Effects of CEOs’ Negative Traits on Corporate Social Responsibility

Abstract

The dark triad, composed of Machiavellianism, psychopathy and narcissism, refers to negative personality traits, which may influence business processes. While negative traits could be an important factor explaining the relationship between a CEO’s immoral and unethical behavior and corporate social responsibility (CSR), there has been minimal research focusing on this relationship. This paper thus attempts to investigate how a CEO exhibiting these negative traits affects CSR, and if an employee’s perception of ethics and social responsibility would mediate the relationship. In addition, this paper considers the moderating effects of an individual performance-based compensation system (IPBCS) between employee’s CSR perception and CSR activities. The data are collected through a survey conducted on 165 employees (companies) in twelve industries. The regression result indicates an inverse relationship between the negative traits of a CEO and an employee’s perception of ethics and social responsibility and CSR activities, and the mediating effect of the perception in the relationship between the negative traits and CSR activities. It also indicates that an IPBCS moderates the relationship between CSR perception and activities. Implications for the study, future research directions, and management approach are discussed.

1. Introduction

CSR has emerged not only as a matter of choice but also as part of strategic decision making. Many studies and academic discussions on corporate social responsibility (CSR) have taken place since the 2000s. Despite considerable attention and research on CSR, few studies have examined the effect of an organization’s internal resources, leadership, and organizational culture on its CSR activities [1]. This suggests the need for better understanding of CSR and top management issues because CEOs have a powerful influence on the introduction, management, and performance of strategies [2] and CEO characteristics are related exemplary corporate social performance [3]. In particular, their personal values are important, and their perception of CSR is a crucial determinant of corporate social performances [4]. Given the development of mass media, CEOs are likely to feel substantial pressure from stakeholders to do more ethical decision makings and thus may place greater emphasis on a positive public image [5].

This study considers the dark triad composed of Machiavellianism, psychopathy and narcissism as a CEO’s negative trait. These personalities have callousness (low empathy), disagreeableness and social exploitativeness as the common core [6]. Although many researches have addressed the dark triad, there is only one empirical study on the relationship between one of the dark triad traits and CSR (Boddy [7] argued again that the lack of corporate social responsibility of corporate psychopaths damages their own organization as well as society) [8]. However, it considers only one of three traits—psychopathy. It investigates the effect of psychopathy on CSR, but there is no direct evidence of a psychopath reducing the organization’s CSR. The respondents (employees) who perceive their leader as revealing a psychopath are just likely to believe that their organization has a lower functioning of CSR. This analysis is not conducted at the organizational level, but at the individual level. Furthermore, information regarding the number of companies studied was not divulged. All dark triad traits, in contrast, are investigated to verify their relationship with CSR perception and activities in this study. Especially, this study attempts to distinguish between an employee’s perception of social responsibility and CSR activities.

In addition, Manner [3] has articulated that the short-term compensation gives a negative relationship with corporate social performances. It is a reasonable assumption that both internal and external factors may influence the relationship between the perception and activities, and thus a compensation system is employed as a moderator for this relationship in the study. Based on the discussion, the following research questions are proposed:

(1)

Is there any relationship among the CEO’s dark triad traits, the employee’s perception of ethics and social responsibility, and CSR execution?

(2)

Does a compensation system moderate the relationship between the perception and the execution of CSR functions?

2. Materials and Methods

2.1. The Negative Trait

The dark triad consists of socially undesirable traits of Machiavellianism, psychopathy, and narcissism [9,10]. They need to be divided into three types, although these traits appear to be linked with lacking ability to compromise, and featuring impulsivity and interpersonal manipulation, and are associated with low conscientiousness [11]. Recent researches have now focused on so-called ‘‘successful’’ psychopaths and narcissists in some context [6], and the dark triad embodies a few of desirable traits like charm, assertiveness, and impression management skills [12]. However, numerous studies have articulated that narcissism has been linked to unethical behavior in CEOs and a need for power [13,14,15], psychopaths can adversely affect productivity [13], and Machiavellianism is associated with diminished organizational commitment [16,17], with a tendency to focus on power and manipulative behaviors [13,18].

Machiavellianism is a strong desire for power without feelings to communicate with others. To achieve a clear goal, it emphasizes practicality with relentless tendency. It is well known that Machiavellianism originated from Machiavelli’s arguments in his famous book “Il Principe”. It includes characteristics of self-beneficial behavior, interpersonal manipulation, deception, and a lack of empathy. In addition, it has been shown to be negatively related with conscientiousness and extraversion [19], and the recognition of the importance of business ethics in achieving organizational success [20].

