Let’s gamble! The polarization of narcissists
The effect of CEO narcissism on firm risk and the moderating role of CFO narcissism
Abstract
Personal characteristics of top managers can influence the output of a firm. Following the upper echelon theory, this research gives insights into the effect of the narcissistic trait on the decision-making process of the CEO and CFO. It explores the relationship between CEO narcissism and firm risk and the moderating role of a narcissistic CFO. A content analysis of 14.601 earnings and investor
calls ranging from 2012 - 2017 gives interesting insights into the narcissistic traits of the top management couple. The dataset consists of 325 firms of the S&P 500 and was analysed using a random effects panel regression. The results indicate a highly significant positive relationship between
CEO narcissism and firm risk and contrary as predicted a weak negative moderating effect of CFO narcissism. It showed that CFO narcissism has only an effect on CEO narcissism if CEO narcissism is
low. Moreover, the content analysis of earnings and investor calls offers a new promising source of data for further research on personal characteristics. However, as the dependent variable is not robust,
the results have to be interpreted carefully.
“Narcissism is voluntary blindness, an agreement not to look beneath the surface.” Sam Keen
June 22, 2020
Philipp Stadlmayr | S4084284
Supervised by M. Weck, MMSc. | Co-assessed by P. Arque-Castells, PhD. MSc BA Strategic Innovation Management | Faculty of Economics and Business
University of Groningen Word count: 11985
Table of Contents
1. Introduction ... 1
2. Theory ... 3
2.1 Upper Echelon Theory ... 3
2.2 Narcissism Theory ... 4
2.3 Group Polarization Theory ... 5
3. Hypotheses development ... 6
3.1 CEO Narcissism and firm risk ... 6
3.2 CFO narcissism as moderator ... 8
4. Methodology ... 9
4.1 Data Sample ... 10
4.2 Analysis ... 14
5. Results ... 14
5.1 Descriptive statistics and pairwise correlation ... 14
5.2 Hypotheses test ... 15
5.3 Robustness test ... 16
5.4 Additional Analyses ... 17
6. Discussion ... 17
6.1 Implications for theory and research ... 19
6.2 Implications for practice ... 19
6.3 Limitations and future research ... 20
6.4 Conclusion ... 21
References ... 22
List of Tables
Table 1 - Variables and data sources ... 13
Table 2 - SIC Codes in sample ... 27
Table 3 - Descriptive statistics and pairwise correlations – coefficients ... 28
Table 4 - Panel regression models – Firm Risk (DAR) ... 29
Table 5 - Robustness check – Models with Firm risk (ICR) ... 30
List of Figures Figure 1 - Conceptual Framework ... 9
Figure 2 - Moderator effect of CFO narcissism ... 27
List of Abbreviations
CATA: Computer aided text analysis CEO: Chief executive manager CFO: Chief finance manager DAR: Debt to assets ratio
EBIT: Earnings before interest and expenses ICR: Interest coverage ratio
SQL: Structured query language UET: Upper echelon theory
1. Introduction
Western societies have become increasingly individualistic in recent decades (Hofstede, 2001). As a result, more and more narcissistic personalities are emerging who prioritise themselves over a group and focus on the "I" rather than the "we" (Foster et al., 2003). Moreover, narcissistic characteristics seem to be especially prevalent in leadership, affecting their surroundings and firm outcome (Chatterjee and Pollock, 2016). How these personality traits influence leadership, especially the head of the company, has not yet been sufficiently researched (Chatterjee and Hambrick, 2007).
Recent literature has linked the narcissism of chief executive officers (CEOs) to riskier strategies such as aggressive R&D spending (Chatterjee and Hambrick, 2007) or increased attitude towards gambling (Lakey et al., 2007). Besides, there is evidence that CEO narcissism is related to financial fraud (Rijsenbilt and Commandeur, 2013), stock market investing (Foster et al., 2011), paying larger acquisition premium (Chatterjee and Hambrick, 2011) or corporate tax avoidance (Olsen and Stekelberg, 2016). The narcissistic trait seems to increase the risk propensity of individuals, leading them to take higher stakes compared to non-narcissists.
Based on the upper echelon theory (UET) by Hambrick and Mason (1984), this research tries to explain how the narcissistic trait influences the decision-making process of the individual and respectively increases the risk of the firm. The UET theory argues that the characteristics of the top managers are reflected in organizational behaviour. Demographic, as well as personal characteristics, act as some sort of lens through which the individual analyses its decisions. Thus, if an executive has an overly risky behaviour caused by narcissism, it should also be visible in risk measurements of the firm. To detect the relationship between CEO narcissism and firm risk, the below research question is formulated:
RQ1: How does the degree of CEO narcissism influence firm risk?
moderating role of the CFO in the relationship between CEO narcissism and firm risk. The second research question is as follows.
RQ2: How does the CFO narcissism influence the relationship between CEO narcissism and firm risk?
To answer the research questions, two hypotheses are developed. These are tested in a longitudinal dataset covering a time period from 2011 to 2017, containing 325 companies and 1525 company years. The narcissism variable is created by applying a content analysis of earnings and investor calls (Belderbos et al., 2017). These calls have a special structure. First, a summary of the financial results of the firm is given. This is followed by a question and answer part, where analysts can ask questions to the top management team. In these parts, the managers have to give spontaneous answers to questions of financial analysts. Hence, it should be more difficult to hide their personality traits compared to the presented summary at the beginning of these calls and thus lead to a stronger measurement of the variable compared to other methods. Both financial and demographic data are collected from the Compustat, Boardex and Datastream databases.
The main results provide evidence that there is a significant positive relationship between CEO narcissism and firm risk. It indicates that if CEO narcissism is high, the risk of a firm is also high. Contradictory to my expectations, the moderating role of CFO narcissism shows a negative, weakly significant effect on CEO narcissism and firm risk. CFO narcissism seems to increase the company's risk only when CEO narcissism is low and CFO narcissism is high. Accordingly, an expected polarization between two narcissists could not be observed.
executives as well as for potential investors in the firm, as high firm risk leads to higher credit ratings or even bankruptcy (Verwijmeren and Derwall, 2010). This study aims to give a better understanding of how narcissism influences the decision-making process of CEO and CFO and consequently increases the risk of the firm.
The thesis is structured as follows: First, the theoretical part including the literature review as well as the conceptual framework is given to elaborate the conceptual core terms used in this study. Next, testable hypotheses are presented, and a conceptual model is designed. The hypotheses should answer the two main research questions mentioned above. Following the methodological part, the intended research design and methods, which mainly focus on literature research, is introduced. Subsequently, the statistical results are presented in relation to the stated hypotheses. Fourth, the interpretation of the regression and its implications for researchers and practitioners are discussed. Finally, limitations are identified, as well as future research directions are drawn.
2. Theory
This section contains the literature review, the development of the hypotheses and the conceptual framework. The literature review contains explanations of the most important terms used in this thesis and its theoretical background, which at the same time leads to the hypothesis development. Two main hypotheses are elaborated, which form the conceptual framework for the relationship between CEO narcissism and firm risk.
