CEO Succession and the Firm’s Entrepreneurial Orientation: The Moderating Role of
CEO Narcissism
June 22, 2020 Julia Smit | S2944545
Supervised by Dr. Philip Steinberg | Co-assessed by Prof. Dr. Pedro de Faria
Master Thesis | Msc BA Strategic Innovation Management | Faculty of Economics and Business | University of Groningen
Word count: 12,504
Keywords: CEO Succession | Entrepreneurial Orientation | CEO Narcissism | Territory Marking
Abstract: Guided by the upper echelon theory, this study examines what influence newly appointed CEO’s have
on the firm’s EO, and more specifically if that relationship is moderated by the presence of narcissistic personality traits. I argue that a CEO succession is positively related to the firm’s EO which is substantiated by the new CEO’s
territory marking. In addition, I expect that a CEO with narcissistic personality traits is more eager to mark its
territory and engage in new opportunities, hence strengthening the effect of one’s succession on the firm’s EO.
Table of content
1. Introduction ... 3
2. Theoretical Background and Hypotheses ... 5
2.1 Entrepreneurial Orientation ... 5
2.2 CEO succession and tenure ... 6
2.3 CEO narcissism ... 7
2.4 Hypotheses development ... 8
2.4.1 CEO Succession and EO ... 8
2.4.2 The Moderating Role of CEO Narcissism ... 9
3. Methodology... 10
3.1 Sample & Data Collection ... 11
3.2 Measurements ... 11
3.2.1 Due diligence and data management ... 11
3.2.2 Independent variable: CEO succession ... 12
3.2.3 Dependent variable: Entrepreneurial Orientation... 12
3.2.4 Moderating variable: CEO Narcissism ... 13
3.2.5 Control variables ... 14
3.2.5.1 CEO control variables... 14
3.2.5.2 Firm control variables... 14
3.2.5.3 Additional control variables ... 15
3.3 Analytical Method ... 16
4. Results ... 17
4.1 Descriptive Statistics and Correlations ... 17
4.2 Regression Results and Hypotheses Testing ... 18
4.3 Robustness analysis ... 19
5. Discussion and Conclusion ... 20
5. 1 Theoretical Implications... 22
5.2 Practical Implications ... 23
5.3 Limitations and Future Research ... 24
References ... 25
Appendices ... 29
Appendix A: Dictionaries for Entrepreneurial Orientation (Short et al.,2010) ... 29
1. Introduction
Driven by an Artificial Intelligence (AI) oriented culture, Tesla was ranked in 2019 as the 9th most
innovative company in the world. According to the Boston Consulting Group, they are pioneers of innovation and the leading explorers of AI in the automotive industry (Columbus, 2019). But what does it take for a firm to become so entrepreneurially oriented? A narcissistic Chief Executive Officer (CEO) like Elon Musk maybe?
Often he is painted by the public as a great narcissist, which he confirms himself by tweeting: ‘If I am a
narcissist (which might be true), at least I am a useful one’. Is it his bold visions and great overconfidence that leads Tesla to be so entrepreneurial? (Coren, 2019). In our contemporary world of continuous innovation firms that manifest entrepreneurial orientations (EO) are here to stay (Wales, Monsen, Mckelvie, 2011). ‘EO refers to
the strategy-making processes that provide organizations with a basis for entrepreneurial decisions and actions’
(Rauch,Wiklund, Lumpkin, & Frese, p.762, 2009).
Previous literature links EO to positive firm performance (Rauch, et al., 2009) and more specifically
connects EO to greater innovation outcomes (Wales, Gupta, & Mousa, 2013). As a consequence, firm strategies often include innovative goals in order to facilitate EO within their firms (Rauch et al., 2009). As result of the great potential for firm outcomes, EO is nowadays far from under researched and has become a cornerstone in literature (Wales, Gupta & Mousa, 2013).
EO research cuts across multiple disciplinary boundaries (Boehm, 2008), one domain receiving much attention are is the field of CEO tenures and life cycles (Grühn, Strese, Flatten, Jaeger & Brettel, 2017; Boling,
Pieper, & Covin, 2015; Richard, Wu & Chadwick, 2009). A main topic in this field is CEO successions. Media
attentiveness has led researchers to explore this public phenomenon (Boling et al., 2015) and have consequently
linked CEO successions to firm level EO(Grühn et al., 2017). Also in the field of upper echelons, CEO
characteristic traits have been linked to EO. According to upper echelon’s theory the executive’s psychological
characteristics, such as the cognitive base and values, form the basis for their strategic choices (Hambrick & Mason, 1984). Especially the CEO personality trait, narcissism, has been of great interest (Chatterjee & Hambrick, 2007; Engelen, Neumann, & Schwens, 2015; Wales, Patel & lumpkin, 2013).
Nonetheless when considering the relationship between a CEO succession and EO outcomes, Grühn et al.
(2017), have focussed on the solo event of a CEO succession on the EO without theorizing about the moderating role of potential narcissistic traits possessed by the new CEO. This moderation of CEO narcissism is however
not unimportant as EO is namely contingent on the characteristics of the firm’s leader (Wales et al., 2011). In
particular, narcissistically characterized CEO’s are interesting as they favour bold actions resulting in either big wins or big losses for the firm (Chatterjee & Hambrick, 2007). Additionally, CEO overconfidence tends to, at decreasing rate, significantly increase the firm’s EO (Engelen et al., 2015). There is thus clear evidence for a relation between CEO narcissism and EO. The presence of narcissistic traits in the new CEO is thus an important moderating mechanism when relating the entrance of a new CEO to the consequent outcome of firm-level EO. Nonetheless, empirical evidence is lacking on the moderating effect of CEO narcissism.
the new CEO’s territory marking behaviour. Furtherly demonstrated by the upper echelon’s theory, I expect that a new CEO with narcissistic personality traits is more eager to mark its territory and engage in new
opportunities, hence increasing the firm’s EO (Engelen et al.,2015). Therefore, I argue that a CEO succession is positively related to the firm’s EO and that this relationship is moderated by CEO narcissism, where a
narcissistic CEO strengthens the relationship
Preceding research of Grühn, et al. (2017) placed absolute change in EO as a result from CEO
replacement. It is proposed that the CEO’s personality, mandate and territory marking behaviour are the
mechanisms bringing about the change in EO after a CEO succession. Especially prevalent in this research is
territory marking. This involves new CEO’s engaging in bold actions in order to gain respect and set an early track record within the firm. This desired admiration and respect one wishes to gain from its followers is also a distinctive character trait of narcissistic personalities (Grühn et al., 2017). The overconfidence that narcissistic CEO’s possess is of good use when engaging in bold actions that CEO’s pursue whilst marking their territory. As this research aims to explore the strengthening effect CEO narcissism on the relationship between a CEO succession and EO, the mechanism of territory marking is most prevalent in this paper.
A newly appointed leader will mark his or her territory in order to distinguish themselves from their
predecessor, which often involves daring moves (Grühn, et al. 2017). These actions resemble the urge of
narcissistic individuals to draw attention to their vision and leadership and participate in competitive behaviour. The relationship that territory marking facilitates between CEO succession and firm-level EO is based upon risky investments and innovative bold actions by the new CEO. Narcissistic leaders are likewise keen on investing in higher risk and higher return projects and will consequently intend to develop an entrepreneurially oriented firm culture (Wales, Patel & Lumpkin, 2013). In greater detail, there is three specific characteristics that narcissistic CEO’s possess that are in line with territory marking behaviour namely the need for praise, overconfidence and
dominance orientation (Kashmiri, Nicol, & Arora, 2017).
I will base my analyses on numerous annual reports and transcripts of quarterly earnings calls of companies in the S&P100 firms between 2008 and 2018. In pursuance of measuring the change in CEO, I will use data from the S&P100 ExecuComp to identify the CEO replacements. EO is measured by a computer-aided-text-analysis (CATA) on the management discussion of firm annual reports (Mckenny, Aguinis, Short & Anglin, 2018). I will identify EO using five dictionaries consisting of innovativeness, proactiveness, risk taking,
autonomy and competitive aggressiveness by Short, Broberg, Cogliser & Brigham (2010). Lastly, I will include a measure for the moderating variable CEO narcissism. The quarterly earnings calls transcripts will be used to determine the relative use of first-person singular pronouns to first-person plural pronouns (Raskin and Shaw’s, 1988). As a measure of robustness, the size of the picture of the CEO in annual reports is deployed.
Second, this research aims to respond to the call of Grühn et al (2017) by including a CEO characteristic. Their previous research on the relationship between CEO successions and EO has thus far lacked to include in this specific personality trait of CEO’s. Whilst, in fact, narcissism has many overlapping components with behavioural actions following a CEO succession and, moreover many CEO’s are often characterized with narcissistic traits. There is thus great relevance in incorporating narcissism as moderator on the effect of a CEO succession on EO. Therefore, this study attempts to fill the literature gap and contribute to the upper echelon literature by conducting an empirical research on the moderating effect CEO narcissism
This paper also makes two practical contributions. First, the paper will help practioners to understand how a CEO succession brings about a change in EO. Moreover, it gives them indications on how this could
potentially negatively or positively affect their firm’s EO and enables them to prepare them for such a change. Second, previous research (e.g. Wales, Patel & Lumpkin, 2013; Kashmiri et al., 2017) often poses that
narcissistic CEO’s greatly influence firms and the organization’s entrepreneurial behaviour. This paper may add nuance to this, and provides practioners a critical notion on whether the appointment of a new narcissistic CEO does actually affect the firm’s EO and if practioners should consequently take narcissistic traits in consideration in their recruitment decisions.
This research will be structured as follows. The second chapter will discuss the theoretical background of the concepts where after the hypotheses will be substantiated. The third chapter is devoted to the methodology which will outline the sample, databases, measurements and analytical method. Thereafter, the fourth chapter will provide the empirical results. The fifth and last chapter discusses the findings, theoretical and practical implications and finally provides limitations and directions for future research.
2. Theoretical Background and Hypotheses
In the following, this research’ theoretical background and formulation of the two hypotheses will be provided. Firstly, I will elaborate on the concept EO, CEO succession, and CEO narcissism. Thereafter, I will end this chapter by presenting this paper’s hypotheses.
2.1 Entrepreneurial Orientation
Multiple definitions of EO exist within literature. Various descriptions of EO are formulated, many describing EO as a multi-dimensional construct. In the meta-analysis of Rauch et al. (2009) EO is reviewed in numerous strategy and entrepreneurship studies. One of the first to define EO as a multi-dimensional construct was Miller (1983). In this research, a three-element definition including risk-taking, innovativeness and proactiveness is constructed. Rauch et al. (2009), find that much of research identifies EO with these three
dimensions (Grühn, et al. ,2017) (Covin et al.,1994). However, Miller’s (1983) three dimensions also provided a
starting point for elaboration of the EO concept. A cornerstone in literature building upon their establishment is the paper by Lumpkin and Dess (1996). Lumpkin and Dess (1996) defined EO as how a new entrance is undertaken. Using this concrete definition, Lumpkin and Dess were able to develop two additional salient dimensions of EO. As a result, they extended on the work of Miller (1983) by adding autonomy and competitive aggressiveness, creating a five-dimensional definition of EO (Lumpkin & Dess, 1996). Nonetheless,
suggest that dimensions of EO may occur in different combinations of elements and thus also influence firm performance differently. Building upon Lumpkin & Dess (1996), this research will deploy their five-element definition of EO. The concepts innovativeness, proactiveness, risk-taking, competitive aggressiveness and autonomy will be defined furtherly.
Innovativeness is described as the degree to which a firm explores the introduction of new products, services or processes. Proactiveness is described as ‘opportunity-seeking, forward-looking perspective
characterized by the introduction of new products and services ahead of the competition and acting in anticipation of future demand’ (Rauch et al., 2009, P.763). This extends the idea of ‘just’ innovation (e.g. new
products introduction) to the commercialization into markets. The dimension of risk-taking refers to discovering
the unknown (Engelen et al., 2015). Consistently, Anderson, Kreiser, Kuratko, Hornsby & Eshima (2015) define
the concept as the managerial approach towards risk and the consequent influence on firm-level strategic
decisions and actions. In addition to Rauch et al (2009), Anderson et. al (2015) elaborate further on competitive
aggressiveness and autonomy. Competitive aggressiveness describes the efforts firms exhibit towards taking out competitors. Prevalent in this dimension is the aggressiveness with which firms respond to rivals (Rauch, et al., 2009). Autonomy refers to lower level employees and middle level employees that independently express their EO-related attitudes and behaviours (Wales, et al., 2011).
Additionally, Covin, et al. (2020) suggest that EO provides firms adaptability, specifically in managing external changes in the environment and internal pressures of inertia. EO can facilitate the renewal of a firm’s internal activities and organization. Consequently, creating and maintaining high levels of entrepreneurial orientation endeavours innovation and enables the introduction of new products, processes or business models.
Resulting from this, EO is found to lead to higher overall performance (Rauch, et al., 2009). This is furtherly
substantiated by Richard et al. (2009) who also plead for a positive influence of EO on firm performance.
2.2 CEO succession and tenure
A CEO succession can have two origins. Firstly, the succession can be an outcome of reduced
organizational performance (Krause, Semadeni and Cannella,2014). The turnover of the previous CEO is then often involuntary and the new executive is appointed with the mandate to initiate changes. On the other hand, a turnover can be voluntary. CEO’s that are of certain age or reach the end of their predefined tenure are to be replaced (Behr & Fehre, 2019).
Previous research (Grühn, et al. ,2017; Boling et al.,2016) not only looked at the single event of the
succession, but also explored the first years of one’s tenure after the succession. In order to better capture the
Following the CEO cycle theory (Grühn et al., 2017), the first wave of changes occurs immediately after
the succession, often related to the new CEO’s specific going-in mandate. Grühn et al. (2017) found that changes
in EO happened primarily in the first three years of the CEO’s tenure. Subsequently, after the first three years the
absolute EO change was stabilizing (Grühn et al., 2017). The findings of Boling et al. (2016) also suggests most
change in EO is initiated in the initial phase. They argue that after a CEO’s tenure, the new leaders often engage in bold and risk-taking behaviours. This leads to the implementation of changes within the firm. Nonetheless Boling et al. (2016) nuance this as they argue that the early-stage entrepreneurial attitude may presumably be limited by social capital and firm knowledge restraints. As a consequence, they predict changes within the first 15 years. This is conflicting with the three years indicated by Grühn,et al. (2017). Inconsistency thus exists with regard to number of years after a succession in which EO changes take place.
In previous research (Behr & Fehre, 2019), a CEO succession is regularly linked to positive firm
performance by the CEO’s Commitment Status Quo (CSQ). Entering the firm, the new CEO shows lower levels of CSQ. They feel unaccountable and unattached to the previous firm strategy. Their psychological and
emotional commitment to the strategy is accordingly lower compared to the previous CEO, who pursued and invested in the strategy. Consequently, with his or her entrance the new CEO often initiates strategic change. The adapting of the firm’s strategic route enables the new CEO to implement changes according to his or her
personality. As a result, this will increase the commitment to the status quo (Behr & Fehre, 2019).
2.3 CEO narcissism
Narcissus, a young man who fell in love with his own reflection. The illustration of this the great self-admiration in the Greek mythology gave rise to the formulation of the concept ‘Narcissism’ (Ellis, 1898). Often seen as a clinical disorder, narcissism was later defined as a personality trait rather than a personality disorder (Chatterjee, Hambrik, 2007). According to Zu & Chen (2014) narcissism ‘is associated with focus,
self-admiration, a sense of entitlement and a sense of superiority.’ CEO’s are often associated with the possession of
narcissistic characteristics. In companies, they occupy prominent and powerful positions and accordingly are treated as the highest and most beloved person in the firm. Consequently, self-overvaluation arises and lies at the origin of narcissistic behaviour of CEO’s (Rijsenbelt, Commandeur, 2013). This is further supported by Kets de Vries who states: "anyone who hopes to rise to the top of an organization should have a solid dose of
narcissism" (2004, p. 188).
Kashmiri et al. (2017) categorize three narcissistic traits, namely overconfidence, the need for praise and dominance orientation. Overconfidence is a cognitive dimension that describes one’s feeling of superiority and uniqueness. This inflated self-view results in overconfidence in one’s decisions and behaviours. The need for praise is characterized as a motivational dimension which describes the desire of one’s inflated self-views to be reinforced amongst others. In line with this, Chatterjee & Pollock (2017) find that CEO’s are in high need of public acclaim of their leadership compared to non-narcissistic CEO’s.
dominancy over the team. However, not all members are willing to adhere to the narcissistic CEO (Chatterjee & Pollock, 2017).
Moreover, CEO narcissism has implications for firms. Their overconfidence tends them to underestimate the risks concerned with new introductions. Therefore, they will steer the firm’s resources more towards
introducing new product introductions or new markets (Kashmiri et al., 2017). Additionally, in the narcissistic
CEO’s efforts to be validated within the firm, they tend to involve in daring innovations. Consequently, CEO’s will direct the firm’s resources and efforts into high-risk and high-return projects (Wales, Patel &
Lumpkin,2013).
2.4 Hypotheses development
2.4.1 CEO Succession and EO As elaborated on before, to increase the CEO’s CSQ, the new executive will initiate strategic change (Behr & Fehre, 2019). Grühn et al. (2017) moreover suggest that such a strategic change in management can consequently trigger a change in EO. The cause of the change in EO followed by a succession can substantiated by three factors of which one is especially prevalent for this paper.
Firstly, personal characteristics of the new CEO are vital in changing the EO. The upper echelons theory states that new CEO’s possess different characteristics than the previous CEO. Strategic choices, amongst which entrepreneurially oriented decisions, will likely be in line with the new CEO’s personality (Hambrick & Mason, 1984). A CEO possessing more confidence than his or her predecessor will likely initiate innovative activities. On the other hand, a CEO who possesses less confidence than his or he predecessor may be inclined to be more preservative in engaging in entrepreneurial behaviour (Engelen et al., 2015). Additionally, the new CEO’s may have previous entrepreneurial experiences and network ties outside the firm (Grühn et al., 2017). These
experiences influence one’s view towards entrepreneurial orientation. As a consequence, a succession in CEO is likely to cause a change in EO as there are new different personality characteristics, experiences in EO projects and outside network ties (Grühn, et al. ,2017). The change in EO can be positive and negative as someone’s personality can favour or disfavour entrepreneurial orientated behaviour, which may be a result of the presence of a prevention focus. Such a focus tends the CEO to desire stability and security over advancement and growth (Gamache et al., 2015).
Secondly, a specific going-in mandate can be commissioned to the new CEO. Krause et al. (2014), found that a succession was often found to be the outcome of poor organizational performance. Usually the succession is involuntary as strategic change is required for the firm’s survival. Hence, a specific going-in mandate will initiate changes in activities and strategy that likely affect the firm’s entrepreneurial pathway (Grühn, et al.,2017). It is nonetheless undefined how this affects EO. The mandate can also oblige the new CEO to take a step back on exploring new opportunities. Lacking performance may signal the firm cut back on innovation efforts as these efforts require large resource devotion (Pisano, 2015). Consequently, the direction of the change in EO caused by mandate can be either positive or negative.
theory, the changes they make are in line with their personalities and personal preferences(Hambrick & Mason,
1984). A high CEO CSQ exists when one beliefs in the powerfulness of the current firm strategy and leaves little
space for adjustment, lower CEO CSQ consequently opts for new strategic positioning (Behr & Fehre, 2019).
CEO’s entering the firm with a low commitment to the current strategy are thus expected to pledge for strategic change. They want to mark their territory and commit themselves to a strategy that they themselves can
psychologically and emotionally invest in. These changes the new CEO implements in order to increase the CSQ and mark one’s territory are often entrepreneurial of nature (Behr & Fehre, 2019). Investing in a new strategic route is the start of creating their legacy. Considering a new leader’s aim to gain respect and set foot towards a legacy, the new leader likely exhibits bold actions with the goal of achieving an early track record (Grühn et al., 2017) By means of implementing these bold actions to mark territory and increase the CSQ, the change in EO will therefore positively affect EO.
Consequently, there are three main mechanisms namely personality characteristics, mandate and territory marking substantiating a change in EO. Personal characteristics and mandate do not suggest a directional change in EO, whereas territory marking leads to a positive directional change in EO. As territory marking is regarded as the main mechanism in this research, I therefore hypothesize the following:
H1a: A CEO succession leads to an increase in the firm-level Entrepreneurial Orientation.
2.4.2 The Moderating Role of CEO Narcissism As specified above, Grühn et al. (2017) place absolute change in EO as a result of CEO replacement. Territory marking is prevalent in this relationship due to its great overlap with narcissistic traits. Thus, I will ground the influence of the narcissistic personality in the mechanism of territory marking. New CEO’s mark his or her territory by initiating strategic change involving bold actions (Behr & Fehre, 2019) (Grühn et al.2017). These changes are daring and the implementation of these
changes requires a certain dose of CEO confidence. Narcissistic CEO’s often possess the kind of traits needed to
implement these changes as they are characterized by the need for praise, overconfidence and a dominance orientation (Rijsenbelt, Commandeur, 2013). I will elaborate further on the strengthening effect of these three narcissistic traits on territory marking.
Firstly, narcissistic CEO often desire the praise of others. They aspire that their ideas are admired by
employees within the firm and constantly seek the praise of others (Kashmiri et al., 2017).Entering the firm, a
new CEO makes bold moves with regard to their business activities. Their foremost goal of this is to show determination and gain respect. This is especially attractive to narcissists who aim to gain respect and praise from others (Grühn et al.,2017). Territory marking activities facilitates this need of narcissists to gain respect from others. Therefore, narcissistic CEO’s aim to gain praise of others will positively influence the relationship between a CEO succession and change in EO.
Secondly, overconfidence is another trait that characterizes narcissistic CEOs. Overconfident CEO’s
recognize and act upon new opportunities in uncertain markets far ahead of others (Engelen et al., 2015). As new
CEO’s enter the firm, they will mark their territories and adjust the current territory in line with their personality (Grühn et al.,2017). CEO’s with narcissistic personalities and a proper dose of overconfidence tend to change the territory and shape it by moving the firm into uncertain markets. Identifying and acting upon these uncertain
outcomes with regard to their proposed actions. This optimism encourages entrepreneurial behaviour amongst
employees and moreover enthuses them (Engelen et al., 2015). In short, overconfident narcissistic CEO’s
proactively shape their territories when they enter the firm by engaging in the uncertain opportunities. Furthermore, in their ability to take their employees along, successive CEO’s with narcissistic traits will positively influence the change in EO. Nonetheless, the concept of autonomy should also be taken in
consideration. As elaborated on before, employees can autonomously exhibit entrepreneurial behaviour (Wales, et al., 2011). Consequently, they can also actively choose to not engage in entrepreneurial behaviour and stick to an entrenched way of working (Covin et al. 2020).
Thirdly, the narcissistic characteristic dominance orientation is characterized by a strong desire towards superiority. It is defined as the joy in engaging in competition in order to show dominancy over their colleagues.
Additionally, they place themselves in situations where they are able to dominate others (Kashmiri et al., 2017).
Narcissistic CEO’s tend to include executives in their TMT that they can control (Chatterjee & Pollock, 2017).
This ambition to gain a strong position in the firm overlaps territory marking behaviour as newly appointed CEO’s are especially keen achieving political foothold in the firm (Grühn et al.,2017). The dominance of the new narcissistic CEO s leads to the avoidance of organizational inertia in the TMT. Consequently, quick decisions can be made which is beneficial for entrepreneurial behaviour as entrepreneurial moves are to be made quick. In short, newly appointed narcissist leaders tend to reduce discussions and dominate the decision-making in the TMT which enables quick response to entrepreneurial opportunities and consequently increases the firm’s EO. However, it should be taken in consideration that in a TMT the lack of narcissism amongst other team members may create a balance. In addition, not all team members are willing to be dominated by the CEO (Wales, Patel & Lumpkin, 2013).
As previously mentioned territory marking and narcissism are highly overlapping behaviours.
Furthermore, the pursuance of territory marking requires a certain dose of narcissistic behaviour. The entrance of a new executive with narcissistic characteristics will therefore strengthen the relationship between a succession and change in EO. This relationship is specifically strengthened by the three narcissistic traits of need for praise, overconfidence and dominance orientation.
Therefore, I expect CEO narcissism to positively moderate the relationship between CEO succession and EO
H2: CEO narcissism moderates the positive relationship between a CEO succession and the firm’s EO,
where the relationship is stronger for a CEO possessing narcissistic traits than for one not possessing these.
3. Methodology
3.1 Sample & Data Collection
The sample in this research is composed of companies from the Standard & Poor’s (S&P) 100 companies, which is a subgroup of the S&P 500 list. The S&P 100 are the largest 100 firms of the S&P 500 US public stock market firms. The use of this secondary data source has two main reasons. Firstly, using publicly listed
companies of the S&P 100 list ensures the availability and accessibility of annual reports, company and CEO characteristics. Secondly, the companies in the S&P 100 list are dispersed over various industries which makes the sample diverse and the research results thus more widely applicable rather than pertinent to one specific industry. This research’ specific sample covers the timeframe between 2008 and 2018 and consists of 150 companies in total. Over this sample’s timeframe of 10 years, new additions but also deletions of firms on the S&P 100 list appeared. Three firms went bankrupt or were acquired by other firms in 2008 and were
consequently removed from the sample. Additionally, new firms emerged in the top 100 and thus more firms than a total of 100 firms were included over the span of ten years.
In this research, all data is of secondary and quantitative nature. Various databases were deployed to collect information about the firms of the sample. These databases will be discussed in the following.
First, the Software for Repository for Accounting and Finance was deployed. From this I retrieved the Stage One 10-K annual report parsed data used for the measure of the dependent variable measure
entrepreneurial orientation. These annual reports are mandatory and filed by the publicly-traded companies, reporting on their financial performance.The files were parsed by Bill Mcdonald and include the management discussions parts which serve as input for the measurement of EO (Loughran & McDonald, 2016).
Secondly, the ExecuComp database was used to collect data for the independent variable CEO succession. Additionally, this database was utilised for measuring the control variables, firm size, CEO age, CEO gender and ‘Past performance’.
Thirdly, the Thomson Reuters DataStream database was used to collect transcripts of quarterly earnings calls of the sample firms for the measurement of the moderating variable CEO Narcissism.
Lastly, Google was consulted in order to gather the control variable age of the firm, by googling the year of origin of each firm.
After collecting all the datasets and importing them into STATA, I merged all the different data sets to create one panel data set. After running the final regression, the sample included 123 Companies and 838 observations, for the robustness measure using the CEO picture the sample is reduced to 573 observations and 94 companies.
In the following, I will elaborate on the measures and datasets used for each of the variables. Various databases were deployed for the gathering of this research’ data
3.2 Measurements
In this next section, the measurements of the variables will be discussed. Firstly, the data management and due diligence of the variables will be discussed. Thereafter I will elaborate further on the measurements the variables.
Firstly, I checked the distribution of the continuous variables, in order to ensure the symmetry of the measurements.
Secondly, I monitored the outliers as they can cause biased results. Single extreme values may influence the results for the whole sample (Stevens, 1984). In case of extreme observations, I used the winsorization technique. I winsorized the dependent variable Entrepreneurial Orientation, the moderating variable CEO narcissism and the control variables CEO age, Firm size and preceding sales. The variables were winsorized without significantly reducing the span of the observations (Dixon, Yuen,1974). In this process, the outliers above and under the set percentiles were consequently replaced by the set percentiles. Further elaboration on the winsorization per each variable will be provided in the description of the respective variables.
Thirdly, I checked with a scatterplot whether there was a linear connection between each of the independent variables with the dependent variable. This was done to ensure the presence of a causal relationship between the independent variables and dependent variable, needed to test the hypotheses. Additionally, I used the lag of each of my control variables, except for the year and industry dummies, and my independent and moderator variables in order to control for endogeneity(Bellemare, Masaki & Pepinsky, 2017).
3.2.2 Independent variable: CEO succession The independent variable CEO succession was captured by using the ExecuComp database’s data on CEO tenure. Following previous research (Grühn et al.,2017), the first three years of a CEO tenure are expected to show and indicate strategic change and thus change in the firm’s EO related to one’s succession. The first three years are regarded as the most vivid time period of a new CEO’s tenure in bringing about changes after his or her succession. Especially in order to capture the actual effects of a succession, a time span of these three years should be included (Grühn et al.,2017). Therefore, the CEO tenure indicating one to three years is labelled as a succession of CEO. A tenure of more than three years is consequently not labelled as a succession. I created a dummy variable for the CEO succession. The variable takes the value ‘1’ for a tenure of one to three years and any other year is rated with the value of ‘0’.
3.2.3 Dependent variable: Entrepreneurial Orientation The dependent variable Entrepreneurial
Orientation is measured with help of computer aided text analysis (CATA) analysis of data from the 10-K
financial reports from 2008-2018.
The text analysis LIWC program was used because such a CATA program is less prone to mistakes and considered a more objective measure compared to human coders. Furthermore, CATA enables a reliable analysis of large sets of data This method of analysis requires the use of dictionaries measuring the concept of EO. The dictionaries used originate from Short et al. (2010) (Appendix A). EO is constructed as one aggregate measure of all five measures of EO, namely innovativeness, risk-taking, proactiveness, autonomy and competitive
aggressiveness (Lumpkin and Dess, 1996). The five-element dictionary by Short et al. (2010) was developed using a deductive approach. In their paper, Short et al. (2010) address the external validity, dimensionality and predictive validity of the dictionaries in order to reassure the quality of the words in the dictionaries.
(Loughran & McDonald, 2016). The 10-k files include management discussions parts which serve as unit of analysis for the measurement of EO. For the purpose of reducing the amount of data I used the fourth quarter of each year per firm as representative for that year. I filed the fourth quarter 10-X reports per subsequent year. In the collected yearly files the firms of this research’s sample can be traced back by using the firm’s TIC.
The next step was the analysis of the data in LIWC. The relative percentages of each of the separate five dictionaries were generated by LIWC and thereafter EO was measured as the aggregate of the percentages of the
five dictionaries. Additionally, in order to control for outliers, I conducted a winsorization at 98th percentile on
the EO variable (Dixon, Yuen,1974).
3.2.4 Moderating variable: CEO Narcissism The data for the moderator CEO narcissism was collected by using the Thomson Reuter Datastream database. In previous research the concept of narcissism was most often measured using the Narcissistic Personality Inventory test (Raskin, Hall, 1988). Given the scope of this thesis, the primary data collection necessary for the NPI index was not feasible. Additionally, response rates to questionnaires were likely to be low. As a result, a more accessible measure of CEO narcissism needed to be
deployed. Accordingly, I use two alternative measures based upon Raskin & hall (1988) and Chatterjee &
hambrick (2007).
Firstly, we build upon the measure of Raskin and Shaw (1988), who notion that the proportion of first-person singular pronouns used in speech is correlated with the NPI score of narcissism. A requisite is that the speech is unplanned and spontaneous. As a result, the transcripts of quarterly earnings calls were most suitable.
The Thomas Reuters DataStream database was used to collect transcripts of quarterly earnings calls of the sample firms. Thereafter, this data was analysed by deploying CATA using the software LIWC. In LIWC the number of first-person singular pronouns were counted, which includes the words ‘I, me, mine, my, myself’. The CEO narcissism measure was then generated by dividing the count of first-person singular pronouns by the use of first-person singular pronouns as well as first-person plural pronouns. The final measure was created as the mean of the four quarters. The main disadvantage of this measurement is the transcripts include much ‘noise’.
Usage of first-person singular and plural pronouns is also counted when the interviewer uses these words and therefore the measure does not only capture the word-use of the CEO. I corrected for the extreme observations in
the CEO narcissism measure by winsorizing the variable at the 0th and 98th percentile.
Secondly, the CEO photograph in the annual reports was used as measure of CEO narcissism for the robustness analysis. The size of their picture is a reflection of the CEO’s position and status within the firm.
Furthermore, the CEO is highly controlling over how his/her picture is depicted in the annual report. This
was depicted alone on less than half a page and four points were assigned when the CEO was depicted alone on more than half a page.
3.2.5 Control variables In this following sector, I will elaborate on the control variables of the research. Firstly, I will focus on the CEO control variables. Thereafter the firm-level variables will be discussed and finally additional control variables are outlined.
3.2.5.1 CEO control variables CEO age
CEO age is a determining factor in his/her view towards risk-taking activities. With the increasing of age CEO’s tend to show more conservative behaviour. Younger CEO’s on the other hand follow riskier strategies (Hambrick & Mason, 1984). Accordingly, older CEO’s invest less in R&D (Serfling, 2014).
Consequently, I control for CEO age as it alters and determines the CEO’s risk-taking actions and decisions with regard to R&D. In order to control for outliers, I winsorized CEO age set at the 1st and 99th percentile (Dixon,
Yuen,1974). After conducting a curve fit estimation procedure, the results guided me in using this variable as continuous variable and not as the natural logarithm (Liu, 2013).
CEO Gender
The gender of the CEO is another variable that is controlled for. Previous research builds upon the notion that men engage in entrepreneurial activity differently than women. According to Lim & Envick (2013), in four out of the five EO dimensions, risk-taking behaviour, innovativeness autonomy and competitive aggressiveness, men score higher than women. The behaviour of men thus tends to support the entrepreneurial orientation of the firm to a greater extent than the actions of women. This categorical variable takes value ‘1’ when the CEO is female and the value ‘0’ when the CEO is male.
CEO power
A CEO’s power is measured by CEO duality. Occupying both the chairman function in the board and
position as CEO increases the CEO’s power and locus of control (Boyd, 1995). As a result, the CEO can use this
power to implement strategies that are according to his/her preference which consequently influences the firm’s entrepreneurial orientation. CEO structural power was measured by assessing one’s duality. This categorical variable takes the value of ‘1’ when the CEO is also chairman of the board and takes the value of ‘0’ when he or she is not concurrently chairman of the board.
3.2.5.2 Firm control variables Firm age
Firstly, grounded in organizational ecology literature I tend to control for firm age. As firms are ageing, resistance to change grows. With the aging of firms, entrenchment in organizational routines increases. This
affects the level of innovation. (Sirén, Hakala, Wincent & Grichnik, 2017). After conducting a curve fit
Firm size
Secondly, firm size impedes renewal in firms. Larger firms have to deal with enhanced complexity due to their size. Furthermore, even though large firms have many resources and capabilities available for EO, radical innovations still face resistance. This is due to the fact that radical ideas challenge and sometimes even question
the established resource (Sirén et al., 2017). Firm size was constructed as the natural logarithm of the number of
employees in order to correct for skewness.
Additionally, I winsorized firm size at 1th percentile. The outliers above and under the 1st and 99 are replaced by
the 1th and 99th percentile (Dixon, Yuen,1974). This was done in order to reduce sensitivity to extreme values
within the observations.
3.2.5.3 Additional control variables Personal characteristics & Mandate
In previous research Grühn et.al (2017) find three main mechanisms that trigger changes in EO, namely personal characteristics, mandate and territory marking. In this research territory marking is used as the main mechanism to explain the change in EO after a CEO succession. Consequently, I will try to isolate the
mechanism of territory marking by controlling for personal characteristics and mandate. Personal characteristics are defined by Klein, Lim, Saltz and Mayer (2004) as demographic characteristics, values, and personality. Most previous research in the upper echelon field focused on the observable demographic attributes of the executives such as age and education. These observable attitudes are consequently used as proxies for the unobservable psychological constructs cognitive base and values. Carpenter, Geletkanycz,& Sanders, (2014) suggest that not only demographics, but personality is moreover essential in measuring values and cognitive base (Wales,Patel & Lumpkin, 2013). However, taking in consideration the scope of this research and feasibility, I will only be able to observe demographic characteristics. In this research, these are the CEO’s age and gender which are included in CEO level control variables.
The mechanism of mandate can partly be controlled for. After a succession, a clear going-in mandate is often the outcome of poor organizational performance (Krauze, Semadeni, Cannella, 2014). Consequently, the weak performance is followed by a mandate to adapt the strategy. Thus, by including a variable to measure the firm’s past growth, mandate is partially controlled for. The firm’s sales growth in the preceding year is used to measure this (Grühn et al., 2017). Nonetheless, it should be taken in consideration that this variable does not fully capture the effect of mandate as mandate is not always a consequence of previous firm performance. I used the natural logarithm to control for the skewness and created a lagged version of the variable in order to capture
the sales of the preceding year. Additionally, I performed a winsorization set at the 2nd percentile in order to
exclude extreme values (Dixon, Yuen, 1974).
Year and industry dummies
3.3 Analytical Method
The sample in this research consists of panel data, also known as longitudinal data. For the companies in this research’ sample various variables such as CEO narcissism are observed over the time period 2008-2018. There are various advantages of using panel data of such as the ability to study dynamic relationships and model differences amongst the companies (Frees, 2004). Furthermore, the outcomes of this research are to a greater extent generalizable over time and thus a pattern can be identified (Grühn et al, 2017). The panel data is unbalanced as there are missing data points.
In order to examine this panel data set, I will use the statistical software program STATA. I will conduct a stepwise hierarchical regression analysis in order to create different models. By creating different models, the effect seperate variables can be identified in a stepwise manner. Panel data combines cross-sectional and time series data. Consequently, there is a need to account for cross section time heterogeneity (Frees, 2004) Two main models could be used to deal with this heterogeneity namely the fixed effect (FE) regression model and the random effects (RE) regression model (Pillai, 2016). The fixed effects model, assumes the unobservable individual effect is a random variable that is correlated to the explanatory, independent, variables. This
regression analysis corrects for the effects of the unobserved variable bias and could therefore provide less biased results with regard to the explanatory power of the observed independent variables on the dependent variable. In addition, this method measures the within-firm variation and does not take in account the effects of independent variables that are time-invariant. The random effects model does not only measure this within-firm variation but also measures firm variation. The random effects regression analysis does measure the effect of time-invariant independent variables but does not correct for the unobserved effects. This research’ main explanatory variable is CEO succession. Over the ten-year time span of this research, CEO successions do not occur yearly, as there are also many long-tenured CEO’s. The value of CEO succession is thus rather stable within firms over time. The fixed model may therefore not obtain significant results and effects due to the small within variance in this explanatory variable of CEO succession. Additionally, psychological traits such as CEO narcissism stay rather constant over time, therefore such a time-invariant variable would not accurately be captured by the fixed effects model. Moreover, in an attempt to explore the second hypothesis, the regression
model should enable a comparison across CEO’s and firms. As a result of the above reasoning, random effects
model is therefore more suitable in this research’ setting. (Pillai, 2016) (Bell, Fairbrother & Jones 2019). In addition, I performed a Hausman test (Hausman & Taylor, 1981) to confirm my initial theoretical argument for using the random effects approach. The null hypothesis stated the random effects model should be used instead of the fixed effects model. The P-value of 0.1789 was not significant and therefore the null
hypothesis was not rejected. This confirms the use of the random effects model. In conclusion, I will therefore
4. Results
4.1 Descriptive Statistics and Correlations
In the following section the results of this research will be discussed. Firstly, the descriptive statistics and correlations will be presented. Thereafter the outcomes of the regression analysis for the testing of the
hypotheses will be elaborated on and finally the robustness check will be discussed.
In table 1 and table 2 (Appendix B) the summary of the descriptive statistics and correlations are depicted. The mean, standard deviations, and minimum and maximum values are represented. In table 1, the summary statistics with regard to the main model are displayed. In table 2, the statistics of the robustness model with the robust measure of CEO narcissism, CEO picture, are presented. The final sample consists of 123 firms and 838 observations. The total number of observations for the robustness check is 573 from 94 firms. Taking in consideration these different sample sizes, the statistics are depicted in separate tables. From the descriptive statistics in table 1, I can derive that in less than half of the observations CEO successions occurred (mean -= 0.359, with 1 indicating a succession). Furthermore, I found that the approximate mean age of the CEO is 57 (mean= 56,733) years. Additionally, there is a significantly larger amount of CEO males compared to females (mean= 0.094, with 1 meaning female). Also in more than half of the instances the CEO is also chairman of the board and thus CEO duality exists (mean=0.653). In the correlations presented in table 1 the correlation between sales and firm size (R=0.835) is found to be very high, exceeding the benchmark of 0.5 (Hemphill, 2003). This can be explained by the fact that larger firms usually have a larger amount of sales and in some researches sales is even used as measurement of the firm’s size. In table 2 similar results are presented, the addition is the correlation of the CEO picture with the other variables.
Additionally, I also performed a Variance Inflation Factor (VIF) test in STATA to check the degrees of multicollinearity. High multicollinearity could make the coefficient unstable and inflate the standard errors for these coefficients. In a simple OLS regression analysis I included the independent variable, dependent variable, interaction effect, control variables and year and industry dummies. CEO succession and the interaction effect of CEO succession and CEO narcissism are highly correlated, with VIF values of 20.91 and 21.24 respectively. These scores exceed the benchmark of 10 (Cohen, Cohen, West, & Aiken, 2003). Consequently, I standardized CEO succession and CEO narcissism by rescaling them with a z-score. As a result, the VIF scores reduced to scores under 1.5 each. The descriptive statistics of the Z-scored variables are not included in tables 1 and 2 as the VIF test was executed afterwards and makes interpretation of the categorical variable CEO succession harder. In the regression analysis, the z-scores will be used in order to control for the multicollinearity. In addition, a RVF plot was used to check the distribution of the residuals. The plot showed a homoscedastic and unbiased
4.2 Regression Results and Hypotheses Testing
This research’ hypotheses will be tested using a random effects model. Results are reported in table 3. In the following I will first present the main findings with regard to the hypotheses, where after I will discuss each of the models in the regression separately.
Hypothesis 1 is predicted in model 2. This model suggest that a CEO succession is negatively and significantly related to EO. The variable CEO succession takes the value of 1 in case of a succession. Taking the value of 1, the level of EO decreases by 0.016 (β = -0.016 and p= 0.010). As this finding is significant, therefore it is assumed that when there is a CEO succession, the firm-level EO is significantly decreases. This presents a conflicting result with the hypothesized positive effect in hypothesis 1.
The results for hypothesis 2 are presented in models 3 and 4. It is found that the moderation effect is negative, the findings are insignificant. Consequently, there is no evidence for the hypothesized positive moderation effect displayed in hypothesis 2.
In model 1 the dependent variable EO, the lagged dummy variables and the industry and year effects are included. The R-squared represents the goodness-of-fit of the model. The statistic indicates which proportion of the variance in the dependent variable is explained by the model’s independent variables. Using a random-effects model, the within, between and overall r-squared are considered. In the first model the within R is 0.0257 the between R is 0.0508, the overall R is 0.0639. Only the variable CEO power shows a significant result (β = 0.027,p= 0.061). From this, we can expect that when there is CEO duality, variable thus takes the value of 1, the level of EO increases with 0.027 keeping all other variables constant.
The second model is used to test the first hypothesis. It builds upon model 1 by adding the independent variable CEO succession. The z-score (standardized value) and lag of this variable is used. The values of R are 0.0389, 0.0471 and 0.0730 for the within, between and overall effects of the models respectively. The overall fit of the model thus increased from 0.0639 to 0.0730. However, this still does not explain much of the variation in the dependent variable EO by the model’s independent variables. Furthermore, CEO power is again significant (β =0.030**, p=0.041), but now at a 5% level instead of a 10% level. I expected the coefficient of CEO succession to be positive. A CEO succession was expected to lead to an increase in firm level EO. Nonetheless, the coefficient is negative (-0.016, p=0,010) and it is significant at a 5% level. The model therefore does not provide empirical support for the first hypothesis and consequently suggest the opposite direction of the effect.
In model 3, the second model is expanded by including the variable CEO narcissism. For this variable, the created z-score (standardized) and lagged value is used. The variable is added, but not yet as interaction effect. Consequently, the variable is presented as an explanatory variable of the dependent variable EO. The coefficient of CEO narcissism is negative and insignificant (-0.003, p=0.717). The main independent variable CEO Table 1: Descriptive Statistics and Correlation
Mean S.D. Min. Max. 8. 9.
1.Entrepreneurial Orientation 0.754 0.173 0.140 1.240 1.000 2.CEO Succession 0.359 0.480 0.000 1.000 -0.034 1.000 3.CEO Narcissism 0.218 0.052 0.038 0.458 0.051 -0.017 1.000 4.CEO Age 56.733 5.684 40.000 73.000 -0.082 ** -0.333 *** -0.099 *** 1.000 5.Preceding Sales 10.104 1.009 7.297 12.362 0.033 0.009 0.060 * 0.042 1.000 6.Firm Size 3.836 1.300 0.000 6.339 -0.009 -0.042 0.109 *** 0.068 ** 0.835 *** 1.000 7.CEO Power 0.653 0.476 0.000 1.000 0.078 ** 0.018 -0.059 * 0.060 * 0.044 0.057 * 1.000 8.CEO Gender 0.094 0.292 0.000 1.000 0.041 0.039 -0.026 -0.059 -0.064 * -0.059 * 0.089 *** 1.000 9.Firm Age 69.296 51.681 1.000 224.000 -0.076 ** 0.017 -0.174 *** 0.051 0.023 0.064 * 0.103 *** -0.024 1.000 N=838
Notes: Significance levels: ***p<0.01, **p<0.05, *p<0.1
7.
succession again shows a significant negative effect, however at a 1% significance level in this model, (=-0.016, p=0.010). Due to rounding off to three digits it may seem as if the significance levels in models 2 and 3 for CEO succession are the same nonetheless in model 3 the CEO succession is significant at 1% instead of 5%. The results remain the same with regard to the other variables, only CEO power exhibits a significant result. The within, between and overall R’s are 0.0390, 0.0483 and 0.0728 respectively. There is thus small decrease in overall R compared to model 2. There is however a small increase in the model’s explanatory power for the between firm effects.
Model 4 tests the second hypothesis and thus adds the interaction effect of the moderating variable CEO narcissism. The interaction effect shows a negative and insignificant result ( -0.003, p=0.489). I expected the coefficient to be positive and significant, therefore this model does not provide empirical evidence for hypothesis 2. The coefficient of the independent variable CEO succession remains negative and significant at a 5% percent significance level. The results remain the same with regard to the other variables, only CEO power exhibits a significant result. The within, between and overall R’s are 0.0401, 0.0472and 0.0711 respectively. The R’s remain rather constant, only the overall R is reduced slightly. The explanatory of the complete model is therefore still low. After adding the main variables in model 2 and models 3 and 4, the explanatory power did not
significantly increase. This means the models do have not much explanatory power.
4.3 Robustness analysis
5. Discussion and Conclusion
In this final chapter of this research paper the discussion and conclusion of the findings are debated. First, I will briefly summarize the results after which the possible theoretical and methodological reasons for my conflicting and/or insignificant results will be explained. Thereafter the theoretical and practical implications will be provided and lastly, I will name the limitations and opportunities for future research.
This study examined the influence newly appointed CEO’s had on the firm’s EO, and more specifically if this relationship was moderated by the presence of narcissistic personality traits. By analysing 123 companies from the S&P100 between 2008 and 2018, I find that in conflict with my first hypothesis there is a negative significant effect of CEO succession on EO. Additionally, I find no significant results for the second hypothesis that proposed a positive moderating effect of CEO narcissism on the relationship between CEO succession and EO.
The possible theoretical and methodological reasons for the lacking empirical evidence for my proposed hypotheses will be elaborated on in the following.
First, the findings of this study did not provide evidence for hypothesis 1 in which I hypothesized a positive effect of CEO succession on EO. Instead, a negative and significant effect of CEO succession on EO
Table 3: Regression Results Hypothesis 1 and Hypothesis 2
Variables Independent Variable . . . . . CEO Succession -0.016 ** -0.016 *** -0.016 ** -0.011 (0.006) (0.006) (0.006) (0.018) Moderator Variable CEO Narcissism -0.003 -0.003 (0.008) (0.008) Interaction Effect
CEO Narcissism x CEO Succession -0.003
(0.005) CEO Age -0.000 -0.002 -0.002 -0.002 -0.003 ** (0.001) (0.001) (0.001) (0.001) (0.001) Preceding Sales -0.012 -0.010 -0.011 -0.011 -0.007 (0.018) (0.018) (0.018) (0.018) (0.022) Firm Size -0.002 -0.003 -0.002 -0.002 0.002 (0.016) (0.016) (0.016) (0.016) (0.019) CEO Power 0.027 * 0.030 ** 0.030 ** 0.030 ** 0.027 * (0.014) (0.015) (0.015) (0.015) (0.016) CEO Gender -0.006 -0.003 -0.003 -0.004 0.023 (0.034) (0.035) (0.036) (0.036) (0.028) Firm Age -0.0002 -0.0002 -0.0002 -0.0002 -0.0003 (0.000) (0.000) (0.000) (0.000) (0.000) CEOpicture -0.001 (0.008) Interaction Effect (Robustness)
CEOpicture x CEOsuccession -0.003 (0.006) Constant 0.897 *** 0.984 *** 0.987 *** 0.991 *** 0.962 *** (0.148) (0.149) (0.149) (0.150) (0.172) Observations 838 838 838 838 573 Number of firms 123 123 123 123 94 R2 Within 0.0257 0.0389 0.0390 0.0401 0.0556 R2 Between 0.0508 0.0471 0.0483 0.0472 0.0814 R2 Overall 0.0639 0.0730 0.0728 0.0711 0.128
Notes: Dependent Variable: Entrepreneurial Orientation
Robust standard errors in parentheses; Significance levels: *** p<0.01, ** p<0.05, * p<0.1. Model 2
was found. This finding is not in line with the findings of existent research by Grühn et al. (2017), who found a positive effect of a CEO succession on EO. A possible theoretical reason can be derived from the mechanisms that were expected to substantiate the CEO succession and EO relationship. In the hypothesis development, three mechanisms were discussed as essential in bringing about the change in EO, namely personality characteristics, mandate and territory marking. Based upon theoretical arguments (Behr & Fehre, 2019; Grühn et al., 2017), I expected a positive effect of territory marking on EO. However, I expected the effects of personality
characteristics and mandate on EO to be unidirectional, suggesting EO could increase or decrease as a
consequence of these two mechanisms. Consequently, the potential negative effects of these two mechanisms on EO could have possibly offset the potential positive effect of territory marking on EO, resulting into a decrease in the EO. Even though, mandate and personality were aimed to control for in this research, in the
methodological reasoning for the findings I will elaborate further on how they might have not adequately been controlled for. Mandate and personality characteristics can exhibit a negative influence on EO in the following ways. The new CEO’s personality can disfavour entrepreneurial behaviour due to a prevention focused
orientation (Gamache et al., 2015) or previous experiences that cause one to negatively evaluate innovation
(Grühn et al.,2017). In addition, mandate can oblige the new CEO to behave less entrepreneurially. Poor firm performance and reduced sales may signal the firm to take a step back and cut down on innovation expenses. Consequently, less may be invested in entrepreneurial activities as resources are scarcer (Pisano, 2015).
Moreover, an additional explanation for the negative significant effect can be derived from the study of
Boling et al. (2016). In the theoretical background, it was elaborated on that Boling and colleagues predicted effects from a CEO succession to show within the first 15 years of their tenure. Inconsistency exists with regard to the three years suggested by Grühn et. al (2017). Boling’s (2016) larger time frame is based upon the
theoretical argument that the early-stage entrepreneurial attitude may presumably be limited by social capital and firm knowledge restraints the new CEO faces. My study’s finding of significant negative effect may thus be related to the fact that the new CEO is still constrained by social capital and firm knowledge during the first three years. Possibly this could result in restraining the CEO from making bold and risky movements and
consequently reducing the EO, instead of increasing it.
The methodological reason for the conflicting result for hypothesis 1 flows from the previously named possible theoretical reasons. In this study, there is the control for the two unidirectional and possible negative mechanisms that influence EO is not adequate enough. Mandate and personality are aimed to be controlled for by including control variables for preceding sales and CEO age and CEO gender. Nonetheless, preceding sales does not fully control for mandate. Mandate is not necessarily arising from the firm’s preceding performance. Brand image or reputational reasons could also cause the firm to commission the CEO to follow a certain strategy (Grühn et al., 2017). In addition, CEO age and CEO gender only capture the demographic aspects of one’s personality which does not fully capture one’s complete personality (Carpenter et al., 2014). Consequently, the possible effects that possibly effect the EO negatively were not accurately enough controlled for.
Firstly, based upon theoretical arguments the expectation was that the dominancy of narcissistic CEO’s would strengthen the newly appointed CEO’s territory marking behaviour and consequently increase the EO. The overlap between dominancy and territory marking behaviour would result into a strengthening effect of a narcissistic personality on the relationships between CEO succession and level of EO. However, CEO’s cannot always dominate everyone. Often, they operate in TMT’s. Team members exhibiting less narcissistic behaviour may therefore offset behaviour and decisions of narcissistic CEO’s. Narcissistic behaviour could therefore be balanced out by fellow team members (Wales et al., 2013). Additionally, in the theoretical background, it was elaborated on that not all team members are willing to adhere to the narcissistic CEO (Chatterjee & Pollock, 2017). Consequently, narcissistic CEO’s may not be able to dominate and push through their bold actions in TMT’s. Accordingly, this could provide an explanation for the insignificant results with regard to the moderating role of narcissism.
Secondly, the second hypothesis was additionally substantiated by the overlap between narcissistic overconfident behaviour and territory marking. Narcissistic CEO’s overconfidence was used to explain the CEO’s influence on employees to act entrepreneurially. However, the theory in this paper also suggested that employees have the autonomy to act entrepreneurially or non-entrepreneurially without mandate from the CEO
(Wales, et al., 2011). As a result, employees can initiate entrepreneurial projects that are not in line with the firm’s strategy towards the EO. Moreover, they may be cognitively entrenched in prior firm actions and thus turn
away from entrepreneurial behaviour (Covin et al. 2020). Consequently, the predicted positive moderating effect
of CEO narcissism may be offset by the employee’s autonomy.
A methodological reason for lacking empirical evidence for hypothesis two may be found in the
accuracy of the measurement of CEO narcissism. This measurement builds upon the research of Raskin and
Shaw (1988), who propose the use of proportion of first-person singular pronouns. The main disadvantage of this measurement is that the transcripts include much ‘noise’. Usage of first-person singular and plural pronouns was also counted when the interviewer used these words and therefore the measure does not only capture the word-use of only the CEO. Consequently, this measurement may not have effectively and accurately measured the CEO’s level of narcissism. In the robustness model the CEO narcissism measurement was replaced by the CEO picture measurement. The explanatory power of the model significantly increased compared to the model using the first-person singular CEO narcissism measure. This may also hint at the inaccurate use of first-person singular pronouns as indicator of CEO narcissism
5. 1 Theoretical Implications
First, this research contributes to the EO literature domain. I attempt to do this by revisiting and
elaborating the research of Grühn et. al (2017) on the relationship between a CEO succession and the firm’s EO. Grühn and his colleagues lay out three substantiating mechanisms. The magnitude and direction of the
effort to explore the single influence of territory marking, instead of also measuring the influence of personality and mandate (Grühn et al.,2017). Additionally, my research may find more support in other studies. The results of this study may be more in line with research stating that effects of a CEO succession will show in 15 rather instead of three years (Boling et al., 2016). As well, this research’ findings find more explanation in studies
(Pisano, 2015; Gamache et al., 2015), suggesting that mandate and personality traits may have offsetting
negative influence on the positive influence of territory marking on EO. The conflicting results of this paper with previous research may alert scholars to furtherly discover the potential individual effects of various mechanisms on EO and, moreover, empirically test these. Therefore, this research provides a starting point for scholars in acknowledging and testing individual effects of the mechanisms on EO. Additionally, this research adds nuance to the existing research (Grühn et al., 2017) by trying to empirically account for the individual effects of mechanisms influencing the firm’s EO and presenting conflicting empirical results on the effect of a CEO succession on EO.
Second, this research aims to respond to the call of Grühn et al (2017) by including a CEO characteristic. This research contributes to the upper echelon literature by introducing the moderating relationship of a CEO narcissistic personality trait on the relationship between a CEO succession and EO. This is important because appointing a new CEO possessing narcissistic traits can a positive significant effect on the firm’s level of EO compared to appointing non-narcissistic CEO (e.g.Kashmiri et al. 2017).This was however not supported by the findings of this paper. Consequently, this research is inconsistent with research that found that narcissistic CEO’s overconfidence in exhibiting bold behaviour positively influences the EO (Kashmiri et al. 2017; Engelen et al., 2015). Rather, this study is more in line with the notion that narcissistic CEO’s cannot influence everyone in their TMT (Wales et al., 2013; Chatterjee & Pollock, 2017) and research (Covin et al., 2020) that states
employees can act autonomy entrepreneurially. Scholars can therefore furtherly explore the moderating effect of CEO narcissism in order to reduce the inconsistency present in the upper echelon literature.
5.2 Practical Implications
First, his research concludes that there may be a negative effect of a CEO succession on EO which is conflicting with previous research (Grühn et al.,2017). A conclusion from this paper is that practioners should critically explore whether a CEO succession in their firm follows the pathway outlined by Grühn et al. (2017) or this paper, in influencing their firm’s EO. Practioners may base this on how they define EO, which is differently measured in this paper than in the study by Grühn et al. (2017). Consequently, this paper provides practioners with knowledge on potential changes in EO outcomes when introducing a new CEO, which enables them to prepare the firm for such changes.
5.3 Limitations and Future Research
The study also has several limitations, which concurrently pave the way for possible future research. There are four limitations that will be discussed in the following.
First, personality characteristics, mandate and territory marking are theoretically substantiating the effect of a CEO succession on EO. In this study personality characteristics and mandate are attempted to be controlled for. However, the control variables preceding sales and CEO age and CEO gender do not sufficiently and accurately control for their effects. As previously elaborated on in the discussion, mandate is not only arising from the firm’s preceding performance captured by preceding sales Grühn et al. (2017) and moreover, CEO age and CEO gender only capture the demographic aspects of one’s personality (Carpenter et al., 2014). Future research could therefore aim to control mandate by using annual reports or newspapers that state the CEO’s going-in mandate. Furthermore, future studies could more accurately control for personal characteristics by examining the deeper layers of one’s personality with help of primary data collection, which exceeded the scope of this research.
Secondly, the time span of this data is a limitation. The panel data set was set from 2008-2018. This research dives into strategic changes with regard to EO. However, strategic changes often require more time to demonstrate effects. The consequences of a CEO succession may not show off as fast as within the proposed three years by Grühn et al. (2017), but may need more time such as the time span of 15 years as suggested by
Boling et al. (2016). They base their argument on the CEO’s initial limited social capital and firm knowledge restraints. Future research could therefore follow the research of Boling et al. (2016) using a time-span of 15 years to explore possible positive and significant effects of a CEO succession on EO.
Thirdly, as mentioned in the discussion, the measurement of CEO narcissism may not measure CEO
narcissism accurately. The measurement used builds upon the research of Raskin and Shaw (1988), which main
disadvantage is the large proportion of ‘noise’. In addition, the robustness model that included the CEO picture as measure of CEO narcissism, exhibited a significant higher r-squared and thus explanatory power of the model. This may also hint at the inappropriate and inaccurate measurement of CEO narcissism as relative use of
first-person singular pronouns to first-first-person plural pronoun.The use of the CEO picture may thus be used in future
research, however, the disadvantage is the amount of lacking data. Other measurements of CEO narcissism may therefore be explored in the future.