Some aspects of psychopathy can lead to success in leadership positions [21], and the trait of psychopathy sometimes appears to be attractive and competent in an organization [8]. However, psychopathy would be willing to falsify financial results to get promotion [7]. Psychopathy is divided into two parts [22]. Primary psychopathy includes such features as selfishness, deficiencies in interpersonal aspects, remorselessness, superficial charm, and exploitation, on the other hand, secondary psychopathy includes antisocial lifestyles and behaviors [23]. It is argued that Machiavellianism and primary psychopathy are essentially identical [22,24]. Williams, Nathanson and Paulhus [25] identified manipulation, a lack of affective experiences, and impulsive or irresponsible behaviors as characteristics of psychopathy. Ray and Jones [26] suggested that attitudes and certain traits associated with psychopathy explain intention to engage in environmental toxic dumping. It can be also described as glibness, superficial charm, conning, a grandiose sense of self-worth, a lack of remorse, emotional shallowness, calculating and cold, a lack of empathy, and a refusal to take responsibility for own actions [8].

Narcissism is a self-loving trait rooted in ancient Greek mythology and encapsulates several different, but interrelated, traits such as persistent attention seeking. Raskin and Terry [27] identified core narcissistic personality traits, including self-admiration, self-absorption, authority, exhibitionism, superiority, entitlement, arrogance, exploitation, self-sufficiency, and extreme vanity [19,28]. However, narcissism can be a positive attribute for a firm or society [15,29]. Narcissism is similar to charisma in that it is self-confident [15], and therefore narcissistic leadership may help an organization adapt to rapid environmental changes and bring about innovation [29,30].

2.2. Theoretical Background and Hypothesis Development

2.2.1. CEOs’ Negative Trait and Employee’s Perception of Ethics and Social Responsibility

The attitude of a responsible leader is an important factor in establishing desirable relationships with stakeholders [31], and CEO’s individual personality or values determine the perception of CSR [32]. Furthermore, Waldman et al. [5] argue that a CEO’s strong morality contributes to a positive effect on CSR results. The dark triad traits will have negative effects on employees and other internal stakeholders, as they are unconscientious and manipulate others in interpersonal relationships [11]. They will also degrade the importance of social responsibility throughout an organization.

Valentine and Fleischman [33] argued that an ethical organizational culture is a key factor in employee’s positive perception and attitude concerning CSR, and a CEO’s negative traits influence an individuals’ attitude or behaviors through organizational culture. Machiavellianism has a negative relationship with business ethics and recognition of the importance of social responsibilities [20]. Similarly, the lack of empathy that is representative of the psychopathy trait could affect an organization negatively and frequently.

Therefore, it expected that these attributes would result in employee’s ignorance of ethics and social responsibility. The narcissistic trait is assumed to negatively influence ethics and social responsibility, because narcissistic CEOs aggrandize their power with self-conviction, despite the possibility of making the wrong decision or judgement [34]. Based on the above discussion, the following hypothesis is proposed:

Hypothesis 1.

The stronger the CEO’s negative traits (Machiavellianism, psychopathy and narcissism), the less likely his or her employee is to perceive the importance of ethics and social responsibility.

2.2.2. CEOs’ Negative Trait and CSR Activities

CSR activities, performed for a variety of stakeholders who influence business practices directly or indirectly, are understood as discretionary [3], altruistic and strategic management decisions [35], and vary according to a company’s unique situation [36,37,38]. Public needs and external requirements may serve as the background of CSR programs [36]. Nevertheless, the main factor for various CSR execution is ultimately an internal decision [3] rather than a stakeholder’s pressure or the flow of new institutional isomorphism [39].

Investment cost for strategic CSR activities follows from business leader’s personal decision-making [40,41]. CSR is not legally mandatory but constitutes managerial activity conducted by virtue of the leader’s authority [42]. Individual traits of CEOs have been identified as an important factor of organizational behavior [43]. Therefore, the ethical attitude of CEOs is critical in determining the level of discretionary CSR activities. The ethics of top management work as an essential driver of a responsible corporate citizen’s activities [33,44,45]. It was known that the decisions of psychopaths lack strategic intent [7]. Psychopaths are concerned with tactical aspects rather than the long term and strategic interests of their organization [46], and are negatively associated with responsibility [7].

With the viewpoint from the CEO’s ethical (or unethical) qualities and discretionary determination of CSR activities, it is assumed that the CEO’s undesirable traits influence the breadth of CSR activities. Babiak and Hare [47] argued that a CEO’s negative traits are related to financial, moral, and legal risks, and also result in decisions which negatively influence society. Based on the above discussion, the following hypothesis is proposed:

Hypothesis 2.

The stronger the CEO’s negative traits (Machiavellianism, psychopathy and narcissism), the less likely his or her organization engages in CSR activities.

2.2.3. Employee’s Perception of Ethics and Social Responsibility and CSR Activities

Organizational behavior theorists find that CSR perceptions influence employees’ attitudes and behaviors [48]. Researchers have shown that if employees perceive their organization is engaged in socially irresponsible activity, they are likely to show negative work attitudes and behaviors [49]. In contrast, if employees perceive their company behaving in a socially responsible manner, they are likely to have positive attitudes [50,51].

Therefore, a company ought to take into account the impacts of its CSR activities on society, based on the perceived importance of ethics and social responsibility, considering the relationship between an employee’s perception of ethics and social responsibility and CSR activities. By doing this, the company acquires legitimacy within their institutionalized circumstances [52]. Ditlev-Simonsen [53] showed that if the employee’s perception of ethics and social responsibility is expanded to include employees, organizational commitment will be increased. Wong and Gao [54] have examined the direct or indirect relationship between organizational commitment and CSR, and found out that an employee in an organization features greater organizational commitment when his or her career development achieves balance with social aspects.

Pastrana and Sriramesh [55] found that an organization can exert a positive influence on organizational members’ performance and customer loyalty through employee’s perceived importance of ethics and social responsibility. In addition, an organization engages in activities that satisfy stakeholders when an employee perceives ethics and social responsibility affirmatively. In brief, it is expected that there is a positive relationship between the employee’s perception of ethics and social responsibility and CSR activities. Based on the above discussion, the following hypothesis is proposed:

Hypothesis 3.

The stronger the organization’s perception of ethics and social responsibility, the more likely it is to engage in CSR activities and the employee’s perception mediates the relationship between the CEO’s negative trait and CSR activities.

2.2.4. The Moderating Effect of Individual Performance-Based Compensation System

A compensation system based on motivational theory may be designed differently depending on the organization and its environment [56], and has various forms such as annual pay systems (merit pay in the U.S.), profit sharing, incentive pay, and wage differentials. Compensation systems may positively influence many aspects of an organization [57,58,59]. However, when only the job performance-based compensation system exists in the organization, self-determination and intrinsic motivation may decrease as a result of excessive attention to external compensation [58,60,61]. Introduction of individual performance-based compensation systems (IPBCS), such as merit pay and incentives, may undermine its human relations, team spirit, and togetherness [62,63].

In addition, an IPBCS can increase the likelihood of employees neglecting long-term performance and resource planning. In particular, an individual performance-based compensation system is likely to have a strong and myopic propensity toward individual performance. Under this compensation system, employees prefer to pursue short-term visible goals, and innovation may be hampered to the detriment of future performance [64].

Compensation systems that are inconsistent with the cultural and institutional background of the organization may cause negative effects on management performance. In a national business system in the Confucian tradition, such as that in South Korea, external rewards are not always effective for high organizational performance, and the traditional high-commitment human resource management system may remain necessary [65]. CSR activities that should be developed through long-term relationships with society involving a variety of stakeholders may be negatively affected by the adoption of an IPBCS. Based on the above discussion, the following hypothesis is proposed and the research model is shown in Figure 1:

Hypothesis 4.

An individual performance-based compensation system (IPBCS) will negatively moderate the relationship between the employee’s perception of ethics and social responsibility and CSR activities.

2.3. Methods

2.3.1. The Sample and Research Procedure

This study employed an online survey method using the KSDC (KSDC established in 1997, provides services such as basic statistical analysis, data provisioning, and data processing and preservation for the efficient social science research. Researchers can conduct on-line survey on KSDC’s website. (http://www.ksdc.re.kr/unisql/engjap/ewa/ewa1111.html)) (Korea Social Science Data Center). The survey was conducted in April 2014. The questionnaire was e-mailed initially from the survey system of KSDC to a variety of person at each firm collected through researchers’ acquaintances. Demographic variables such as gender, age, education level, employee tenure, the position at the firm, and types of jobs were controlled to remove any response bias. Every respondent who received the e-mailed questionnaire was encouraged to participate in the survey by researchers’ phone calling directly. Through this process, the data completed by 207 respondents among 267 persons were collected automatically through the KSDC system.

In order to attempt to minimize selection bias from oversampling one industry alone, a wide variety of firms were included in the analysis. A total of 207 responses were checked to avoid any duplication by e-mail and IP addresses of respondents. Among firms with two or more respondents, only one response was randomly selected for use in this study. Nine respondents from unidentified firms as well as twelve respondents from nonprofit organizations were excluded (This study focused on for-profit businesses). As a result of CSR being performed at the firm level, the sample was also analyzed at firm level.

Consequently, a total of 165 responses were included in the sample. Of the firms sampled, 38.8% of the firms were SMEs with fewer than 300 employees, and 84.2% were operating for more than 10 years. Only four firms had female CEOs, and 52.1% of the CEOs held their positions for more than three years. In addition, 70.3% of the CEOs were between the ages of 50 to 65. Industries included in the sample are summarized in

Table 1. Industries in the sample.

