. Last, the CEO has on average 1.702 qualifications, with a minimum of 0 qualifications and a maximum of 6 qualifications.

20

Table 4 presents the Pearson correlation coefficients between CEO narcissism, internationalization, financial reporting risk and the control variables. From the table it can be perceived that internationalization is positively correlated with CEO narcissism (0.0916). This is in line with the previous research of Oesterle et al. (2016). In addition, financial reporting risk is also positively correlated with CEO narcissism (0.1330) and internationalization (0.2616). A correlation between variables of more than 0.700 or less than -0.700 was considered as multicollinearity. As shown in Table 4, none of the values exceeds 0.700 or is less than -0700, so there is no high level of multicollinearity between the variables in this research.

In addition, I performed a variance inflation factor (VIF) test to check whether high correlation coefficients lead to multicollinearity. This measure is a more precise way to measure multicollinearity. A VIF value of 10 or higher indicates that multicollinearity forms a problem. The VIF was stated below each regression model. Looking at the highest VIF-values, the VIF is 1.24 in the OLS regression and 1.26 in the Propensity Score Matching. From this viewpoint, this means that the outcomes of the analyses were reliable and not affected by multicollinearity between variables.

Variable (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)

(1) CEO Narcissism 1.0000

(2) Internationalization 0.0916 1.0000 0.0001

(3) Financial Reporting Risk 0.1330 0.2616 1.0000 0.0000 0.0000 (4) Firm Size 0.0852 0.2022 0.2552 1.0000 0.0002 0.0000 0.0000 (5) Financial Leverage -0,0454 -0.0109 -0.1052 -0.0000 1.0000 0,0514 0.6414 0.0000 0.9983 (6) Firm Performance -0.0241 -0.0131 -0.0934 -0.0231 0.2068 1.0000 0.3012 0.5728 0.0001 0.3224 0.0000 (7) CEO Age -0.0802 0.1154 -0.0378 0.0665 -0.0504 -0.0430 1.0000 0.0006 0.0000 0.1048 0.0043 0.0307 0.0651 (8) CEO Tenure -0.2316 0.0174 -0.0403 -0.1096 0.0147 0.1075 0.3316 1.0000 0.0000 0.4546 0.0835 0.0000 0.5283 0.0000 0.0000

(9) CEO Network Size 0.0595 0.0837 0.0516 0.2356 0.0177 -0.0052 -0.0933 -0.1422 1.0000 0.0107 0.0003 0.0269 0.0000 0.4468 0.8219 0.0001 0.0000

(10) CEO Qualifications 0.1348 0.1556 0.0673 0.1053 -0.0058 -0.0334 0.0231 0.0130 0.3479 1.0000 0.0000 0.0000 0.0039 0.0000 0.8043 0.1513 0.3209 0.5766 0.0000

(11) Closely Held Shares -0.1205 -0.1263 -0.0084 -0.1274 -0.0363 -0.0656 -0.0768 0.0561 -0.1421 -0.0852 1.0000 0.0000 0.0000 0.7173 0.0000 0.1192 0.0049 0.0010 0.0161 0.0000 0.0002

*** p<0.01, ** p<0.05, * p<0.1 TABLE 4 Pearson Correlations

21

Main Analysis

Structural Equation Modeling is used to test the hypotheses of this paper. Before the analyses are executed, all continuous variables in the data are winsorized at the top and bottom 1 percent levels to mitigate the influence of outliers. In addition, the data is standardized.

Structural Equation Modeling is used because it simplifies testing since it is designed to test complicated mediation models in a single analysis (MacKinnon, 2008). In addition, standard regression implies a statistical relationship, while Structural Equation Modeling implies a functional relationship, which means that the indirect and direct effects are more appropriately expressed using Structural Equation Modeling instead of OLS regression (Gunzler et al., 2013). In addition, the research of Danner et al. (2015) concluded that Structural Equation Modeling can improve scientific insights.

Structural Equation Modeling is executed to answer the main question: What is the mediating

effect of internationalization on the relationship between CEO narcissism and financial reporting risk?. The direct effects of these tests are presented in Table 5. As shown in the

table, for this analysis there are 1842 observations used from listed companies in the Netherlands and United Kingdom. In addition to the firm-level and CEO-level control variables, the year-fixed effects, industry-fixed effects and country-fixed effects are added to the analysis. Due to data limitations of the compensation data, this measure is not used in the main analysis. The CEO narcissism score consists of the picture score and the first-person singular pronouns.

Model 1 shows the direct effect of CEO narcissism on internationalization. As presented, CEO narcissism has a positive sign and is significant at 10 percent, with β1 = 0.045 and

z-value = 1.96. Model 2 shows the effect of CEO narcissism and internationalization on the company’s financial reporting risk. CEO narcissism is positive and has a significance level of 1 percent, with β1 = 0.092 and z-value = 3.89. Internationalization is significant at 1 percent

and has a positive sign, with β2 = 0.216 and z-value = 9.01.

Based on these results, hypothesis 1 is accepted: CEO narcissism will increase the company’s financial reporting risk. Hypothesis 2 describes the positive relationship between CEO narcissism and internationalization. As shown in Model 1, this hypothesis is supported. In addition, hypothesis 3 is supported. When a company performs in international activities, it

22

has a positive effect on company’s financial reporting risk. In hypothesis 4, the mediating effect of internationalization on the relationship between CEO narcissism and financial reporting risk is predicted. The results reveal that internationalization mediates the relationship between CEO narcissism and financial reporting risk. So, hypothesis 4 is accepted as well.

(1) (2)

Variable Model 1 Model 2

Dependent variable Internationalization Financial Reporting Risk

Constant -0,142 -0,110 (0.098) (0.101) Test variables CEO Narcissism 0.046** 0.096*** (0.023) (0.024) Internationalization 0.216*** (0.024) Control variables Firm Size 0.154*** 0.249*** (0.031) (0.032) Financial Leverage 0.004 -0.271*** (0.071) (0.073) Firm Performance 0.029 -0.060** (0.029) (0.030)

Closely Held Shares -0.094*** -0.017

(0.022) (0.023)

CEO Age 0.058*** -0.113***

(0.022) (0.022)

CEO Tenure 0.035 0,036

(0.022) (0.023)

CEO Network Size 0.056*** -0,008

(0.021) (0.022)

CEO Qualifications 0.073*** 0,004

(0.021) (0.022)

CEO Gender Yes Yes

Year-fixed effects Yes Yes

Industry-fixed effects Yes Yes

Country-fixed effects Yes Yes

Observations 1842 1842

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

TABLE 5

23

Robustness Checks

In this section, three analyses are performed to check if the results in the main analysis are robust. First, I use the variables from the main analysis to perform OLS regression analysis instead of Structural Equation Modeling. Second, I perform an analysis with another proxy for CEO narcissism. Third, I use propensity score matching (PSM) to test the endogeneity of CEO narcissism.

Regression Analysis

Table 6 presents the coefficient estimates, robust standard errors and the significance level of the regression analyses for the four hypotheses. As shown, again 1842 observations are used in these analyses. The R-squared varies between 16.5% and 31.7%. The results are clustered by the International Securities Identification Number (ISIN) of a company.

Model 1. The OLS regression in Model 1 is performed for the independent variable

“CEO narcissism” and the dependent variable “financial reporting risk”. As shown, CEO narcissism has a positive sign and is significant at 5 percent, with β1 = 0.106 and t-value =

2.46, which is consistent with hypothesis 1 that CEO narcissism will increase the company’s financial reporting risk. This result is consistent with the result in the main analysis.

Model 2. The OLS regression in Model 2 is performed to test the relationship between

the independent variable “CEO narcissism” and the mediator “internationalization”. The hypothesis predicted that CEO narcissism would increase the internationalization of a company. As shown in Table 6, this hypothesis is not supported. Internationalization has a positive sign and has no significance level, with β1 = 0.046 and t-value = 0.99. This is not

consistent with the results in the main analysis, where this relationship was significant.

Model 3. The OLS regression in Model 3 is executed to test the relationship between

the mediator “internationalization” and the dependent variable “financial reporting risk”. As shown in Table 6, internationalization has a positive sign and is significant at 1 percent, with β1 = 0.250 and t-value = 5.16. These results are consistent with hypothesis 3, which predicted

that internationalization would increase the company’s financial reporting risk. This result is consistent with the result in the main analysis.

24

(1) (2) (3) (4)

Variable Model 1 Model 2 Model 3 Model 4

Dependent variable Financial Reporting Risk Internationalization Financial Reporting Risk Financial Reporting Risk

Constant -0.112 -0.897** 0.089 0.240 (0.345) (0.374) (0.312) (0.217) Test variables CEO Narcissism 0.106** 0.046 0.096** (0.043) (0.047) (0.042) Internationalization 0.220*** 0.216*** (0.043) (0.043) Control variables Firm Size 0.283*** 0.154*** 0.250*** 0.249*** (0.064) (0.056) (0.061) (0.059) Financial Leverage -0.270*** 0.004 -0.285*** -0.271*** (0.104) (0.116) (0.105) (0.104) Firm Performance -0.054 0.029 -0.058 -0.060 (0.050) (0.058) (0.047) (0.048)

Closely Held Shares -0.037 -0.094** -0.024 -0.017

(0.051) (0.044) (0.051) (0.051)

CEO Age -0.101** 0.058 -0.115** -0.113**

(0.048) (0.051) (0.046) (0.045)

CEO Tenure 0.044 0.035 0.015 0.036

(0.042) (0.040) (0.042) (0.042)

CEO Network Size 0.004 0.056 -0.012 -0.008

(0.039) (0.043) (0.041) (0.040)

CEO Qualifications 0.020 0.073 0.014 0.004

(0.043) (0.048) (0.042) (0.042)

CEO Gender Yes Yes Yes Yes

Year-fixed effects Yes Yes Yes Yes

Industry-fixed effects Yes Yes Yes Yes

Country-fixed effects Yes Yes Yes Yes

Observations 1842 1842 1842 1842

R-squared 0.165 0,317 0.193 0.200

Highest VIF 1.24 1.24 1.24 1.24

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

TABLE 6 Regression Analysis

Model 4. The OLS regression in Model 4 is performed to test the relationship between

the dependent variable “financial reporting risk”, the mediator “internationalization” and the independent variable “CEO narcissism”. As shown in Table 6, CEO narcissism has a positive sign and is significant at 5 percent, with β1 = 0.096 and t-value = 2.30. Internationalization

also has a positive sign and is significant at 1 percent, with β2 = 0.216 and t-value = 5.07. The

coefficient of narcissism is lower in Model 4 than in Model 1. Therefore hypothesis 4, internationalization mediates the relationship between CEO narcissism and financial reporting risk, is supported. This is consistent with the results from the main analysis.

25

Another Proxy for CEO Narcissism

Chatterjee and Hambrick (2007) used a proxy for the CEO narcissism score that contains the prominence of the CEO’s photograph, the prominence in press releases, the use of first-person singular pronouns and the CEO’s relative cash and non-cash compensation. Due to data limitations, I excluded the CEO’s relative cash and non-cash compensations in the main analysis. In this section of the study, I perform an additional test with the CEO narcissism score including relative cash and non-cash compensation to validate the results of the main analysis.

The relative pay is used because CEOs have considerable domination in the establishment of their own compensation (Bebchuk and Fried, 2004). In addition, they have total control over the pay of the other executives working in their company (Chatterjee and Hambrick, 2007). Following their research, there are two measures used of the relative compensation. The relative cash pay was measured through the CEO’s cash compensation (salary and dividend) divided by that of the second-highest-paid executive in the company. The relative noncash pay (deferred income, stock options, stock grants) was measured in the same way.

Table 7 presents the results of the Structural Equation Modeling that is performed in the main analysis, but with CEO narcissism including relative cash and non-cash compensation. As shown, 1182 observations are used, compared with 1842 in the main analysis. Model 1 shows the direct effect of CEO narcissism on internationalization. As presented, CEO narcissism has a negative sign and no significance. These results are not the same as in the main analysis, where it was positive and has a significance level of 5 percent. Model 2 shows the effect of CEO narcissism and internationalization on the company’s financial reporting risk. CEO narcissism is positive and has a significance level of 1 percent in the main analysis and is positive and has a significance level of 5 percent in this analysis. Internationalization is significant at 1 percent and has a positive sign in the main analysis. As shown in Table 7, these results are the same in this analysis.

In sum, the results of the main analysis are not robust with the CEO narcissism score including relative cash and non-cash compensation.

26

(1) (2)

Variable Model 1 Model 2

Dependent variable Internationalization Financial Reporting Risk

Constant -0.004 0.001 (0.117) (0.122) Test variables CEO Narcissism -0.026 0.069** (0.029) (0.030) Internationalization 0.259*** (0.030) Control variables Firm Size 0.147*** 0.211*** (0.034) (0.036) Financial Leverage 0.054 -0.339*** (0.085) (0.089) Firm Performance 0.023 -0.054 (0.036) (0.037)

Closely Held Shares -0.032 -0.071**

(0.029) (0.030)

CEO Age 0.066*** -0.113***

(0.026) (0.027)

CEO Tenure 0.016 -0.012

(0.025) (0.026)

CEO Network Size 0.097*** -0.026

(0.027) (0.028)

CEO Qualifications 0.055** 0.023

(0.024) (0.025)

CEO Gender Yes Yes

Year-fixed effects Yes Yes

Industry-fixed effects Yes Yes

Country-fixed effects Yes Yes

Observations 1182 1182

*** p<0.01, ** p<0.05, * p<0.1

TABLE 7

Structural Equation Modeling

27

Propensity Score Matching

For the last robustness check, propensity score matching is used to test the endogeneity of CEO narcissism. First, the propensity scores are estimated in a logit model, where the dependent variable is “high_narcissism” and the explanatory variables are the control variables. “High_narcissism” is based on the conditional probability of a CEO being highly narcissistic. I define a CEO as highly narcissistic when the CEO narcissism score exceeds the median of -0.0126. Each highly narcissistic CEO and not highly narcissistic CEO are matched with the closest propensity score from the sample. Using the propensity-matched sample, the regressions are repeated.

Table 6 presents the coefficient estimates, robust standard errors and the significance level of the Propensity Score Matching for the four hypotheses. As shown, 1022 observations are used in these analyses. The R-squared varies between 15.7% and 34.8%. The results are clustered by the International Securities Identification Number (ISIN) of a company.

Model 1 indicates that the financial reporting risk increases when the CEO is narcissistic. This result is significant at 5 percent, with β1 = 0.135 and t-value = 2.44, which is consistent with

the result in the main analysis. Model 2 indicates that the internationalization of a company increases when they have a narcissistic CEO. This result is not significant, with β1 = 0.018

and t-value = 0.29. This is not consistent with the results in the main analysis, however it is consistent with the results in the other robustness checks. Model 3 indicates that financial reporting risks are increasing when the internationalization is high. This result is significant at 1 percent, with β1 = 0.232 and t-value = 4.23, which is consistent with the results in the main

analysis. Finally, Model 4 contains the relation between CEO narcissism, internationalization and financial reporting risk. CEO narcissism is significant at 5 percent, with β1 = 0.131 and

t-value = 2.47, which has a significance of 1 percent in the main analysis. Internationalization is significant at 1 percent, with β2 = 0.230 and t-value = 4.21. This result is the same as in the

main analysis. In addition, the coefficient of CEO narcissism is lower in Model 4 than in Model 1, which indicates that hypothesis 4 is accepted.

28

(1) (2) (3) (4)

Variable Model 1 Model 2 Model 3 Model 4

Dependent variable Financial Reporting Risk Internationalization Financial Reporting Risk Financial Reporting Risk

Constant -0.477 -0.616 -0.334 -0.335 (0.657) (0.501) (0.673) (0.593) Test variables CEO Narcissism 0.135** 0.018 0.131** (0.056) (0.061) (0.053) Internationalization 0.232*** 0.230*** (0.055) (0.055) Control variables Firm Size 0.247*** 0.133 0.214*** 0.216*** (0.082) (0.081) (0.077) (0.070) Financial Leverage -0.332*** -0.069 -0.321*** -0.316*** (0.120) (0.118) (0.121) (0.118) Firm Performance -0.092 0.046 -0.100 -0.103 (0.072) (0.076) (0.068) (0.069)

Closely Held Shares -0.080 -0.070 -0.076 -0.063

(0.061) (0.056) (0.063) (0.062)

CEO Age -0.118** 0.070 -0.135** -0.134**

(0.056) (0.064) (0.054) (0.053)

CEO Tenure 0.056 -0.035 0.058 0.064

(0.057) (0.058) (0.061) (0.059)

CEO Network Size 0.030 0.084* 0.010 0.011

(0.043) (0.045) (0.047) (0.044)

CEO Qualifications -0.018 0.060 -0.031 -0.032

(0.055) (0.050) (0.055) (0.054)

CEO Gender Yes Yes Yes Yes

Year-fixed effects Yes Yes Yes Yes

Industry-fixed effects Yes Yes Yes Yes

Country-fixed effects Yes Yes Yes Yes

Observations 1022 1022 1022 1022

R-squared 0.157 0.348 0.184 0.197

Highest VIF 1.26 1.26 1.26 1.26

*** p<0.01, ** p<0.05, * p<0.1

TABLE 8 Propensity Score Matching

29

Discussion and Conclusions

In this chapter, the findings concerning the results are explained. Based on these findings, the implications are explained. This chapter will finish with the limitations and some ideas for future research.

Conclusions

This study examines two factors that influence the number of financial reporting risks in financial reports. First, executives have many reasons to conduct dishonest or unethical behaviour in company’s financial reports. Duchon and Drake (2009) suggest that extreme narcissistic CEOs are more often associated with unethical business practices because they lack a moral identity. Therefore, this study investigates the influence of CEO narcissism on financial reporting risk. Second, Glassman (2006), commissioner for the SEC, expressed concerns about the complexity of international accounting standards, which can lead to more financial reporting risks. Another danger that can arise from internationalization activities is the foreign currency risk (Bartram and Bodnar, 2007), which can result in more financial reporting risks. Therefore, this study examines the influence of internationalization. In summary, to extend the understanding effects of financial reporting risk, this research examines the effect of CEO narcissism and internationalization on financial reporting risk. However, De Vries and Miller (1985) concluded that narcissistic CEOs often opt for excellent, spectacular and risky projects, which can lead to overconfidence in international acquisitions. In addition, Oesterle et al. (2016) concluded that the more narcissistic a CEO is, the higher the growth of a company’s international activities. To extend on this, I have also examined the mediating role of internationalization on the relationship of CEO narcissism and financial reporting risk.

The first research question that I have answered is about the relation between CEO narcissism and financial reporting risk. This study presents a significant positive effect of CEO narcissism on financial reporting risk. This finding is consistent with previous research, which concluded that narcissists take more risks than other people. Reason for this can be the attraction of admiration and attention (Zhu and Chen, 2015). According to Duchon and Drake (2009), another reason for this can be that narcissists become self-obsessed and use sense of entitlement, denial and rationalizations to justify decisions they make.

30

Second, the relation between CEO narcissism and internationalization is examined. The study shows a significant positive effect of CEO narcissism on internationalization, which means that the more narcissistic the CEO is, the more a company performs in international activities. This result is consistent with the expectations of the study and previous findings, such as Malmendier and Tate’s study (2008), which find that narcissistic CEOs often overestimate their ability to generate returns from a cross-border project. This resulted in the overpaying for target companies and the undertaking of value-destroying mergers.

Third, I examined the relation between internationalization and financial reporting risk. The findings indicate a significant positive effect of internationalization on financial reporting risk, which means the higher the degree of a company’s international activities, the higher the number of financial reporting risks in financial reports. This may be due to the use of more complex financial reporting, which is necessary when performing in international activities. In addition, the increase of a firm’s sales into foreign countries will introduce accounts in the financial reports that are translated from foreign currencies (Lee and Jang, 2010), which lead to more financial reporting risks and affects the company’s value.

Last, I answer the main research question: “What is the mediator effect of internationalization

on the relationship between CEO narcissism and financial reporting risk?”. The results show

a mediating effect of internationalization on the relationship between CEO narcissism and financial reporting risk. This effect is made robust with propensity score matching, which has shown the same results.

Implications

This study contributes to the existing literature and provides new, valuable insights. By investigating the influence of CEO narcissism and internationalization on financial reporting risk, this study filled a gap within the literature. This study was the first that tested the mediating effect of internationalization on the relationship between CEO narcissism and financial reporting risk. It appears that the mediating effect exists, what could impact the material risks in a company’s financial report. The study has implications for investors, auditors and other stakeholders. Auditors should emphasize the impact of CEO narcissism and internationalization on the company’s financial reports. For international stakeholders, the findings of this study might be relevant as well. They need to understand the impact of internationalization on the financial reporting statements they receive.

31

Limitations and Future Research

While the findings, for the biggest part, show significant and positive relationships between the main variables, this study has some limitations to consider. First, in all three robustness tests, the direct relation between CEO narcissism and internationalization shows no significance. Second, the sample consists of large listed companies from the Netherlands and United Kingdom, which limits the generalizability to private companies and smaller companies. Third, the hand-collected data on CEO narcissism and financial reporting risks is gathered by a group of people who conducted research in the same area, which can lead to different interpretations. Fourth, I only consider the sales-dimension as a proxy for internationalization and did not use other proxies, for example cross-border acquisition. These limitations suggest in turn opportunities for future research. First, if the relations can be identified for other countries or another set of companies, this might verify and extent the insights of this study. In addition, a greater sample with enough compensation data for the CEO narcissism score will develop this study. Second, using another proxy for internationalization can extend the research. For example, the number of operating and geographic segments can be used. It can be argued that the likelihood of financial reporting risks is higher for companies with greater operational complexity. Last, the study can be verified if the data was collected by one person, or by making clear decision rules for the data collectors.