2.1 Upper Echelon Theory
According to upper echelon theory (UET), characteristics of top managers are able to influence the strategic orientation of the firm and accordingly its financial performance (Hambrick and Mason, 1984). The actions influencing strategic orientation are defined as large, long-term undertakings at firm level that require a considerable deployment of resources. They are expensive, complex, use a large number of resources and associated with a high amount of risk (Connelly et al., 2010).
due to limited cognitive and temporal resources. The brain then picks not the optimal but the best solution regarding the situation, highly influenced by the personal characteristics of the individual.
According to the UET, these interpretations of top managers directly influence their behaviour and consequently the company’s decision-making process. The theory, therefore, sees the organization as a reflection of the characteristics of its top managers (Hambrick and Mason, 1984). In the existing literature, evidence is found for the effect of demographic characteristics of a manager such as gender (Khan et al., 2013), tenure (Orens and Reheul, 2013) or age (Hambrick and Mason, 1984) on output variables of the firm such as firm performance. Most of the research focuses on the demographic characteristics of managers. Chatterjee and Hambrick (2007) point out that cognitive data of the top-level managers are more difficult to retrieve than demographic data. This may be because managers are often unwilling to participate in surveys that examine their personality, or that they are trained to give socially acceptable answers and thus hide their underlying psychological characteristics.
Yet, merely looking at the demographic variables may provide an incomplete picture of the decision-making effect. Hambrick and Mason (1984) argue that research should also focus on the personality traits of top management. These traits can play an even more critical role in influencing the perception of the situation and the resulting decisions. Hence, they can help to complete the picture of the effects underlying the upper echelon theory. The character trait narcissism has been identified by several types of research as a personality dimension that has a high influence on the decisions of top management teams (Gerstner et al., 2013). For example, narcissism is related to the emergence of leadership (Gaddis et al., 2015), which does not mean that all top managers are narcissists, but that some variation of narcissistic tendencies can be expected.
2.2 Narcissism Theory
The condition of narcissism describes a clinical condition of self-love and was first defined in 1898 by Havelock Ellis. The concept was taken up and specified by the psychoanalyst Sigmund Freud. According to Sigmund Freud (1914), every human being carries a certain amount of narcissism in different forms from birth on. Freud (1914) and Otto Rank (1911) further pointed out that narcissism can be seen as a protective mechanism of the individual, guarding against undesirable emotions such as a lack of self-esteem or self- confidence.
appear dominant, overbearing, self-confident or exhibitionistic. They can have a charming side although they are exploitative and envious (Levy et al., 2011). In contrast, individuals with the second type are more insecure, defensive, hypersensitive to the evaluation of others and fearful (Britton, 2000). Both types have in common to satisfy their own needs at the expense of others and have unrealistically high expectations of themselves. (Levy et al., 2011; Ronningstam, 2014).
Several studies have proven that narcissism is a stable personality trait (Cragun et. al, 2019). However, the focus was mainly on the overt features of narcissism and does not capture the vulnerable site of the trait. Individuals who reveal their narcissistic traits show this through exaggerated self-perception, increased self-esteem, empathy, and dominance (Morf and Rhodewalt, 2001; Resick et al., 2009). Narcissists are accused of manipulating others (Capalbo et al., 2018), fraud (Rijsenbilt and Commandeur, 2013), bullying (Regnaud, 2014), manipulate policies (Buyl et al., 2019; Olsen et al., 2014) and to be overconfident in their behaviour (Campbell et al., 2004, Hribar and Yang, 2016). The lack of self-esteem drives narcissists to constantly look for attention and to get affirmation from their peers. In this way, they are constantly scanning the environment for possible opportunities to satisfy this need (Carlson et al., 2011). Narcissists tend to blame others for their own failures (John and Robins, 1994) and are resistant to criticism (Kernis and Sun, 1994). Other studies showed that narcissism is an important leadership trait due to the pursuit of power and prestige and the tendencies to manipulate (Lubit, 2002). Narcissism is also part of the dark triad which is associated with a callous-manipulative interpersonal leadership style (Paulhus and Williams, 2011) and positively related to risky behaviours (Crysel et al., 2013).
2.3 Group Polarization Theory
Group polarization is a phenomenon that arises within group discussions. The main effect is that members of a group move to a more extreme point compared to their previous standpoint through a discussion. This effect is even more pronounced within people who are already prone to extremes and are able to confidently defend them (Sunstein, 2002; Stoner, 1968).
others. Hence, people change their previously created opinions in order not to damage their self-image in front of others. In order to not be seen as weak or cautious and to appear more experienced, they begin to take a more extreme attitude than they originally had (Posner, 1997; Blascovich and Ginsburg, 1974).
The second explanation deals with persuasive argument theory of Burnstein and Vinokur (1977). The standpoint with the most convincing arguments will probably be the preferred opinion of the group and the individual will adjust his position accordingly. If a group tends in one direction, the dominant opinion will have the largest number of arguments supporting that opinion. Furthermore, the opinion of the individual will change even more in the direction of the dominant opinion, as it is assumed that the entire pool of arguments is limited (Sunstein, 2002). Thus, if an opinion has more compelling arguments than other opinions in the discussion, it's more likely that this opinion will be accepted by the majority of the group.
3. Hypotheses development
3.1 CEO Narcissism and firm risk
Previous research indicates that the CEO has the largest influence and power within the firm (Quigley and Hambrick, 2015) and that the CEO’s personality traits influence the decisions they make (Chatterjee and Hambrick, 2007; Hiller and Hambrick, 2005; Judge et al., 2006). This research is based on upper echelon theory, developed by Hambrick and Mason in 1984. As the theory implies that personal demographic and psychological characteristics of the top management influence their decision-making process. The process consists of three parts: the field of vision, the selective perception, and the interpretation of the gathered information (Hambrick, 2007).
feedback, not all possible options of a decision can be detected. Decisions made by a narcissist, therefore, have a higher risk of failure as the information on which the decision is based is incomplete due to the inability of accepting feedback by others.
Furthermore, the narcissistic CEO does not see any possible negative effects of his decisions but instead blames others in case of failure (Campbell et al., 2000). On the other hand, a narcissistic CEO believes that positive events only happen due to his ingenious thinking and acting (Kashmiri et al., 2017). Because of the resistance to feedback and the inability to think outside the box, narcissism weakens also the CEO’s selective perception. Possible risks are underestimated and one's own abilities are overestimated.
Studies have shown that the personality of a narcissistic individual is perceived as emotionless and cold (Morf and Rhodewalt, 2008). They find it difficult to empathize with other people in order to understand their different ways of thinking and views (Patel and Cooper, 2014). Narcissists do not care about the collateral damage of their actions and so act without consideration for others. This increases their willingness to take risks, as they have no bad conscience (O’Reilly et al., 2014) and believe that the consequences of their actions are free of costs (Lobuts and Pennewill, 1986).
Narcissists have a constant demand for attention (Wales et al., 2013). They have an excessive desire to be the centre of attraction and an increased need for admiration. To satisfy this demand, they make decisions that are visible to others and give them the desired attention (Carlson et al., 2011). The higher the risk of this decision, the higher the expected reward can be if the investment is successful. However, as the narcissist is so confident of himself and does not associate failure with his decisions, he will not perceive the potential risk as such (Chatterjee and Hambrick, 2011). As a result, even higher risks will be taken in future decisions, to satisfy the constant need for attention and admiration.
In summary, I argue that the personality trait of narcissism has a high negative influence on the decision-making process of the CEO and hence on the risk of the firm. Narcissism leads the CEO to weaken his or her environment and possible control levels, which would prevent the CEO from taking wrong decisions, which in turn raise the risk of the firm (Chatterjee and Pollock, 2017). In addition, the narcissistic CEO is immune to feedback (Campbell et al., 2000), with the result that the CEO is unable to learn from negative events, which could improve the ability to detect wrong decisions. Besides, their blurred perception of reality will lead to an increase of bold and risky actions (Campbell et al., 2004) since an assessment of what is right and what is wrong no longer exists. That is why I hypothesise:
3.2 CFO narcissism as moderator
The exchange of opinions within a group can lead to a process of shifting one's own position towards that of the group. These new positions are more extreme than the original individual opinion, caused by the discussion within the group. The effect is called group polarization. The theory is based on two mechanisms, namely social influence on behaviour and the pool of limited arguments as explained in the theory section (Burnstein and Vinokur, 1977; Sanders and Baron, 1977). Both mechanisms have an influence on the shift in the decision and thus the decision-making process of an individual. The decision-making process is based on three parts, the field of vision, the selective perception, and the interpretation of the gathered information (Hambrick, 2007).
Narcissists have a desire to dominate others and to be permanently the centre of attention (Campbell et al., 2004). They will try to use discussions with other members of the top management team as a platform for their self-expression. In addition, they will strive to present their ideas and concepts in a novel and pompous way to make an impression on others and to underline their wisdom and glory. In order not to appear cautious or weak in front of others, the group members will excel the dominant opinion of the narcissist and as a result shift to a more extreme view (Sunstein, 2002). This behaviour will create an information bubble, as one's own opinions may not be fully exposed to the group and thus relevant information representing a different view is not shared. The narcissist will get a feeling of superiority, as he does not receive critical feedback to his opinions (Chatterjee and Pollock, 2017). Moreover, the dominant appearance and the convincing nature of the narcissist will convince other participants of the group to align with his opinion. Hence, the field of vision will be biased, as other group members will not fully expose their own thoughts to the group.
The more homogeneous the group is, i.e. the more similar the individual group members are to each other in, for example, personal characteristics, the more likely the effect of group polarization will occur (Sunstein, 2002). So, their field of vision may be rather limited compared to a more heterogeneous group. Furthermore, as arguments will align over time due to the homogeneity of the individuals, the selective perception of the group members may be biased due to a lack of an outside view. In addition, a shift in the group will be even greater if a member of the group takes a more extreme view, such as a narcissist (Sunstein, 2002).
decisions. Hence, the CFO can be seen as some sort of control mechanism within the top management team, supporting the CEO in his decisions but also taking care of the financials of the firm. However, if the CFO is narcissistic, he will seek recognition and attention and be more likely to neglect potential risks to achieve his goals (Campbell et al., 2004). He will feel familiar with the narcissistic CEO as they share the same values and perceptions and thus support the extreme tendencies of the CEO within group discussions. This, in turn, will increase the total pool of arguments in favour of the opinion of the CEO. Hence, the probability is increased that the group will even more tilt towards the dominant opinion of the CEO.
In summary, I argue that narcissism has an influence on group polarization, and this has an influence on the decision-making process of the individual. Narcissistic CEOs present their ideas with exaggerated self-confidence and dominance. In order to not be seen as weak and cautious in front of the group, the other members of the group will shift to a more extreme (the narcissistic CEOs) opinion as they had previous to the group discussion. The more homogenous the group is, the more likely this effect will occur. Hence, if the CFO has similar traits, he will share similar values and thus support the extreme views of the CEO instead of preventing it. Accordingly, the second hypothesis is as follows
H2: The relationship between CEO narcissism and firm risk is moderated by CFO narcissism.
The conceptual model is in this research is as follows: Figure 1 - Conceptual Framework
4. Methodology
For this study a mixed methodology approach was used. Starting with a computer-aided text analysis (CATA) (Belderbos et al., 2017; Short et al., 2010) to derive the degree of the narcissism of the top management team from the transcribed earnings and investor calls. The method allows searching for specific single or multiple word constructs within a text to derive the frequency of these constructs. The data was enriched with financial and company-specific
CEO Narcissism CEO Narcissism
CFO Narcissism
data from Datastream as well as demographic data of the top management team from BoardEx and Compustat. The transcribed earnings and investor calls were derived from Seeking Alpha. The gathered data were analysed with several programs. First, I used an SQL script for the CATA analysis to search the investor transcripts for keywords applying the dictionary of Raskin and Shaw (1988). Furthermore, I used a self-written Python code to derive the year, quarter, and ticker symbol from the titles of the transcripts. Third, the statistical Software STATA was used to combine and analyse the datasets.
4.1 Data Sample
The dataset consists of S&P 500 firms, listed from the year 2011 till 2017 in the S&P 500. For this sample, 14.601 earning calls and conference calls were downloaded from Seeking Alpha. The S&P 500 presents well-established firms, headquartered in the United States. Hofstede (2001) states that the United States is one of the most idiosyncratic countries in the world. Consequently, on closer analysis, a high degree of narcissistic tendencies can be expected. Furthermore, the use of the S&P 500 index prevents possible cultural bias, which derives from cultural variation (Douglas, 1978) and is thus suitable for this research.
To retrieve the independent variable narcissism, I used the question and answer part of the investor and earning calls. The dataset of spoken words is particularly well suited to look for personality traits. Managers do not know in advance the questions asked by analysts. Consequently, no adequate preparations can be made, compared to the first part of the investor calls where the managers present their results. Accordingly, as they have to give spontaneous, non-prepared answers, it is more difficult to hide their true personality (Larcker and Zakolyukina, 2012). Furthermore, compared to a survey-based method, the text analysis method does not have a strong response and non-response bias. As in the case of a survey, respondents can manipulate the results by giving wrong answers as they, for example, want to hide their personality or do not want to participate at all (Duncan and Kalton, 1987).
regressions. This led to a final sample of 325 (58,5%) companies and 1.525 (42,5%) company years. To test if there is a significant difference between the removed and not removed data of the CEO and CFO, I applied a Kolmogorov-Smirnov two-sample test (Westphal and Bednar, 2005). The test showed no significant difference between the two samples in any of the used terms of gender, age, tenure, board size, share of women on board, employees, Tobin’s Qt-1, narcissism index etc.
The average age of the CEOs is 57 years (sd=5,45) and for the CFOs 52,6 years (sd=5,45). Furthermore, in terms of gender, males are overrepresented compared to women. The CEOs are primarily male (95,3%), as well as the CFOs (87,3%). Moreover, there are more women within CFOS (mean = 0.13) than within CEOs (mean = 0.05). 93% of the firms are listed each year in the S&P 500 index. As displayed in table 2, the majority of firms in the sample are distributed in the manufacturing industry, followed by the finance and the services industry.
Dependent Variable. The dependent variable firm risk is measured by the total debt to assets ratio (DAR). The equation gives an indicator of how many assets of a firm are financed through debts. A high percentage indicates a high risk that the firm is not able to meet its obligations (Graves and Waddock, 1994). To reduce outliers that could influence the results of the regression, I winsorized each dependent and alternative dependent variable by 1% using the winsor2 command in STATA. The function makes it possible to reduce the borders of the sample with no loss of precious data points (Salkind, 2010). Moreover, the data for the calculation of the dependent and alternative dependent variables were gathered from Datastream.
𝑑𝑒𝑏𝑡 𝑡𝑜 𝑎𝑠𝑠𝑒𝑡𝑠 𝑟𝑎𝑡𝑖𝑜 (𝐷𝐴𝑅) = 𝑡𝑜𝑡𝑎𝑙 𝑑𝑒𝑏𝑡 𝑡𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠
To test for robustness, I used the interest coverage ratio as an alternative dependent variable. This variable shows how well a firm can pay its interests on the outstanding debts. The lower the value, the higher the risk of the firm not paying back their debts and thus going bankrupt (Greuning and Brajovic-Bratanovic, 2003).
𝑖𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑐𝑜𝑣𝑒𝑟𝑎𝑔𝑒 𝑟𝑎𝑡𝑖𝑜 (𝐼𝐶𝑅) = 𝐸𝐵𝐼𝑇
𝑖𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑒𝑥𝑝𝑒𝑛𝑠𝑒𝑠
T
counted how often a person used the singular and plural pronouns in a conversation and divided the results by the total number of words used in the whole conversation. The singular pronouns were then divided by the plural pronouns. A higher number of singular pronouns compared to plural pronouns give a tendency for a narcissistic trait.
To get the singular and plural word counts of each manager, I used an SQL script together with the keyword dictionary of Raskin and Shaw (1988). To test if the word count was successful and the script reliable, I controlled 10% of the transcripts by hand. The individual narcissism index (singular pronoun divided by plural pronoun) was calculated using the STATA and normalized to 500 words according to Krippendorf (2013). The CEOs have been identified by merging the Compustat and Boardex data to the sample. Therefore, I merged the data by using a key with the ticker symbol, year, last name and first letter of the first name of each manager. For the robustness tests, I used the distance of the CEO salary compared to the CFO salary as an alternative independent variable. This alternative measurement has already been used by Hambrick & D'Aveni (1992) to calculate CEO narcissism.
Moderator. The same method (Raskin and Shaw, 1988) as for CEO narcissism was used to calculate the moderator variable CFO narcissism. Thus, the sum of singular pronouns and plural pronouns was divided by the total amount of words used and then divided by each other. A high value indicates a high narcissistic trait. The gathered value was normalized to 500 words according to Krippendorff (2013). Afterwards, the CFO has been identified by their name using the Compustat and Boardex database.
employees per year has been used as firm-specific control variables. To take time-delayed effects into account, I added a one year lagged and logged value of firm performance (Tobin's Q(t-1)). The data sources and calculation of the used variables are displayed in table 1.
Environmental level-factor. Year and industry dummy variables are used to control for trends that occur over time. I used a four digit SIC Code as industry classification.
Table 1 - Variables and data sources
Variables Operationalization Source
Dependent variable
Firm risk (DAR) Total debt divided by total assets Datastream
Firm risk (ICR) EBIT divided by interest expenses Datastream
Independent variable
CEO narcissism (Wordcount)
Singular WC divided by the total amount of WC divided through plural WC divided by the total amount of WC
Investor transcripts
CEO narcissism (Compensation)
(Relative noncash pay + relative cash pay) / 2 Relative noncash pay = CEO options divided by CFO options
Relative cash pay = CEO compensation divided by CFO compensation
Compustat
Moderator variable
CFO narcissism Singular WC divided by the total amount of WC
divided through plural WC divided by the total amount of WC
Investor transcripts
Control variables
Firm size Logarithm of employees/year Datastream
Tobin’s Q t-1 Lagged value of logarithm Tobin's Q Datastream
CEO age Year of birth - Year Compustat/Boardex
CEO tenure Year of becoming manager - Year Compustat/Boardex
CEO sex Binary – 1: male, 0: female Compustat/Boardex
CFO age Year of birth minus Year Compustat/Boardex
CFO sex Binary – 1: male, 0: female Compustat/Boardex
Share of independent board Sum of independent directors on board divided by total board size
Boardex
Share of woman on board Sum of women on board divided
by total board size
Boardex
Board size Total amount of board members Boardex
Environmental-level factor
Time period Dummy variable for each year in the sample -
Industry Dummy variable for each Industry in the sample.
Four digits of the Standard Industrial Classification (SICCode)
4.2 Analysis
A panel data regression analysis was applied to test the hypothesis in this research. A panel regression is an efficient way to analyse longitudinal time-series data (Baltagi, 2013). The two most common approaches are fixed effects and random effects analysis. Therefore, a Hausmann test has been conducted, to test which of these two regressions is applicable for the sample. If the test is significant, I can reject the null hypothesis, that the error terms are not correlated with the independent variables, and accept the alternative hypothesis, that they do correlate (Greene, 2012). In the used sample, the Hausmann test was not significant and thus a random-effects model was applied. In this way, the random effects panel equation for firm risk and CEO narcissism is as follows:
Firm Riski,t = 𝛽0 + 𝛽1 ∗ CEOnarci,t + 𝛽2 ∗ CEO𝑎𝑔𝑒i,t + 𝛽3 ∗ CEOteni,t + 𝛽4 ∗ CEOsexi,t + 𝛽5 ∗ shareIndepi,t + 𝛽6 ∗ shareWomi,t + 𝛽7 ∗ BoardSizei,t + 𝛽8 ∗ FirmSizei,t +
𝛽9 ∗ Tobin’s Q,i,t-1 + 𝛴 𝑦𝑒𝑎𝑟 𝑑𝑢𝑚𝑚𝑖𝑒𝑠 + 𝛴 SIC 𝑑𝑢𝑚𝑚𝑖𝑒𝑠 + 𝜀𝑖
To test for moderator effect, that the narcissistic CFO influences the relationship between the narcissistic CEO and firm risk, I also used panel regression. As the first step, the CEO narcissism and CFO narcissism variable were standardized. Through standardization of a variable, different types of variables can be compared to contribute equally to the regression analysis (Bring, 1994). It further reduces the possibility of collinearity. The Hausmann test showed a p-value above 0.05, hence no significance. Accordingly, a random-effects panel regression was used to test the second hypothesis.
5. Results
In the following section the descriptive statistics are displayed, as well as the results from the panel regression and the correlation matrix. Furthermore, additional analysis has been conducted to support my arguments.
5.1 Descriptive statistics and pairwise correlation
---
Insert table 3 (Correlation & Descriptives) about here ---
The dependent variable firm risk (DAR) shows a positive correlation with one year lagged and logged Tobin’s Qt-1 (0.26), a negative correlation with board size (-0.07) and firm size (-0.14). The independent variable CEO narcissism positively correlates with CFO narcissism (0.25), Tobin’s Qt-1 (0.08), board size (0.05), CEO sex (0.08), CEO tenure (0.17) and share of independent board members (-0.10). CFO narcissism shows a significant positive correlation with CFO age (0.05) and CFO sex (0.08).
5.2 Hypotheses test
In table 4 the results of the random-effects model regression are displayed. Year and industry dummies are included in the model, but not reported. The table consists of four models. The first model includes the dependent variable firm risk (DAR) and the control variables. The second model test Hypotheses 1, the third model test for the moderator effect and in the fourth model I changed the calculation of the independent variable CEO narcissism, to test for robustness. In the second Model, the independent variable CEO narcissism (WC) is included. In Model 3 I added the moderator variable CFO narcissism as well as additional control variables CFO age and CFO gender.
As can be observed from Model 1, table 4, three control variables found to be significant, namely Tobin’s Qt-1 (β = 0.0362, p < .01), firm size (β = -0.0189, p < .01) and share of independent board members (β = 0.0914, p < .1). The model itself is significant with an adjusted R2 of 66.8.
In Model 2, I added the independent variable CEO narcissism to test the first hypothesis. The model shows a positive significant relationship between firm risk and CEO narcissism (β = 0.0342, p < .01). Therefore, Hypothesis 1, a positive relationship between firm risk and CEO narcissism is supported. Furthermore, three control variables have been significant, namely Tobin’s Qt-1 (β = 0.0371, p < .01), firm size (β = -0.0187, p < .01), share of independent board members (β = 0.1010, p < .1), with an adjusted R2 of 66.99.
narcissism is low. If both the CEO and CFO have a high narcissistic trait, there is no polarization effect that raises firm risk. Hypothesis 2 could therefore not be supported, as I predicted a positive moderation effect. Furthermore the model shows a significant relationship between Tobin’s Qt-1 and firm risk (β = 0.0373, p < .01), firm size (β = -0.0183, p < .01), share of independent board members (β = 0.0939, p < .1), with an adjusted R2 of 66.95.
To test if the variance of the estimated coefficients is influenced by multicollinearity, I used the variation inflation factor (VIF). The factor resulted in a mean value of 2.79 and is thus below the threshold value of 10, so no multicollinearity could be determined between the variables (Alin, 2010).
---
Insert table 4 (Panel regression models – Firm Risk (DAR)) about here ---
---
Insert Figure 2 - Moderator effect of CFO Figure 2 (Moderating effect of CFO narcissism) about here
---
5.3 Robustness test
---
Insert Table (Robustness check – Models with Firm risk (ICR) about here ---
5.4 Additional Analyses
To take a deeper look into the data set and to support my research, I conducted several additional statistical tests. First, I analysed if there are differences between the “structured part” (first part, summary) and the “unstructured part” (question & answers) of the earnings and investor calls. Therefore, I applied a two-sample t-test. This test can be used to check if there are differences in the means of two samples (Goulden, 1939). The output showed significant difference between the structured and unstructured part for CEO narcissism (mean structured = 0.12, mean unstructured = 0.39, p<0.00) as well as CFO narcissism (mean structured = 0.10, mean unstructured = 0.32, p<0.00). Furthermore I controlled if there are differences between men and women in the level of narcissism for CEOs (♀ mean = 0.32, ♂ mean =0.39, p<0.00) and CFOs
(♀ mean = 0.28, ♂ mean = 0.33, p<0.00). The output showed a significant difference of
narcissism between the gender in both CEO and CFO groups.
6. Discussion
Generally, this research seeks to answer the following research questions: How does the degree of CEO narcissism influence firm risk and how does the CFO narcissism influence the relationship between CEO narcissism and firm risk. The purpose of this work was to explain how CEO narcissism influences the decision-making process of the individual and thus the risk of the firm. Furthermore, this study aimed to contribute to the existing literature by identifying a new moderator of narcissism and firm risk. We recall that this research is based on the upper echelon theory, which identified the narcissistic trait as an underlying personality dimension, able to influence the outputs of a firm (Chatterjee and Hambrick, 2007). Moreover, narcissism occurs to be important to give new insights into management leadership, especially on the top management level (Zhu and Chen, 2014).
admiration drive the narcissists to take bold and risky actions, neglecting the risk of potential failure. Thus, decisions made based on the incomplete information set of the narcissistic CEO will have an impact on the firm, by raising the risk of the firm. These findings are in most parts coherent with existing literature (Campbell et al., 2004).
Second, I expected a positive moderating effect of CFO narcissism on CEO narcissism and firm risk. The results provide a weakly significant negative effect, contrary to the expectations. Therefore, the second hypothesis was rejected. CFO narcissism seems to only have an effect on CEO narcissism if CFO narcissism is high and CEO narcissism is low. However, if both measurements are high, there is no effect on firm risk. Previous research indicates that the CEO and the CFO have different incentives for financial reporting. In case of failure, the CFO faces a greater loss of reputation compared to the CEO (Mian, 2001) and bears the majority of the consequences (Feng et al., 2011). A loss of reputation is difficult for narcissists to accept, as it would danger their whole narcissistic construct. Therefore, narcissistic CFOs may have incentives to mitigate the risky ideas of CEOs if the stakes are too high instead of further pushing it. Also, having more than one narcissist in a group may lead to conflicting incentives, as both want to have the full attention and admiration of the others. Thus, they may try to denigrate each other's ideas and praise their own ideas. In order to reach a compromise, they may meet at the lowest common denominator and not push to a higher level. In the case of a narcissist, however, the lowest denominator probably already carries a high risk. The results showed that if narcissism is high for the CEO, it is also high for the CFO. Consequently, this research is in line with Chatterjee and Pollock (2017), who stated that narcissists want to have other narcissists in their inner circle.
investigate the different calculations of firm risk and the influence of the CEO on these measurements.
6.1 Implications for theory and research
Previous research (Campbell et al., 2004; Chatterjee and Hambrick, 2011; Lakey et al., 2008) has focused on narcissism and individual risk-taking. Yet, it has not linked narcissism directly to firm risk. This study added to existing theories by using a different approach and consequently by providing more insights into CEO narcissism and its effect on firm risk. In addition, a new moderating effect was identified, namely CFO narcissism, which influences the relationship between CEO narcissism and firm risk. Moreover, this study highlights the difficulties in using a suitable calculation of firm risk variables that result from a wide variety of possible measurements.
Additionally, this research contributes significantly to the research methodology by applying the content text analysis approach to identify the personal characteristics of individuals. The results show that there is a significant difference in the level of narcissism between the “structured” and “unstructured” part of the earnings and investor calls. So that the unstructured part has an average narcissistic value three times higher than the structured part. This is an essential point for future research with CATA analysis, as spontaneously spoken words seem to provide a stronger measure of a character trait. Lastly, the outcome of the narcissism variable also supports research of Ingersoll et al. (2019), as a significant difference in the level of narcissism between a woman and male CEOs as well as CFOs could be found in the additional analysis.
6.2 Implications for practice
that above a certain risk level, additional narcissistic individuals (CFO) within the top management team do not increase the risk level, but neither do they reduce it.
6.3 Limitations and future research
Although this study offers some interesting new insights into narcissism and firm risks, the results still need to be examined based on the limitations of the study. This, in turn, gives an outlook on future research on the selected topic.
First, regarding the narcissism variable, several data points have been lost due to a lack of data, especially at the CEO and CFO level. Apart from that, the question and answers part of earnings and investor calls to calculate the value of narcissism have their limitations. In the question and answer part, the analysts can directly ask the executives about their opinions on a certain topic. Thus, by answering a direct question, the possibility to use singular pronouns may be higher compared to a normal conversation, resulting in a bias of the narcissism variable. To overcome this, a dummy variable could be created that for example takes direct questions from analysts who start with "you" into account. This research observes just five years within the S&P 500, therefore future research could enlarge the observed timespan as well as taking a look at specific industries as the influence of a CEO may be lower in highly regulated industries such as the banking industry.
Second, several restrictions also apply to the dependent variable firm risk. Industries such as manufacturing, for example the service sector, have a different structure of debt to assets. Since about 40% of the companies in the sample are in the manufacturing industry, this could be the main reason for the effect of the regression. Nevertheless, I have addressed this issue by using the 4-digit SIC codes as dummy variable. Besides, it might be interesting to examine the differences in short-term and long-term debt. As narcissists may tend to attract attention in the short-term, and consequently affect short-term debts to a greater extent. Further research could shed light on other measures of firm risk, especially those that use variables that can be directly influenced by the CEO.
Third, the S&P 500 index only contains firms that were able to meet a certain condition. For this reason, a survival bias may exist in the sample (Elton et al., 1996). Firms with a severe risk of failure may not be represented in the data, as they have dropped out of the sample, e.g. due to a decline in their market share or bankruptcy. One way to overcome this bias could be to include in the sample all firms that were represented in the S&P 500 during the selected period and have dropped out of the sample but have not filed for bankruptcy.
that may influence risk-taking such as CEO power (Lewellyn and Muller‐Kahle, 2012), CEO stock holdings (Rogers, 2000) or firm cash holdings (Tong, 2012). Additionally, to check the reverse causality, the lagged variable firm performance could be extended to further years. By applying a one-year lagged variable, I partly addressed this issue in the research.
6.4 Conclusion
References
Alin, A. (2010). Multicollinearity. Wiley Interdisciplinary Reviews: Computational
Back, M. D., Schmukle, S. C., & Egloff, B. (2010). Why are narcissists so charming at first sight? decoding the narcissism-popularity link at zero acquaintance. Journal of
Personality and Social Psychology, 98(1), 132–45.
Baltagi, B. H. (2013). Econometric analysis of panel data (5th ed.). Wiley.
Barry, C. T., Kerig, P. K., Stellwagen, K. K., & Barry, T. D. (Eds.). (2011). Narcissism and machiavellianism in youth: implications for the development of adaptive and maladaptive behavior. American Psychological Association.
Belderbos, R., Grabowska, M., Leten, B., Kelchtermans, S., & Ugur, N. (2017). On the use of computer-aided text analysis in international business research. Global Strategy
Journal, 7(3), 312–331.
Blascovich, J., & Ginsburg, G. P. (1974). Risky shifts and gambling: what's at stake?
Proceedings of the Division of Personality and Society Psychology, 1(1), 246–248.
Bring, J. (1994). How to standardize regression coefficients. The American Statistician, 48(3), 209–213.
Britton, R. (2000). Hyper-subjectivity and hyper-objectivity in narcissistic disorders. Fort Da, 6(2), 53-64.
Burnstein, E., & Vinokur, A. (1977). Persuasive argumentation and social comparison as determinants of attitude polarization. Journal of Experimental Social Psychology, 13(4), 315–332.
Buyl, T., Boone, C., & Wade, J. B. (2019). Ceo narcissism, risk-taking, and resilience: an empirical analysis in u.s. commercial banks. Journal of Management, 45(4), 1372– 1400.
Cain, N. M., Pincus, A. L., & Ansell, E. B. (2008). Narcissism at the crossroads: phenotypic description of pathological narcissism across clinical theory, social/personality psychology, and psychiatric diagnosis. Clinical Psychology Review, 28(4), 638–56. Campbell, W. K., Bonacci, A. M., Shelton, J., Exline, J. J., & Bushman, B. J. (2004).
Psychological entitlement: Interpersonal consequences and validation of a new self-report measure. Journal of Personality Assessment, 83, 29–45.
Campbell, W. K., Reeder, G. D., Sedikides, C., & Elliot, A. J. (2000). Narcissism and comparative self-enhancement strategies. Journal of Research in Personality, 34,329– 347.
Capalbo, F., Frino, A., Lim, M. Y., Mollica, V., & Palumbo, R. (2018). The impact of ceo narcissism on earnings management. Abacus, 54(2), 210–226.
Carlson, E. N., Vazire, S., & Oltmanns, T. F. (2011). You probably think this paper's about you: narcissists' perceptions of their personality and reputation. Journal of Personality
Chatterjee, A., & Hambrick, D. C. (2007). It's all about me: narcissistic chief executive officers and their effects on company strategy and performance. Administrative
Science Quarterly, 52(3), 351–386.
Chatterjee, A., & Hambrick, D. C. (2011). Executive personality, capability cues, and risk taking: How narcissistic ceos react to their successes and stumbles. Administrative
Science Quarterly, 56(2), 202–237.
Chatterjee, A., & Pollock, T. G. (2017). Master of puppets: how narcissistic ceos construct their professional worlds. Academy of Management Review, 42(4), 703–725.
Connelly, B. L., Hoskisson, R. E., Tihanyi, L., & Certo, S. T. (2010). Ownership as a form of corporate governance. Journal of Management Studies, 47(8), 1561–1589.
Cragun, O. R., Olsen, K. J., & Wright, P. M. (2019). Making ceo narcissism research great: a review and meta-analysis of ceo narcissism. Journal of Management, (2019).
Crowe, M. L., Lynam, D. R., Campbell, W. K., & Miller, J. D. (2019). Exploring the structure of narcissism: Towards an integrated solution. Journal of Personality, 87, 1151–1169. Crysel, L. C., Crosier, B. S., & Webster, G. D. (2013). The dark triad and risk behavior.
Personality and Individual Differences, 54(1), 35–40.
Douglas, M. (1978). Cultural bias (Ser. Occasional paper (royal anthropological institute of great britain and ireland), no. 35). Royal Anthropological Institute.
Duncan, G. J., & Kalton, G. (1987). Issues of design and analysis of surveys across time.
International Statistical Review, 97-117.
Elton, E. J., Gruber, M. J., & Blake, C. R. (1996). Survivorship bias and mutual fund performance. The Review of Financial Studies, 9(4), 1097–1120.
Feng, M., Ge, W., Luo, S., & Shevlin, T. (2011). Why do cfos become involved in material accounting manipulations? Journal of Accounting and Economics, 51(1), 21–36.
Finance, 251–273.
Foster, J. D., Keith Campbell, W., & Twenge, J. M. (2003). Individual differences in narcissism: inflated self-views across the lifespan and around the world. Journal of
Research in Personality, 37(6), 469–486.
Foster, J. D., Reidy, D. E., Misra, T. A., & Goff, J. S. (2011). Narcissism and stock market investing: correlates and consequences of cocksure investing. Personality and
Individual Differences, 50(6), 816–821.
Freud S. (1914). On narcissism. In the Standard Edition of the Complete Psychological Works of Sigmund Freud (vol 14) (ed J Strachey): 67–102. Hogarth Press.
Gerstner, W.-C., König A., Enders, A., & Hambrick, D. C. (2013). Ceo narcissism,
audience engagement, and organizational adoption of technological discontinuities.
Administrative Science Quarterly, 58(2), 257–291.
Goulden, C. H. (1939). Methods of statistical analysis. J. Wiley & Sons, Inc. Graves, S. B., & Waddock, S. A. (1994). Institutional owners and corporate social
Greene, W. H. (2012). Econometric analysis (7th edition, international, Ser. The pearson series in economics). Pearson.
Greuning, H. van, & Bratanovic, S. (2003). Analyzing and managing banking risk: a
framework for assessing corporate governance and financial risk (2nd ed., Ser. World bank e-library). World Bank.
Grijalva, E., Harms, P. D., Newman, D. A., Gaddis, B. H., & Fraley, R. C. (2015). Narcissism and leadership: a meta-analytic review of linear and nonlinear relationships. Personnel
Psychology, 68(1), 1–47.
Hambrick, D. C. (2007). Upper echelons theory: an update. The Academy of Management
Review, 32(2), 334–343.
Hambrick, D. C., & D'Aveni, R. A. (1992). Top team deterioration as part of the downward spiral of large corporate bankruptcies. Management Science, 38(10), 1445-1466. Hambrick, D. C., & Mason, P. A. (1984). Upper echelons: the organization as a reflection of
its top managers. The Academy of Management Review, 9(2), 193–206.
Hiller, N. J., & Hambrick, D. C. (2005). Conceptualizing executive hubris: the role of (hyper-) core self-evaluations in strategic decision-making. Strategic Management Journal, 26(4), 297–319.
Hofstede, G. (2001). Culture's consequences: comparing values, behaviors, institutions, and organizations across nations (2nd [rev. and updated]). Sage.
Hogan, R., Raskin, R., & Fazzini, D. (1990). The dark side of charisma. In K. E. Clark & M. B. Clark (Eds.), Measures of leadership (p. 343–354). Leadership Library of America. Hribar, P., & Yang, H. (2016). Ceo overconfidence and management forecasting.
Contemporary Accounting Research, 33(1), 204–227.
Ingersoll, A. R., Glass, C., Cook, A., & Olsen, K. J. (2019). Power, status and expectations: how narcissism manifests among women ceos. Journal of Business Ethics, 158(4), 893–907.
Jeong, S.-H., & Harrison, D. A. (2017). Glass breaking, strategy making, and value creating: meta-analytic outcomes of women as ceos and tmt members. Academy of
Management Journal, 60(4), 1219–1252.
Jiang, X. J., Petroni, K. R., & Yanyan Wang, I. (2010). Cfos and ceos: who have the most influence on earnings management? Journal of Financial Economics, 96(3), 513–526. John, O. P., & Robins, R. W. (1994). Accuracy and bias in self-perception: individual
differences in self-enhancement and the role of narcissism. Journal of Personality and
Social Psychology, 66(1), 206–19.
Judge, T. A., LePine, J. A., & Rich, B. L. (2006). Loving yourself abundantly: relationship of the narcissistic personality to self-and other perceptions of workplace deviance, leadership, and task and contextual performance. Journal of Applied Psychology, 91(4), 762
Kashmiri, S., Nicol, C. D., & Arora, S. (2017). Me, myself, and i: influence of ceo narcissism on firms’ innovation strategy and the likelihood of product-harm crises. Journal of the
Kernis, M. H., & Sun, C.-R. (1994). Narcissism and reactions to interpersonal feedback.
Journal of Research in Personality, 28(1), 4–13.
Khan, W. A., & Vieito J. P. (2013). Ceo gender and firm performance. Journal of Economics
and Business, 67, 55–66.
Krippendorff, K. (2013). Content analysis: an introduction to its methodology (3rd ed.). Sage. Lakey, C. E., Rose, P., Campbell, W. K., & Goodie, A. S. (2008). Probing the link between
narcissism and gambling: the mediating role of judgment and decision-making biases.
Journal of Behavioral Decision Making, 21(2), 113–137.
Larcker, D., & Zakolyukina, A. (2012). Detecting deceptive discussions in conference calls: detecting deceptive discussions in conference calls. Journal of Accounting Research, 50(2), 495–540.
Levy, K. N., Ellison, W. D., & Reynoso, J. S. 2011. A historical review of narcissism and narcissistic personality. In W. K. Campbell & J.D. Miller (Eds.), The handbook of narcissism and narcissistic personality disorder: Theoretical approaches, empirical findings, and treatments: 3-13. Hoboken, NJ: Wiley.
Lewellyn, K. B., & Muller-Kahle, M. I. (2012). Ceo power and risk taking: evidence from the subprime lending industry. Corporate Governance: An International Review, 20(3), 289–307.
Lobuts, J. F., & Pennewill, C. L. (1986). Risk-free decision-making. Journal of Business
Ethics, 5(1), 29–37.
Lubit, R. (2002). The long-term organizational impact of destructively narcissistic managers.
The Academy of Management Executive (1993-2005), 16(1), 127–138.
Mian, S. (2001). On the choice and replacement of chief financial officers. Journal of
Financial Economics, 60(1), 143–175.
Morf, C. C., & Rhodewalt, F. (2001). Unraveling the paradoxes of narcissism: a dynamic self-regulatory processing model. Psychological Inquiry, 12(4), 177–196.
O’Reilly, C. A., Doerr, B., Caldwell, D. F., & Chatman, J. A. 2014. Narcissistic CEOs and executive compensation. The Leadership Quarterly, 25: 218-231.
Oehmichen, J., Braun, D., Wolff, M., & Yoshikawa, T. (2017). When elites forget their duties: the double-edged sword of prestigious directors on boards. Journal of
Management Studies -Oxford-, 54(7), 1050–1078.
Olsen, K. J., & Stekelberg, J. (2016). Ceo narcissism and corporate tax sheltering. Journal of
the American Taxation Association, 38(1), 1–22.
Olsen, K. J., Dworkis, K. K., & Young, S. M. (2014). Ceo narcissism and accounting: A picture of profits. Journal of Management Accounting Research, 26: 243-267. Orens, R., & Reheul, A.-M. (2013). Do ceo demographics explain cash holdings in smes?
European Management Journal, 31(6), 549–563.
Paulhus, D. L., & Williams, K. M. (2002). The dark triad of personality: narcissism,
machiavellianism, and psychopathy. Journal of Research in Personality, 36(6), 556– 563.
Posner, R. A. (1997). Social norms and the law: an economic approach. The American
Economic Review, 87(2), 365–369.
Quigley, T. J., & Hambrick, D. C. (2015). Has the "ceo effect" increased in recent decades? a new explanation for the great rise in america's attention to corporate leaders. Strategic
Management Journal, 36(6), 821–830.
Raskin, R. N., & Shaw, R. (1988). Narcissism and the use of personal pronouns. Journal of Personality, 56, 393–404.
Regnaud, D. A. 2014. The relationship between top leaders’ observed narcissistic behaviors and workplace bullying. Doctoral dissertation, Walden University, Minneapolis, MN. Resick, C. J., Whitman, D. S., Weingarden, S. M., & Hiller, N. J. (2009). The bright-side and
the dark-side of ceo personality: examining core self-evaluations, narcissism, transformational leadership, and strategic influence. The Journal of Applied
Psychology, 94(6), 1365–81.
Rijsenbilt, A., & Commandeur, H. 2013. Narcissus enters the courtroom: CEO narcissism and fraud. Journal of Business Ethics, 117: 413-429
Rogers, D. A. (2002). Does executive portfolio structure affect risk management? CEO risk-taking incentives and corporate derivatives usage. Journal of Banking & Finance, 26(2-3), 271-295.
Ronningstam, E. (2014). Beyond the diagnostic traits: A collaborative exploratory diagnostic process for dimensions and underpinnings of narcissistic personality disorder.
Personality Disorders: Theory, Research, and Treatment, 5(4), 434–438.
Salkind, N. J. (2010). Encyclopedia of research design. Sage.
Sanders, G. S., & Baron, R. S. (1977). Is social comparison irrelevant for producing choice shifts? Journal of Experimental Social Psychology, 13(4), 303–314.
Serfling, M., 2014, CEO age and the riskiness of corporate policies. Journal of Corporate Statistics, 2(3), 370-374.
Short, J. C., Broberg, J. C., Cogliser, C. C., & Brigham, K. H. (2010). Construct validation using computer-aided text analysis (CATA) an illustration using entrepreneurial orientation. Organizational Research Methods, 13(2), 320-347.
Stoner, J. A. (1968). Risky and cautious shifts in group decisions: The influence of widely held values. Journal of Experimental Social Psychology, 4(4), 442–459.
Sunstein, C. R. (2002). The law of group polarization. Journal of Political Philosophy, 10(2), 175–195.
Tong, Z. (2010). CEO risk incentives and corporate cash holdings. Journal of Business
Finance & Accounting, 37(9‐10), 1248-1280.
Wales, W. J., Patel, P. C., & Lumpkin, G. T. 2013. In pursuit of greatness: CEO narcissism, entrepreneurial orien-tation, and firm performance variance. Journal of Management
Studies, 50: 1041-1069.
Westphal, J. D., & Bednar, M. K. (2005). Pluralistic ignorance in corporate boards and firms' strategic persistence in response to low firm performance. Administrative Science
Quarterly, 50(2), 262–298.
Zhu, D. H., & Chen, G. (2015). Narcissism, director selection, and risk-taking spending: narcissism, director selection, and risk-taking spending. Strategic Management
Journal, 36(13), 2075–2098.
Appendix
Figure 2 - Moderator effect of CFO narcissism
Table 2 - SIC Codes in sample
Sic Codes Industry Count
1000 – 1499 Mining 104
2000 – 3999 Manufacturing 607 4000 – 4999 Tranportation, Communications,
Electric, Gas and Sanitary service 212 5000 – 5199 Wholesale Trades 41 5200 – 5999 Retail Trade 95 6000 – 6799 Finance, Insurance & Real Estate 284
7000 – 8999 Services 164 35 36 37 38 39 40 41
Low CEO Narcissism High CEO Narcissism
Table 3 - Descriptive statistics and pairwise correlations – coefficients
Variables Count Mean Std. Dev. (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)
(1) Firm Risk (DAR) 1525 00.29 00.16 -1.00
Table 4 - Panel regression models – Firm Risk (DAR)
Model 1 Model 2 Model 3 Model 4
Variables Firm Risk (DAR) Firm Risk (DAR) Firm Risk (DAR) Firm Risk (DAR)
Tobin’s Qt-1 (-0.04** (0.04** (0.04** (0.04** -(0.01) (0.11) (0.11) (0.10) Firm size (-0.02** -0.02** -0.02** -0.02** -(0.00) (-0.00) (0.00) (0.00) Board size (-0.02 (0.02 (0.02 (0.02 -(0.01) (0.19) (0.02) (0.02)
Share of independent on Board -(0.09+ (0.10+ (0.10* (0.09+
-(0.05) (0.05) (0.05) (0.05)
Share of women on board -(0.00 (0.00 (0.01 -0.00
-(0.03) (0.03) (0.03) (0.03) CEO sex (-0.02 -0.02 -0.02 -0.02 -(0.01) (0.01) (0.01) (0.01) CEO age -(0.00 -0.00 -0.00 (0.00 -(0.00) (0.00) (0.00) (0.00) CEO tenure (-0.00 -0.00+ -0.00 -0.00+ -(0.00) (0.00) (0.00) (0.00)
CEO narcissism (Wordcount) (0.03** (0.00** (0.01) (0.00) CFO narcissism (0.00 (0.00) Moderator effect -0.00+ (0.00) CFO sex -0.00 (0.00) CFO age (0.00 (0.00)
CEO narcissism (Salary) (0.00*
(0.00) Constant (0.37* (0.37* (0.38* (0.37* (0.15) (0.15) (0.15) (0.15) N 1525 1525 1525 1525 WaldChi2 726.93 738.41 737.03 731.19 (Prob > Chi2) 0.00 0.00 0.00 0.00 Adjusted R2 66.80 66.99 66.95 66.81
Notes: Standard Errors in parenthesis. Year and Industry dummies included but not reported. (DAR = Debt to assets ratio)
Table 5 - Robustness check – Models with Firm risk (ICR)
Model 1 Model 2 Model 3
Variables Firm Risk (ICR) Firm Risk (ICR) Firm Risk (ICR)
Tobin’s Qt-1 31.98*** 31.96*** 31.97*** (3.84) (3.85) (3.85) Firm size 3.77* 3.77* 3.77* (1.71) (1.72) (1.72) Board size -3.02 -3.05 -2.79 (5.86) (5.87) (5.89)
Share of independent on Board -15.34 -15.51 -15.98
(17.09) (17.13) (17.16)
Share of women on board 13.35 13.27 12.88
(11.92) (11.93) (11.96) CEO sex 5.46 5.47 5.35 (4.60) (4.60) (4.63) CEO age -0.08 -0.07 -0.07 (0.21) (0.21) (0.21) CEO tenure -0.25 -0.24 -0.24 (0.20) (0.21) (0.21)
CEO narcissism (Wordcount) -0.74 -0.17
(3.66) (0.76) CFO narcissism -0.19 (0.86) Moderator effect 0.63 (0.68) CFO sex -1.08 (3.00) CFO age 0.05 (0.16) Constant -16.17 -16.24 -20.19 (40.37) (40.42) (41.51) N 1525 1525 1525 WaldChi2 396.32 395.19 398.48 (Prob > Chi2) 0.00 0.00 0.00 Adjusted R2 46.37 46.35 46.49
Notes: Standard Errors in parenthesis. Year and Industry dummies included but not reported. (ICT = Interest coverage ratio)