Industry	Sample Size N%
Petrochemicals	18	10.9%
Metals and minerals	4	2.4%
Electronics	17	10.3%
Automobile	10	6.1%
Food and beverage	4	2.4%
Information and telecommunications	16	9.7%
Other manufacturing	9	5.5%
Construction and real estate	13	7.9%
Retail and distribution	11	6.7%
Financial and banking services	34	20.6%
Public services	7	4.2%
Other services	22	13.3%
Total	165	100.0%

.

2.3.2. Measurement

Questionnaire items were derived from established scales from previous literature, translated into Korean considering English illiteracy and cultural differences (First, one of the researcher translated the scales, and then another researcher, management professor, who had received a doctorate in the U.S. reviewed the translated scales and revised vague expressions). For the independent variables, namely the CEO’s dark triad traits, the ‘Dirty Dozen scale’ developed by Jonason and Webster [66] was used. This scale is composed of 12 items, four items for each of Machiavellianism, psychopathy, and narcissism. The Dirty Dozen scale is the simplest of all the various scales for the dark triad, so that it is appropriate for this analysis employing a number of other items. Because each employee who was well aware of the CEO responded of the CEO, items from the Dirty Dozen scale were revised for this survey. For instance, the item “I have used deceit or lied to get my way” was revised to “The CEO of my company has used deceit or lied to get his/her way”.

If it is needed to identify a person’s characteristics, many studies have supported the self-other or self-peer agreement [67,68,69,70,71,72]. This means that when it is difficult to ask questions directly to someone, there is no significant difference between the results of direct confirmation and the responses through his or her peers. Thus, this study method is deemed reasonable and reliable. Mahaffey and Marcus [73] claimed that unethical propensity and anti-social personality are observed easily even by those not educated about observation and evaluation, and therefore the method of this study is deemed acceptable. The issue of consistency can be tested for reliability. As Cronbach α of the variable is larger than .9 in our study, the survey results of the dark triad are reliable. Respondents were asked to indicate the extent to which they thought of their CEO as “Machiavellian, psychopathy, and narcissistic” on a six-point Likert-type scale, ranging from “not at all” (1) to “very much” (6).

The employee’s perception of ethics and social responsibility were estimated using “company-PRESOR” [74], a revised version of PRESOR [75], to investigate the extent to which a company perceives the importance of ethics and social responsibility. For example, the item “Being ethical and socially responsible is the most important thing a firm can do” was revised to “My company believes that being ethical and socially responsible is the most important thing it can do” [74]. Therefore, the higher the score from respondents, the more likely the respondent’s perception of the company places importance on ethics and social responsibility.

CSR activities were tested using the stakeholder-focused evaluation scale [76], composed of CSR issues from the perspective of stakeholders. The analysis employed a six-point Likert-type scale as well as other scales. This scale includes seven items for IPBCS such as performance-based incentives and a substantial salary gap, and gives a score of 0 (“no”) or 1 (“yes”). Therefore, the maximum possible score was seven, and the minimum, zero. The analysis employed the measurement method used in Kim and Kim [77] using IPBCS as a moderator.

This analysis is conducted in the same way (survey method) with the same respondents using dependent and independent variables, therefore, common method bias was removed. First, the order of each question of all the scales used in this study were rearranged, because respondents could interpret different meanings depending on the context and the order of questions [78]. Second, the scale, SDS-17 [79], for social desirability was used as a control variable. SDS-17 is composed of 16 items. The scale was developed recently, and its content was tailored for impression management, not self-deceptive enhancement. Therefore, it was considered appropriate for the analysis [80].

According to previous analysis [80,81], the reliability and validity of SDS-17 is suitable for application in South Korea and the U.S., as well as Germany where it was developed. SDS-17 uses a dichotomous scoring method giving a score of 0 (“no”) or 1 (“yes”). The total score for items was used in the analysis. In addition, eight variables were included in the analysis as control variables. These were firm-related variables (size, history and industry), CEO-related variables (tenure, age, and gender), and respondent-related variables (tenure and position).

2.4. Factor Analyses

2.4.1. CEO’s Negative Trait

Jonason and Webster [66] developed the Dirty Dozen scale and noted that Machiavellianism, psychopathy, and narcissism may be treated as separate constructs from each other or treated as one latent variable. However, factor analysis found Machiavellianism and psychopathy (Ma-Psy) to be one factor compositely, and narcissism to be another factor in this study. Miller et al. [24] showed that Machiavellianism and psychopathy manifested nearly identical empirical profiles and both were significantly related to disinhibitory traits. