The Crystal as a
Mirror for Powerful
Narcissists
The Effects of CEO
Narcissism and Power on
CSR Reporting
abstract
Recent research highlights the importance of finding corporate social responsibility (CSR) disclosure determinants. This study therefore empirically investigates the direct influence of Chief Executive Officer (CEO) narcissism and the moderating influence of CEO power on CSR disclosure. Based on impression management and upper echelon theory, narcissistic CEOs are expected to reinforce their inflated self-views by engaging in CSR disclosure, as this can bring social praise and attention. This positive association can be strengthened by the amount of power the CEO possesses, since the approach/inhibition theory states that powerful individuals will exert more trait-consistent behaviour, having more available economic and social resources and being unconstrained by others. After examining 223 of the largest Dutch organisations, participating in the 2016 CSR Transparency Benchmark (TB), the results support the prediction that CEO narcissism positively influences CSR disclosure, whereas the positive moderating influence of CEO power is not supported. These findings provide theoretical insights about CSR disclosure determinants and further enrich the measurement of CEO narcissism. Practically, the findings provide insights for the decision making of various organisational stakeholders, as CEO narcissism proves to have an impact on organisational decisions and outcomes.
MASTER THESIS
University of Groningen
Faculty of Economics and Business MSc Accountancy
Name: Milan van Rees Student number: s2479850
Address: Zuiderpark 1002, 9724 AK, Groningen Phone Number: 06-20451995
E-mail: m.van.rees@student.rug.nl Supervisor: prof. dr. D.A. de Waard Second assessor: drs. L.M. Wielens Word Count: 12.308
1 Introduction 5 2 Scientific contribution 7 3 Theoretical framing 10
3.1 Impression management theory 10 3.2 Upper echelon theory 11 3.3 Approach/inhibition theory 12
4 Hypotheses generation 14
4.1 CEO narcissism 14
4.2 CEO power 16
5 Method 18
5.1 Measurement of CEO narcissism 19 5.2 Measurement of CEO power 21 5.3 Measurement of CSR disclosure 22 5.4 Control variables 23 5.5 Statistical model 24
6 Results 25
6.1 Measurement of CEO narcissism 25 6.2 Measurement of CEO power 26 6.3 Descriptive statistics 27 6.4 Correlation analysis 27 6.5 Regression analysis 28 6.6 Robustness checks 29 6.7 Endogeneity 31 7 Conclusion 33 7.1 Findings 33 7.2 Theoretical implications 34 7.3 Practical implications 35 7.4 Limitations and future research 35
8 References 37
Appendix A 42
1. INTRODUCTION
Recently, research showed that the percentage of students exhibiting narcissistic personality traits has increased by more than a half since the 1980s (Twenge & Forster, 2010). This personality trait named after Narcissus, a mortal in the Greek mythology that fell in love with its own reflection, is therefore presumably a commonplace character trait for tomorrow’s leaders (New York Times, 2016). Even today’s managers are more narcissistic than the average individual (Financial Daily, 2008). This is not remarkable, as managers should be convinced of their own qualities to aspire an executive position. It is therefore valuable to investigate this character trait and the business impact it has, which may be especially relevant for the individual possessing the highest executive position in the organisation, the CEO.
Within the business world, CSR reporting has been emerging over the past several years and it therefore seems destined to become a key part of the overall accounting reporting framework (Tschopp & Heufner, 2015). Whereas disclosure of financial information is highly regulated, the disclosure of CSR information is largely voluntarily (De Villiers & Marques, 2016). According to Moser & Martin (2012), CSR disclosure can both show conformance to societal expectations and provide market participants with additional information. Consequently, CSR disclosure may create positive economic results, by decreasing the cost of equity capital and increasing the forecast accuracy of analysts (Dhaliwal et al., 2011; Dhaliwal et al., 2012). To report on CSR activities, companies can utilize various frameworks, such as the Sustainability Accounting Standards Board (SASB), Global Reporting Initiative (GRI) and the International Integrated Reporting Council (IIRC) (Weber et al., 2016).
Subsequently, the performance of organisations that report on CSR can be ranked. International examples of these CSR rankings are: CRO magazine’s 100 Best Corporate Citizens, Ethisphere’s Most Ethical Companies and Corporate Knight’s Global 100 (Huang, 2013). Next to CSR performance rankings, organisations can also be compared on their disclosure of CSR related activities. In the Netherlands, due to the Transparency Benchmark (TB) of the Ministry of Economic Affairs, the CSR activities of the largest Dutch companies are ranked annually (Ministry of Economic Affairs, 2016). The TB ranks the organisational reporting on CSR by assessing disclosures about economic, environmental and social value creation, in the short-, medium- and long-term. Ranking and benchmarking the disclosure of CSR activities may be even more valuable for stakeholders than absolute CSR performance, since transparency on CSR activities eases the dialogue with shareholders on relevant topics,
leading to better decision making (Ministry of Economic Affairs, 2016). The TB is intended to provide transparency on CSR reporting by assessing the content and quality of CSR reports. The participating companies in the TB contend for the Crystal Prize, which is a leading award in the area of social reporting in The Netherlands (Ministry of Economic Affairs, 2016). The obtained scores by the TB contenders however differ significantly, and it is therefore relevant to investigate possible antecedents of these differences. One of these antecedents may be the level of CEO narcissism, as narcissists may use CSR disclosure to endorse their inflated self-views by creating social praise and attention (Vazire & Funder, 2006). According to findings of Michelon et al. (2014), CSR reporting can be used substantively or symbolically. Using the substantive approach, organisations are genuinely interested in transparency due to their sense of accountability to stakeholders. On the other hand, using the symbolic approach, CSR reporting is used to create an inaccurate image to influence perceptions of stakeholders. Narcissistic CEOs may have a great interest in influencing the perceptions of stakeholders, as stakeholders can be considered as suppliers of the CEO’s narcissistic needs (Wallace and Baumeister, 2002). These arguments are in line with impression management and upper echelon theory, since the impression management theory states that narcissistic individuals are relatively more motivated to focus on CSR disclosure and upper echelon theory assumes that organisational outcomes, such as CSR disclosure, are a reflection of the strategic choices of managers and these choices are in turn a result of their (narcissistic) characteristics (Leary & Kowalski, 1990; Hambrick & Mason, 1984)
Concerning the relationship between CEO narcissism and CSR disclosure, CEOs may increase their efforts to influence stakeholder perceptions when their degree of power within the organisation grows, since they can opportunistically use their power to achieve social praise and build reputation (Li et al, 2016). Building on approach/inhibition theory, powerful individuals have the resources to act in their own interests and are not constrained by the interference of others, causing them to live more by their (narcissistic) traits (Keltner et al., 2003).
CEO narcissism and power may therefore be antecedents of the dissimilarity in TB scores between contending companies. Hence, the following research question is discussed in this paper:
What is the influence of CEO narcissism and power on CSR disclosure, as indicated by TB scores?
In the next section, the scientific contribution of this research question will be discussed. Subsequently, relevant theories concerning the research question will be presented. This theoretical framing section focuses on both the impression management, upper echelon and approach/inhibition theory. Then, the hypotheses generation will be discussed, which will make specific predictions about the direct influence of CEO narcissism and the moderating influence of CEO power on CSR disclosure. Thereafter, an examination of the methodology and the associated data collection will be presented. Making use of the described methodology, the results will be showed in the following section. Lastly, the conclusion will discuss the main findings, limitations and suggestions for future research.
2. SCIENTIFIC CONTRIBUTION
Previous research has been done on the relationship between CEO narcissism and various corporate activities. For example, Chatterjee & Hambrick (2007) have investigated the relationship between CEO narcissism and corporate actions. They concluded that narcissistic CEOs engage in bold and highly visible choices, whereas less narcissistic CEOs engage in more incremental and less visible choices. Petrenko et al. (2016) found that CEO narcissism is positively correlated with CSR, as CSR gives the CEO a platform for attention and for reinforcement of their positive self-views. Within this context, the value loaded character of CSR activities can offer exhibitionism opportunities. Moreover, CSR engages in sources of narcissistic supply, such as media attention and praise (Wallace and Baumeister, 2002). Narcissistic CEOs are also unwilling to take socially irresponsible actions, as that can bring easy criticism from investors, media, the public or employees (Petrenko et al., 2016). Research of Gerstner et al. (2013) suggests that narcissistic CEOs will invest heavily in discontinuous technology, especially when these investments attract attention and admiration from external actors. Research of Oesterle et al. (2016) reveals that narcissistic CEOs are more willing to expand their company internationally, as they want to achieve excessive admiration and feel an intense need for power and prestige. The risks that go along with leading internationally operating organisations are probably underestimated by narcissistic CEOs, as they believe to overcome these risks by using their own talents.
Previous research has therefore provided valuable insights about the effects of CEO narcissism on corporate activities, such as organisational decision making (Chatterjee & Hambrick, 2007), CSR engagement (Petrenko et al., 2016), technological investments (Gerstner et al., 2013) and internationalisation (Oesterle et al., 2016). However, the literature does not provide empirical evidence about the relationship between CEO narcissism and the disclosure of corporate social activities. This is remarkable, as research shows that CSR reporting is highly valuable, by signalling future financial performance to investors and communicating private information about the organisation (Chen et al., 2016), presumably lowering the cost of equity capital (Dhaliwal et al. 2011). Furthermore, corporate social behaviour has become an important aspect of business society, because of increased attention by press and society (Hooghiemstra, 2000). Consistently, Huang (2013) has proposed a further examination of CEO specific characteristics as antecedents of CSR and the disclosure hereof in future research. It is therefore relevant to bridge this gap in the literature by conducting a research on the influence CEO narcissism has on CSR disclosure.
This study also enriches the measurement of the CEO narcissism concept by adding two new unobtrusive measures to the Chatterjee & Hambrick (2007) framework. The first added measure is the length of the CEO’s letter to shareholders. Research states that this letter is one of the major external communication channels of the CEO, which can be used for impression management goals (Hooghiemstra, 2000), influencing the external perceptions about the qualities of the narcissistic CEO. To positively shape the impression of readers, improving external perceptions about the CEO’s leadership and his/her reputation, managers may need more words to deliver this massage (Patelli & Pedrini, 2014). Secondly, the CEO’s LinkedIn prominence is added. This large online professional network creates a platform for narcissists to achieve self-presentational goals (Guillory & Hancock, 2012), causing them to construct a ‘display window’ out of their public profile.
Regarding CEO specific characteristics, the role of CEO power may be relevant as well. Before, CEO power has been mostly observed in relationship with firm performance (e.g. Han et al., 2016; Lee et al., 2015; Bebchuk et al., 2011). The results mainly conclude that higher levels of CEO power go along with more opportunistic and rent-seeking behaviour. Furthermore, higher levels of CEO power tend to cohere with more discretion in strategic decision making (Han et al., 2016). It is therefore relevant to investigate the relationship between CEO power and the strategic choices the CEO takes. Hence, the area of CSR disclosure, as part of strategic decision making of CEOs, is an unexploited topic of research. Within this context, the relatively new approach/inhibition theory, derived from the social psychology, may generate useful insights in other disciplines, such as the business world (Keltner et al., 2003). Besides, this thesis also complies with the call of Bebchuk et al. (2011) to investigate the relationship between CEO power and decision making.
In addition Giannarakis et al. (2014) state that more research is needed for the investigation of CSR disclosure determinants. The majority of research has previously focused on external factors that influence CSR, whereas fewer analysed the role of internal factors at the individual level (Petrenko et al., 2016). Examples of previous research are the examination of the role of institutional determinants, such as the corporate, social and legal system (Garcia-Sanchez et al., 2016), the role of external stakeholders, such as media pressure (Garcia-Sanchez et al. 2014) and governance mechanisms, such as independence of board members (Fuente et al., 2017). These perspectives may suggest that companies facing similar institutional, external and governance circumstances may respond similarly in terms of CSR disclosure, whereas evidence concludes that organisational responses are heterogeneous (Walls & Beronne, 2016). This may however be due to the underexposed role of managerial personal characteristics, such as CEO narcissism and power. Thus, this research extends the knowledge about CSR disclosure determinants.
Next to the development of theoretical knowledge, this research also contains practical relevance for the business world. Firstly, the board of directors could exploit the knowledge of the interaction between CEO narcissism, power and CSR disclosure in their CEO appointment decision. Fitzsimmons & Callan (2016) state that the selection and appointment of CEOs by board is increasingly based on leadership characteristics that cause ethical outcomes. Consequently, boards that focus on attaching value to CSR disclosure may consider the level of narcissism of CEOs. Next, shareholders take CSR disclosure into account when making investment decisions, since CSR activities impact the economic (future) performance of organisations by creating probable cost savings, revenue growth or risk reduction, which
can be indicated by disclosure of CSR activities (Bliss, 2015). Using this study’s results, shareholders may be able to improve their evaluation of the motives for CSR disclosure, and hence make better investment decisions. Lastly, governmental bodies, such as the Dutch Ministry of Economic Affairs, are interested in the antecedents of CSR policies and activities of organisations within their sphere of influence, because this may lead to facilitation of dialogues between stakeholders and improved business performance (Ministry of Economic Affairs, 2016a).
3. THEORETICAL FRAMING
In this section, the upper echelon, impression management and approach/inhibition theory are discussed to develop insights about the relationship between narcissism, power and CSR disclosure.
3.1 IMPRESSION MANAGEMENT THEORY
The impression management theory, derived from social psychology, is concerned with how individuals present themselves to others to be perceived favourably by others (Hooghiemstra, 2000). Within this view, corporate communication aims at the protection or enhancement of image or reputation. Managers are therefore more willing to report ‘good news’, and reluctant to disclose ‘bad news’ (Hooghiemstra, 2000). Leary & Kowalski (1990) argue that impression management involves two distinct processes, which are ‘impression motivation’ and ‘impression construction’.
Firstly ‘impression motivation’ deals with the conditions under which people are motivated to manage public impressions (Leary & Kowalski, 1990). Sometimes, people withhold from any action to create a certain impression by others, whereas under other circumstances, people are motivated to conduct actions that create impressions in the minds of others. To determine the motivation of an individual, the impression’s goal-relevance and desired value of these goals should be determined in order to assess the motivation of individuals to engage in impression management.
Next, ‘impression construction’ deals with the precise way of conducting the impression management tactic (Leary & Kowalski, 1990). There are various mechanisms to execute impression management, like self-description and non-verbal behaviour, which may have various consequences on the targeted audience. To determine the precise construction of impression management, an individual considers its current or potential social image, its desired and undesired identity image and its self-concept.
Within the impression management theory, corporate communications, such as CSR reporting, are being used as ‘impression construction’ instruments to influence people’s perceptions. Narcissistic CEOs may be motivated to engage in ‘good news’ CSR disclosure (Hooghiemstra, 2000), as this may positively influence their image or reputation, creating a
form of narcissistic supply (Petrenko et al., 2016). Moreover, Li et al. (2016) state that CSR related initiatives mostly serve the selfish management’s image, instead of serving shareholder’s preferences. Thus, a narcissistic personality has, in terms of Leary & Kowalski (1990), various inherent ‘impression motivations’ to engage in positive CSR disclosure.
Next to the focus on generation of positive CSR perceptions, narcissistic CEOs are strongly reluctant to disclose negative CSR information, as this can bring criticism from various stakeholders, shaping their image in an undesirable way (Petrenko et al., 2016). In specific, negative CSR may lead to scandals, discrimination or lawsuits, which may create extreme negative emotional effects for narcissists (Rhodewalt & Morf, 1998). This is due to the asymmetry in reactions of narcissists on positive and negative outcomes as opposed to non-narcissistic individuals, as success is relatively more positively interpreted and failure is relatively more negatively interpreted. Building on these findings, impression management can play a vital role in the decision making about the extent of CSR disclosure by managers, by means of the fostering of positive CSR disclosure and the avoidance of negative CSR disclosure (Hooghiemstra et al., 2000).
All in all, narcissistic CEOs will presumably be more focused on CSR disclosure than their less narcissistic counterparts, due to their inherent ‘impression motivations’ and their tendency to give a positive substance to their ‘impression construction’. Narcissistic CEOs are therefore more willing to spend organisational resources on CSR disclosure, as they acknowledge that corporate social behaviour is commonly attributed to themselves (Petrenko et al., 2016). The increased spending of resources and focus on CSR disclosure may consequently lead to a higher ranking in the TB.
3.2 UPPER ECHELON THEORY
Moreover, the influence of CEO narcissism on strategic decision making is often observed from the upper echelons theory (e.g. Oesterle et al., 2016). This theory states that the constrained capacity of managers and the situation’s complexity causes a limited perception of situations (Oesterle et al., 2016). First, the cognitive base and values of managers, described as physiological upper echelon characteristics, create a screen between the situation and the perception of the situation by the manager (Hambrick & Mason, 1984). Next, also observable upper echelon characteristics, such as age, education and tenure, influence the manager’s perception.
Combining physiological and observable upper echelon characteristics, strategic actions, such as acquisitions and innovations, can be predicted (Hambrick & Mason, 1984), consequently influencing organisational performance. Thus, organisational outcomes are a reflection of the strategic choices of managers and these choices are in turn a result of their observable and physiological upper echelon characteristics (Hambrick & Mason, 1984). This shows that the manager’s personality heavily influences the decision making in an organisation, which is at the hearth of the upper echelons perspective (Oesterle et al., 2016).
Previously, merely the observable aspects of CEOs within the upper echelon approach were taken into account, such as age, tenure, functional background and education (Oesterle et al., 2016). Next to these observable characteristics, more sophisticated aspects of the
CEO’s personality, such as narcissism, can be incorporated as well, being a part of his or her physiological upper echelon characteristics.
Research has already made clear that, using the upper echelons theory, narcissism influences risk taking (Chatterjee & Hambrick, 2007) and internationalisation decisions (Oesterle et al., 2016) of CEOs. It is therefore reasonable to expect that the amount of CSR disclosure of an organisation, being a strategic decision, is determined by the level of narcissism of the CEO. That is, narcissistic CEOs may be more willing to take visible choices that may cause social praise (Vazire & Funder, 2006), and hence make strategic decisions such as the disclosure of CSR information, which fulfil that demand.
3.3 APPROACH/INHIBITION THEORY
Whereas the abovementioned impression management and upper echelon theory are mainly relevant for the relationship between narcissism and disclosure, the influential role of power concerning this relationship can be explained by the approach/inhibition theory. This theory is founded by Keltner et al. (2003), who were aiming for an integrative way to explain the influence of power on human behaviour, and the consequent acting of individuals within the social environment.
The theory states that dimensions of approach and inhibition determine the response of an individual to its environment (Keltner et al., 2003). The behavioural approach system is concerned with the regulation of behaviour related to areas such as achievement and social attachment. The behavioural inhibition system is concerned with the regulation of behaviour related to threats, uncertainty and punishment. Research has shown that these two systems have different effects on the motivations, emotions and actions of individuals (Lewellyn et al., 2012).
When individuals possess power, mainly approach related behaviour will be activated (Keltner et al., 2003). This has, as Kelter et al. continue, two reasons. Primarily, power is associated with relatively higher amounts of financial and social resources, such as money, attraction, esteem and praise. Second, when experiencing power, individuals recognise that they have the freedom to act without interference of others. Contrastingly, individuals with less power have scarcer social and financial resources and are more constrained by others, creating a relatively greater tendency towards inhibition related behaviour.
As part of the approach related behaviour that is activated by the possession of power, individuals are more sensitive to rewards, neglecting potential threats, and thus see others ‘through a lens of self-interest’ (Keltner et al., 2003: 271). When making decisions, powerful individuals will therefore be more inclined to see opportunities that can create social rewards that are in line with their self-interest. Contrastingly, less-powerful individuals exhibit more inhibition related behaviour and will thus focus more on threats, instead of opportunities, and on the interests of others.
Moreover, the tendency towards approach related behaviour is especially relevant in combination with personal traits. As one of their main findings, Keltner et al. (2003: 266), concluded that ‘power enhances the expression of trait-consistent behaviour’, such that, for example, exchange oriented individuals act in a self-serving fashion, when given power.
That is, powerful individuals, as previously mentioned, have the resources to act in their own interests and are not constrained by the interference of others, causing them to live more by their traits. Contrastingly, less powerful individuals have less resources and feel constrained in their actions, causing them to act more in external interests.
Combining the findings of the approach/inhibition theory, several useful insights regarding the antecedents of CSR disclosure can be named. Firstly, the resource-availability and freedom of powerful CEOs may enhance their tendency towards actions that satisfy their self-interests in terms of achievement and social attachment. As the upper echelon and impression management theory already showed, narcissists have especially great interest in these areas, in which a positive reputation and image can be shaped. Besides, narcissism can be observed as a personality trait (Raskin & Terry, 1988). The approach/inhibition theory states that powerful individuals live more by their traits, and hence, the narcissism of powerful CEOs may emerge relatively more than the narcissism of their less powerful counterparts.
4. HYPOTHESES GENERATION
This sections deals with the specific predictions about the direct relationship between CEO narcissism and CSR disclosure and the moderating role of CEO power. The concepts of CEO narcissism and power will also be explained in more detail.
4.1 CEO NARCISSISM
Freud (1921) was the first to mention the term narcissism. In these years, narcissism was largely observed as a clinical disorder that could be identified by excessive self-admiration and self-preoccupation by individuals (Chatterjee & Hambrick, 2007). Within this context, the Diagnostic and Statistical Manual of Mental Disorders (DSM) describes narcissism as a disorder characterised by ‘a pervasive pattern of grandiosity, need for admiration, and lack of empathy present in a variety of contexts’ (American Psychiatric Association, 2001). This personality dimension is caused by a combination of genetic factors and early parental relations, and may be slightly changed during an adult’s life due to experiences and environmental stimuli (Chatterjee & Hambrick, 2007).
Later, narcissism was observed as a personality dimension, which can be rated among various scales (Raskin & Terry, 1988). For the scope of this study, narcissism is also observed as personality dimension, rather than a clinical disorder. A narcissistic person is then defined as an individual ‘with very inflated self-views and preoccupied with having those self-views continuously reinforced’ (Campbell et al., 2004: 298).
Narcissism has cognitive and motivational elements. Firstly, narcissists find themselves relatively more intelligent, creative and competent from a cognitive perspective (Chatterjee & Hambrick, 2007). Additionally, from a motivational perspective, narcissists constantly seek reaffirmation of their superiority from others. Research suggests that leaders are chosen on the match of people’s characteristics with an ideal conception of a prototypical leader (Grijalva et al., 2015). The cognitive and motivational elements of narcissistic people are, as Grijalva et al., continue, ‘leader like’ characteristics, which causes narcissistic individuals to emerge as leaders within groups, as these characteristics match with the ideal conception of prototypical leaders.
However, the personality trait not only influences the chance of becoming a leader, yet it also influences the subsequent decisions the leader takes. A CEO can be regarded as the most important intra-firm leader, and, consequently, the influence of CEO characteristics on strategic decision making is significant (Oesterle et al., 2016). This influence can be traced back to both the formal (e.g. legal authority and responsibility) and symbolic (e.g. compensation and media coverage) power of CEOs. Focusing on strategic decision making, one of the recent topics of interest is CSR and the disclosure hereof. (Tschopp & Heufner, 2005). Petrenko et al. (2016) state that the personal values of narcissistic CEOs are especially influential in discretionary decision making, such as the decision to engage in CSR activities and subsequent disclosure, because these decisions are difficult to evaluate. Besides, CSR disclosure can generate ‘narcissistic supply’ from external (e.g. media) and internal (e.g. employees) sources, creating possibilities for CEO exhibitionism (Wallace & Baumeister, 2002). These statements are in accordance with the upper echelons perspective, which states that organisational outcomes are a reflection of the strategic choices of managers and these choices are, amongst others, a result of their physiological (e.g. narcissistic) characteristics (Hambrick & Mason, 1984).
In addition, CEO narcissism may influence the engagement in CSR disclosure, as indicated by TB scores, due to the influence of the character trait on Corporate Social Performance (CSP). As previous research and impression management theory have revealed, narcissistic CEOs are more willing to participate in CSR activities, increasing their CSP (Petrenko et al., 2016; Hooghiemstra, 2000). Furthermore, organisations tend to disclose less specific and complete information if performance is negative, and vice versa (Guay et al., 2016), leading to a cost-benefit analysis of the desired level of disclosure. Organisations with high CSP should therefore be more inclined to specifically disclose their strong social performance. Specific and complete disclosure leads to a higher score in the TB, as this benchmark is largely set up by factors that are involved with corporate social activities, such as prevention of bribery and corruption, paying taxes, impact on the labour market and environmental impact (Ministry of Economic Affairs, 2016).
Besides, the most transparent organisation wins The Crystal, a price that is accompanied with public attention (Ministry of Economic Affairs, 2016a). Narcissistic CEOs will see this price as another opportunity to reinforce their inflated self-views. This argument is in line with literature, which states that CEOs with narcissistic character traits are especially sensible for social praise and public attention (Vazire & Funder, 2006). Consequently, the price may, as a form of narcissistic supply, enhance the CEO’s feeling of superiority (Petrenko et al., 2016). Research also shows that narcissistic CEOs might be more proactive in their approach to organisational developments (Wales et al., 2013). Recently, major changes are occurring in the domain of CSR reporting, such as the implementation of 2014/95/EU, which makes reporting about environmental and social aspects mandatory for large public-interest entities in the near future (European Commission, 2016). It is therefore reasonable to expect that narcissistic CEOs will proactively react to these legislative changes by implementing these standards earlier than other less-narcissistic CEOs, which leads to more transparency and hence higher TB scores.
At last, narcissistic CEOs may favour to execute the opportunity for assessment and self-analysis in the TB relatively more to reinforce their inflated self-views (Campbell et al., 2004). This may cause higher TB scores, as participating in the self-assessment may ensure that all appropriate and available accountability information is included, as opposed to a direct review by the accounting firm when neglecting the self-assessment (Ministry of Economic Affairs, n.d.).
Based on the arguments above mentioned, the following hypothesis is provided:
H1: CEO narcissism positively influences CSR disclosure
4.2 CEO POWER
CEO power can be defined as ’the CEO’s ability to effectively run the organisation, extend his or her own tenure, and pursue strategies that will be personally benefitting’ (Gavin, 2014: 48). CEOs can achieve this by the mobilisation of resources to direct strategic actions (Walls & Berrone, 2015). The (formal) power of CEOs is an implicit result from the nature of the job, but the amount of social influence can differ. More specifically, the board of directors and the top management team significantly influence the amount of power a CEO has (Walls & Berrone, 2015). CEOs with greater power have the opportunity to ‘imprint their personal inclinations on major organisational outcomes’ (Chen & Zhu, 2014: 15), as they have more influence on major organisational decisions, limiting the influence of others. CEO power can be further divided into 4 subdomains, those being: structural, ownership, expert and prestige power (Finkelstein, 1992).
Structural power is concerned with the organisational structure and hierarchical authority (Finkelstein, 1992). This source of power is derived from the organisational position a CEO has, which causes CEOs to directly and indirectly control the actions of employees.
Ownership power is focused on the strength of the CEO’s position in relationship with shareholders (Finkelstein, 1992). If a CEO possesses larger proportions of the company’s shares, he/she has more control over the board members of the company and can pursue his/her own goals.
Expert power stresses the knowledge and experience of CEOs in critical organisational areas (Finkelstein, 1992). When a CEO possesses more expert power, he/she can deal better with organisational uncertainties and can make better strategic choices.
Prestige power deals with the personal status a CEO has in an organisation (Finkelstein, 1992). That is, the perceptions of stakeholders inside and outside the organisation about the power a CEO has, are influenced by the CEO’s internal and external reputation.
Recent research found that the possession of these various sources of power influences the psychological processes of executives (Lewellyn et al., 2012). Building on the approach/ inhibition theory, the possession of power has effect on individual motivations, emotions and action, by stimulating specific parts of the behavioural approach system. This causes more powerful CEOs to be mainly focused on upside consequences of opportunistic actions, such as the rewards that may be related to these actions (Lewellyn et al, 2012).
These findings suggest that CEOs possessing power will not always engage in profit-maximizing organisational decision making, as the level of power may cause a CEO to aggressively pursue his or her own interests (Gavin, 2014). Moreover, the power of CEO’s reveals itself in the social performance and image of the firm, since CEOs can also lead the decision making concerning these organisational domains (Li et al., 2016). This opens the way for narcissistic CEOs to engage in more CSR activities and the disclosure hereof, to further reinforce their inflated self-views (Campbell et al., 2004). This is in line with the findings of Keltner et al. (2003), who pose that powerful individuals will be exerting more trait-consistent behaviour. Thus, powerful CEOs will use their power and consequent discretion in decision making to give substance to their narcissistic personality, which causes CEO power to positively moderate the relationship between CEO narcissism and CSR disclosure, as indicated by the TB score.
Based on the arguments above mentioned, the following hypothesis is provided:
5. METHOD
To conduct an investigation on the relationship between CEO narcissism and power on CSR reporting, a data sample of companies participating in the 2016 TB benchmark is being used, which assesses companies’ CSR disclosure for the reporting year 2015. In 2016, 512 companies that met the TB transparency benchmark participating criteria were part of the research group. Within this group, 262 companies have not obtained a score. Non-scoring organisations are not relevant for the scope of this study, since no causal relationships between our independent and dependent variables can be analysed for these organisations. Hence, these organisations are excluded from the data sample. Next, 26 organisations did not report on one or more financial indicators (i.e.: total assets, return on assets and leverage), and are therefore also excluded from the sample. Subsequently, 2 excluded organisations were led by co-CEOs, troubling the intended individual measurement of narcissism and power. Lastly, 1 company is excluded as it had no supervisory board by the end of 2015. A detailed presentation of the sample selection is presented in table 1.
TABLE 1: SAMPLE SELECTION
Firms participating in the 2016 Tranparency Benchmark 512 LESS:
Firms that obtained no score (262) Firms that did not report on financial indiactors (26) Firms that are managed by CO-CEOs (2) Firms without a supervisory board (1)
Final Sample 223
The resulting sample therefore consists of 223 organisations from 17 different sectors. The sectors services (49), banks and insurers (36) and food and beverages (23) supply the largest share of organisations within the sample. The relevant data on various variables is hand-collected and derived from the 2015 annual and CSR reports of the 223 TB contenders.
5.1 MEASUREMENT OF CEO NARCISSISM
Research measuring CEO narcissism, the independent variable, predominantly uses the Narcissistic Personality Inventory (NPI) as instrument to analyse this personality trait (e.g. Petrenko et al., 2016). This index, founded by Raskin & Terry (1988), uses the DSM criteria to measure narcissism. These DSM criteria observe impairments in personality functioning (e.g. exaggerated self-appraisal, lack of empathy and a need for personal gain) and the presence of pathological personality traits (e.g. antagonism, grandiosity and attention seeking) as essential features of narcissism (American Psychiatric Association, 2001). The NPI can both be used as a self-reporting tool, but is merely used as a rating instrument for third-party experts that analyse short videos of CEOs (Petrenko et al., 2016).
Chatterjee & Hambrick (2007) and Zhu & Chen (2015) however suggest that CEO narcissism can be measured by using several unobtrusive measures of personality. This methodology is more feasible, since self-reporting questions about personality traits such as narcissism would yield low response rates and would be greatly influenced by a social desirability bias. Moreover, the use of unobtrusive measures is also favoured above video sampling, because publicly available data can be used and reliability concerns regarding inter-rater differences can be avoided.
The validity of unobtrusive indicators to measure CEO narcissism is further supported by findings of Zhu & Chen (2015) that show a correlation coefficient of .84 between third-party video ratings, using the NPI and the measurement via unobtrusive indicators. Additionally, Chatterjee & Hambrick (2007) concluded that 40 experienced security analysts rated the narcissism of CEOs in line with the scores of unobtrusive indicators, indicated by a correlation coefficient of .82.
Specifically, CEO narcissism will be measured by forming a CEO Narcissism Index (CNI). The CNI is built upon three previously known unobtrusive indicators, being (1) the CEO’s prominence in the company’s annual report, (2) the CEO’s use of first-person singular pronouns and (3) the CEO’s relative pay (Zhu & Chen, 2015; Chatterjee & Hambrick, 2007), and two new unobtrusive indicator, being (4) length of the CEO’s letter to shareholders and (5) the CEO’s LinkedIn prominence.
The CEO’s prominence in the company’s annual report (PromAnnual) is one of the indicators
of CEO narcissism, because CEOs normally have specific preferences about the portrayal of their photo within annual reports (Chatterjee & Hambrick, 2007). It can be argued that more narcissistic individuals are more inclined to show their importance and leadership in one of the major external communication channels of the organisation. Therefore, narcissistic CEOs will have photographs that are relatively larger than their non-narcissistic counterparts. The precise measurement of this indicator is included in appendix A.
The CEO’s use of first-person singular pronouns (First-Singular) is another indicator of
narcissism (Chatterjee & Hambrick, 2007). Narcissistic CEOs tend to have a great interest in outside communications with stakeholders, as these actors can enforce their inflated self-views (Petrenko et al., 2016). Consequently, a narcissistic CEO will use more first-person singular pronouns (I, me, mine, my, myself), as opposed to first-person plural pronouns (we, us, our, ours, ourselves) (Chatterjee & Hambrick, 2007). In specific, this indicator can be measured as the number of first-person singular pronouns divided by the number of first-person plural
pronouns in one of the company’s major press releases, being the letter to shareholders. The precise measurement of this proxy is included in appendix A.
The CEO’s relative pay (RelativePay) is another measure that can capture narcissism, since CEOs
have the ability to almost entirely control the compensation of other executives. Narcissistic CEOs will use this ability to create a pay gap between them and the executive with the second-highest compensation, to emphasize their importance and superiority (Chatterjee & Hambrick, 2007). This indicator can be measured as the CEO’s total compensation divided by the total compensation of the second-highest compensated executive in the firm. The precise measurement of this proxy is included in appendix A.
Additionally, this study will enrich the measurement of CEO narcissism by appending another unobtrusive indicator, being the length of the CEO’s letter to shareholders
(LetterShareholder). The prominence of CEOs in the company’s press releases already showed
that narcissists have a great interest in influencing perceptions of outside actors by external communications (Chatterjee & Hambrick, 2007). This may be especially true for the personal message of CEOs to shareholders, being a ‘periodic, widely read, written, signed, and public presentation of the firm’s goals, actions, and results’ (Patelli & Pedrini, 2014: 20). Similarly, Hooghiemstra (2000) states that impression management usually manifests itself in these letters. To positively shape the impression of readers, improving external perceptions about the CEO’s leadership and his/her reputation, managers may need more words to deliver their message. Previous research of Patteli & Pedrini (2014) has already shown that, supporting this line of reasoning, positive messages in the CEO’s letter to the shareholders are lengthier than negative messages. The precise measurement of this indicator is included in appendix A.
Lastly, the CEO’s LinkedIn prominence (PromLinkedIn) is another unobtrusive indicator considered in the measurement of narcissism. Research suggests that LinkedIn, an online platform which allows people to post resumes and connect with others, is a major medium for individuals to achieve self-presentational goals, aimed at the creation of positive impressions (Guillory & Hancock, 2012). They state that social networking profiles are ‘designed to convey impressions to an audience’ (Guillory & Hancock, 2012: 136). LinkedIn is currently the world’s largest professional network, with over 467 million professional users. Narcissists may therefore find this online medium a useful place to convince this large audience of their superior competencies, using their individual public profile extensively as a ‘display window’. Narcissists may therefore be relatively more inclined to use LinkedIn actively than average individuals. There are several ways to create a ‘display window’ out of a public profile, of which four are explicitly considered in this study. These four indicators are built on four conceptual elements of narcissism, mentioned in Chatterjee & Hambrick, (2007: 365), being: ‘leadership/ authority’, ‘self-absorption/self-admiration’, ‘superiority/arrogance’ & ‘exploitativeness/ entitlement’.
Firstly, similarly to our first proxy, narcissistic CEOs can use their photograph to imprint their leadership and authority to possible audiences, showcasing themselves. Less narcissistic individuals may not have this goal, and will therefore be less inclined to include a photograph to enrich their public profile. Secondly, narcissists may show their absorption and self-admiration by including a personal message on their public profile, apart from their working experience and education section, in which they can emphasize their extraordinary self.
It is up to individual choice to use this section, which should cause narcissists to be more inclined to include such a story. Thirdly, narcissistic executives can display their superiority or arrogance, by providing a detailed job description, pinpointing their superiority over others in their role as CEO. On the other hand, less narcissistic individuals may only include their job title and employer, without including a detailed description about the job. Fourthly, exploitaitiveness and entitlement can be shown in the ‘Skills & Endorsements’ section on the public profile of a CEO. That is, this section can be specifically used to gain esteem and respect of others, by leaving them the possibility to endorse his/her skills. Less narcissistic CEOs may be less inclined to actively use this section on their public page, as users can voluntarily choose to add this section to their profile. The precise measurement of this proxy is included in appendix A.
In order to construct the CNI an Exploratory Factor Analysis (EFA) has been executed. EFA has the power to assess the relative importance of the five CNI proxies, given that the proxies represent the underlying structures of CEO narcissism (Reio and Shuck, 2015). The main goal of EFA is to capture all original information in as less factors as possible. That’s why the statistical technique of Principal Component Analysis (PCA) will be used, since PCA can generate one distinct CNI score, which summarises the five CEO narcissism proxies. Besides, PCA is observed as the highest-quality EFA technique deployed in research (Reio & Shuck, 2015). Note that higher CNI score indicate more narcissistic CEOs.
5.2 MEASUREMENT OF CEO POWER
Building on the four (i.e.: structural, ownership, expert and prestige) CEO power dimensions of Finkelstein (1992), this study forms a CEO Power Index (CPI). The formation of a CPI is valuable, since the examination of individual elements of power has created empirically mixed and inconsistent results (Gavin, 2014). Following the methodology of Lisic et al. (2016), at least one specific indicator for every power category is determined.
To measure the extent of structural power, the formal position of the CEO is most relevant. Therefore, previous research has extensively examined CEO duality, the situation where a CEO is also chairman of the board of directors, as an indication of structural power (Gavin, 2014). Since duality is legally prohibited in The Netherlands, this study uses, consistent with Sarkar et al. (2008), the ratio between the number of executive board members compared to supervisory board members (ExecRatio) as structural power measure. That is, the task of the supervisory board is to evaluate, supervise and discipline the CEO and other executives, which can be achieved by controlling the agenda of board meetings and the information flow within these meetings (Gavin, 2014). Consequently, if the amount of supervisory board members is relatively low, the CEO (and other executives) gain more power.
Ownership power is determined by two proxies. Firstly, the CEO can have share ownership
(ShareOwn) in the firm, which may enhance the control a CEO has over the board members
of the company, creating power and possibilities to pursue his/her own goals (Finkelstein, 1992). Next, a CEO or his/her first-degree relative that is also the founder (Founder) of the company may have more ownership power (Lisic et al., 2016).
Two proxies are being used as an indication of CEO expert power. The tenure (Tenure) of CEOs enriches their knowledge and experience about the firm and its environment, which
allows them to make better decisions (Gavin, 2014). Also, CEOs may derive power from former executive positions (NumExec), which they held in the organisation, with executive positions being vice-president, president, CFO, COO or General Manager (Lisic et al., 2016).
Prestige power is measured by the number of memberships the CEO has in other corporate boards (BoardMem) (Lisic et al., 2016), since the maintenance of relationships and social ties with important business people positively influences CEO’s reputation inside and outside the organisation. That is, when a CEO is member of the board of directors in another organisation, these organisations also see the merits of the CEO’s opinion and service, leading to more CEO prestige (Lewellyn et al., 2012).
Similar to the formation of the CNI, the CPI will be formed by using the EFA technique CPA, to capture the relative importance of the CPI proxies that represent the underlying structures of CEO Power, into one CPI score (Reio and Shuck, 2015). Subsequently, higher CPI scores indicate more powerful CEOs.
5.3 MEASUREMENT OF CSR DISCLOSURE
The dependent variable, CSR disclosure, is measured by the TB score of an organisation, which is an annual research, conducted since 2004, on the content and quality of CSR reporting of the largest Dutch companies (Ministry of Economic Affairs, 2016). Both private sector companies as well as governmental institutions are included within the research group. The objective of the TB is to increase transparency on CSR reporting, since transparency enables stakeholders to form a view about the CSR performance of an organisation, and hence, start a dialogue on this matter (Ministry of Economic Affairs, 2016a). CSR reporting can be done by the provision of a separate CSR report or by integrating CSR matters in the annual report. These forms of reporting can provide comprehensive understanding of economic, environmental and social value creation, in the short-, medium- and long-term.
Participation in the TB is not voluntary, as different criteria, such as turnover, amount of employees and public listing, determine participation. Organisations that participate in the TB can score on eight different perspectives, those being:
1. Company and business model 2. Policy and results
3. Management approach 4. Relevance 5. Clearness 6. Reliability 7. Responsiveness 8. Coherence
A distinction is made between categories 1-3, which are content-oriented and categories 4-8, which are quality-oriented. These criteria are based on the GRI and IIRC frameworks. In both the content-oriented and quality-oriented categories, organisations can earn 100 point, resulting in a maximum potential score of 200 points. The aim of the TB is to increase the achieved score of the participating companies every year.
Based on the total score of the participating organisation, it is ranked in one of the five following groups: 1. Leaders 2. Followers 3. Peloton 4. Laggards
5. Companies with zero scores
This study incorporates 223 Dutch companies with a positive TB score and sufficient information on factors included in table 1 regarding their 2015 reporting. Within this sample, the maximum score is 199 point and the minimum score is 5 point.
5.4 CONTROL VARIABLES
Previous research has shown that several other variables can influence the degree of CSR disclosure. To control for confounding effects, control variables on both industry (sector), corporate (size, profitability and growth) and corporate governance (board size) level are included in this study.
The industry (sector) in which the organisation is active can have an impact on the amount of CSR disclosure (Luo et al., 2017). In particular, organisations active in the polluting industry experience higher amounts of public pressure, creating incentives to engage relatively more in CSR reporting. To control for this effect, a dummy variable of 1 is given to organisations in the sector ‘Energy, oil & gas’, and a dummy variable of 0 is given to organisations active in other sectors.
On the corporate level, the size (size), measured as the natural logarithm of total assets, of the company may influence the extent of CSR disclosure, since larger organisations have more stakeholders that are interested in corporate disclosure and have more resources to report on CSR (Craven & Marston, 1999). Next, relatively high levels of organisational profitability
(profitability), measured by the annual Return on Assets (ROA), can cause organisations to
be more subject to media exposure, on which they respond in the form of CSR disclosure (Garcia-Sanchez et al., 2014). Additionally, firms with high levels of organisational leverage
(leverage), operationalized by the debt to equity ratio, may experience more creditor pressure
and therefore have less resources for and focus on voluntary social disclosure (Brammer & Pavelin, 2006).
On corporate governance level, the size of the supervisory board (boardsize) can impact organisational outcomes, such as CSR disclosure, as larger boards are relatively more experienced and diversified and may therefore demand more CSR disclosure to fulfil their monitoring role (Garcia-Sanchez et al., 2014).
Table 2: Measurement of dependent and independent variables
VARIABLE PROXY MEASUREMENT
DEPENDENT VARIABLE
CSR disclosure TB score Score determination based on CSR transparency including content and quality indicators
INDEPENDENT VARIABLES
EXPLANATORY VARIABLE
CEO narcissism CNI Score determination based on five unobtrusive CEO narcissism indicators
MODERATING VARIABLE
CEO power CPI Score determination based on six CEO power proxies, derived from the four CEO power sources
CONTROL VARIABLES
Polluting industries Sector Value 1 for companies in the sector ‘Energy, oil & gas’, 0 otherwise Organisational size Size Natural logarithm of total assets at
year end
Organisational Profitability Net profit/ total assets at year end Organisational leverage Leverage Shareholders equity/ liabilities at
year end
Size of the board Boardsize Size of the supervisory board at year end
5.5 STATISTICAL MODEL
This research conducts an ordinary least squares method on the following statistical model: (1) TB-score = ß0 + ß1(CNI) + ß2(CPI) + ß1(CNI)*ß2(CPI) + ß3(Sector) + ß4(Size) + ß5
(Profitability) + ß6 (Leverage) + ß7 (Boardsize) + ε
Provided that ßi are coefficients and ε is the error term. The various variables are further defined in table 2.
6. RESULTS
This section will firstly discuss the coherence and internal validity of the CEO narcissism and CEO power indicators that form the CNI and CPI, respectively. Next, the descriptive and correlation statistics will be presented. Lastly, the regression analysis will show the results of the hypotheses testing, and, thereafter, these regression results will be tested for robustness and endogeneity.
6.1 MEASUREMENT OF CEO NARCISSISM
The formation of the CNI consist of five narcissism proxy variables. The means, standard deviations and correlations between these proxy variables are presented in table 3.
Table 3: CNI proxies
Mean S.D. 1. 2. 3.* 4. 5. 1. PromAnnual 2.57 1.47 2. First-Singular .05 .09 .42*** 3. RelativePay* 1.39 .55 .29*** .14 4. LetterShareholder 755.83 603.90 .47*** .29*** .07 5. PromLinkedIn 2.82 1.53 .18*** .20*** -.17** .08 *N=132, **p < .05 (two-tailed), ***p < .01 (two-tailed)
As the table indicates, only 132 observations about the relative pay of CEO’s could be obtained. This is caused by the unavailability of executive compensation information in annual reports of merely relatively small companies within the sample. In line with the expectations, most of the proxies show positive correlation coefficients. However, the correlation between RelativePay and PromLinkedin (r=-.17, p < .05) is an unexpected exception, as all proxies are presumed to capture specific elements of the narcissistic character of CEOs, indicating consistent high or low scores on the five proxies, expectantly leading to positive correlation coefficients. Also, the Cronbach’s alpha is calculated to test for internal consistency. The obtained Cronbach’s alpha of .59 is on the low end, but levels between 0.50 and 0.60 are still acceptable for this type of preliminary research (Nunnally, 1967).
To summarise the five CNI proxies into one variable, a PCA is executed. To control the adequacy of the CNI a Bartlett test of sphecerity is conducted, which provides evidence for a significant relationship among the proxies (χ2=76.86, p=0.00). Then a Kaiser-Mayer-Olkin
(KMO) test is performed, to check if the created CNI score in the PCA efficiently reflects the original variables. The KMO test yields an outcome of .62 which is well above the identified threshold of .5 (Hair and Black, 2009). This means that the KMO value is sufficient for a PCA in this setting. In table 4 the results of the PCA are presented. The results highlight that the first 2 proxies (PromAnnual & LetterShareholder) already explain 60.11% of the standardized variance in CEO narcissism. Accordingly, these factors play a major role in the formation of the CNI, as shown in the factor loading column.
Table 4: Principal Component Analysis CNI
Eigenvalue Factor loading % Of variance Cumulative % 1. PromAnnual 1.83 0.85 36.50 36.50
2. LetterShareholder 1.81 0.64 23.62 60.11
3. First-Singular 0.90 0.62 17.89 78.01
4. RelativePay 0.65 0.46 12.97 90.98
5. PromLinkedIn 0.45 0.32 9.02 100.00
6.2 MEASUREMENT OF CEO POWER
Similarly, the six proxy variables forming the CPI will be analysed. Table 5 shows the means, standard deviations and correlation coefficients between these proxy variables.
Table 5: CPI proxies
Mean S.D. 1. 2. 3. 4. 5. 6. 1. ExecRatio .57 .31 2. ShareOwn .43 .50 -.07 3. Founder .09 .29 .20** .36** 4. Tenure 5.54 7.09 .11 .30** .47** 5. NumExec .75 .80 .01 .37** .20** .26** 6. BoardMem 1.30 1.68 -.09 -.16** -.16* -.05 -.03 *p < .05 (two-tailed), **p < .01 (two-tailed)
The table displays mixed results. The major part of the CEO power proxies positively and significantly correlate as expected. Contrastingly, the amount of outside board memberships (BoardMem) a CEO has, being an indicator of a CEO’s prestige power, shows solely negative (significant) correlations with the other proxies. The Cronbach’s alpha of the six proxies amounts .40, below the identified threshold of Nunnally (1967). Nevertheless, when the BoardMem proxy is excluded, the Cronbach’s alpha increases to .51, just above the threshold. In the remainder of this analysis, the CPI will therefore be computed without the enumeration of the BoardMem proxy.
To form the CPI score that includes the five remaining proxy variables into one variable, a PCA is executed. The Bartlett’s test of sphecerity indicates a significant relationship between the proxies (χ2=142.67, p=0.00). Next, the KMO test yields a value of .66, being well above
the identified threshold of 0.5 (Hair and Black, 2009). The PCA can thus also be run to form the CPI. In table 6 the results of the PCA are presented. The results highlight that the first 3 proxies (Founder, Tenure & ShareOwn) already explain 77.78% of the standardized variance in CEO power. Accordingly, these factors play a major role in the formation of the CPI, as shown in the factor loading column.
Table 6: Principal Component Analysis CPI
Eigenvalue Factor loading % Of variance Cumulative % 1. Founder 2.02 0.76 40.33 40.33 2. Tenure 1.08 0.74 21.55 61.89 3. ShareOwn .80 0.71 15.90 77.78 4. NumExec 0.63 0.59 12.67 90.45 5. ExecRatio 0.48 0.23 9.55 100.00 6.3 DESCRIPTIVE STATISTICS
Table 7 presents the descriptive statistics of the investigated dependent, independent and control variables in this study. The results show that, on average, companies scored 111.69 points in the 2015 TB. Furthermore, the CNI and CPI scores show means of 0 and standard deviations of 1, which is caused by the generation of these variables using PCA (Reio and Shuck, 2016). Additionally, only approximately 5% of the sample originates from polluting industries. The average amount of total assets, amounting €20.325.030.000, is relatively high, since the sample consists of the largest Dutch companies. The results also show the average profitability and leverage of the sample, being 4.23% and 36.39% respectively. Lastly, the average size of the supervisory board of the organisations included in the sample is 5.81.
Table 7: Descriptive statistics
Mean S.D. Min. Max.
TB score 111.69 56.76 5 199 CNI .00 1.00 -1.79 2.13 CPI .00 1.00 -1.08 4.32 Sector 0.05 0.22 0 1 Size (*€1.000.000) 20325.03 86046.20 6.43 841769 Profitability (%) 4.23 9.28 -38.93 50.48 Leverage (%) 36.39 23.13 -19.37 99.99 Boardsize 5.81 2.37 0 14 6.4 CORRELATION ANALYSIS
Table 8 displays the correlation coefficients between the various variables. Coherent with the expectations, CSR disclosure, operationalized by the TB score, has a positive and significant correlation with the index for CEO narcissism (r=0.35, p=0.00). This is an indication for the confirmation of H1. Organisational size, operationalized by total assets (r=0.27, p=0.00), and board size (r=0.22, p=0.00) significantly cohere with CSR disclosure as well. Contrastingly, the correlation coefficient between the CEO narcissism and the CEO power index is negative (r=-0.01, p=0.89), indicating a rejection of H2.
The highest correlation coefficients between control variables were found between size and leverage (r=-0.28, p=0.00) and size and supervisory board size (r=0.28, p=0.00). As a general rule, a correlation coefficient greater than |0.7| between two independent variables can bring up multicollinearity issues, the situation in which two or more independent variables in a regression model are strongly correlated (Valentine, 1969). The correlation coefficients in this study remain well below this limit.
Table 8: Correlation coefficients
1. 2. 3. 4. 5. 6. 7. 8. 1. TB score 2. CNI 0.35** 3. CPI -0.82 -0.01 4. Sector 0.12 -0.03 -0.05 5. Size*** 0.27** 0.14* -0.01 0.08 6. Profitability*** 0.13 0.01 0.09 0.02 -0.07 7. Leverage*** -0.06 0.02 -0.05 -0.02 -0.28** 0.03 8. Boardsize 0.22** 0.13 -0.09 0.07 0.28** 0.10 -0.15*
*p<.05 (two-tailed), **p<.01 (two-tailed) ***Prior to the correlation analysis, winsorizing is applied to this variable to control for extreme values.
6.5 REGRESSION ANALYSIS
To perform a regression analysis using the ordinary least squares method, homoscedasticity is assumed. To test this assumption, a Breuch-Pagan test is executed, given that under H0 homoscedasticity is present. The statistical results reveal that the assumption holds (LM=2.71, p=0.91).
Table 9 provides the results of the regression analysis. Model 1 is a controls-only model that explains approximately 16.4% (adj. r2=0.164) of the variance in CSR disclosure and is significant
(F=9.175, p=0.000). In line with the literature, this study finds that the size (ß=10.999, p=0.000) and profitability (ß=0.848, p=.020) of organisations significantly influence CSR disclosure (Craven & Marston, 1999; Garcia-Sanchez et al., 2014). Contrastingly, no significant relationships are found between CSR disclosure and polluting industries (ß=13.141, p=0.420), leverage (ß=0.283, p=0.088) and the size of the board (ß=1.043, p=0.522) (Luo et al., 2017; Brammer & Pavelin, 2006; Garcia-Sanchez et al., 2014).
Model II presents the test statistics for the assessment of hypothesis 1. Building on impression management and upper echelon theory, narcissistic CEOs are expected to observe CSR disclosure as an opportunity to positively influence outside perceptions about the CEO (Petrenko et al., 2016). Hence, CSR disclosure can generate sources of ‘narcissistic supply’ in the form of media attention and praise (Wallace and Baumeister, 2002). Consequently, hypothesis 1 predicts that more narcissistic CEOs will disclose more CSR information. This hypothesis is confirmed in model II, since the coefficient is positive and significant (ß=15.587, p=0.000). Model II explains approximately 24.0% of the variation in CSR disclosure (adj. r2=0.240) and is significant (F=11.023, p=0.000)
Model III adds the interaction term of CEO narcissism and CEO power. Based on the approach/inhibition theory, approach related behaviour will come relatively more to the fore
if individuals possess high levels of power (Lewellyn et al., 2012). Powerful CEOs will therefore use their power and consequent discretion in decision making to give substance to their narcissistic personality. Accordingly, hypothesis 2 predicts a positive interaction between CEO narcissism and CEO power. This hypothesis is not supported in model III, since the interaction term is negative and insignificant (ß=-0.195, p=0.954). Model III explains similar amounts of variation in CSR disclosure as model II (adj. r2=0.237) and is also significant (F=9.601, p=0.000).
To control for multicollinearity, the highest variation inflation factor (VIF) for every model is provided. The VIF’s of model 1, model 2 and model 3 with values of 1.469, 1.543 and 1.547 respectively, stay well below the identified threshold of 10 (Kennedy, 1992).
Table 9: Regression analysis
Variables Model I Model II Model III
Independent variables CNI (H1) - 15.587 (3.425)** 15.577 (3.438)** CPI - -5.719 (3.565) -5.719 (3.565) CNI*CPI (H2) - - -0.195 (3.362) Control variables Sector 13.141 (16.279) 16.404 (15.578) 16.364 (15.629) Size 10.999 (1.995)** 9.507 (1.949) ** 9.512 (1.956)** Profitability 0.793 (0.379)* 0.848 (0.363)* 0.849 (0.364)* Leverage 0.283 (0.165) 0.190 (0.159) 0.189 (0.159) Boardsize 1.043 (1.628) 0.354 (1.568) 0.355 (1.572) Constant -140.262 (42.128) ** -101.827 (41.305)** -101.939 (41.446)** Observations 223 223 223 F-test 9.715** 11.023** 9.601** Adj. R2 0.164 0.240 0.237 Highest VIF 1.469 1.543 1.547 **p<0.01 (one-tailed), *p<0.05 (one-tailed) 6.6 ROBUSTNESS CHECKS
The empirical results supporting the first hypothesis may be biased, since the significant results may be caused by confounding variables that influence the chance of being a narcissistic CEO. That is, simply comparing CSR disclosure outcomes between narcissistic and non-narcissistic CEOs may not entirely prove the direct relationship between the two, as organisations led by narcissistic CEOs possibly possess a set of characteristics that significantly differ from organisations led by non-narcissistic CEOs. To control for these effects, this study, coherent with the literature, applies propensity-score matching (Lisic et al., 2016).
Propensity-score matching preliminary identifies a treatment and a control group. In this study, the treatment group consists of above average narcissistic CEOs and the control group of below average narcissistic CEOs, based on CNI scores. Consequently, these groups are matched on several characteristics, using the estimated chance of receiving the treatment
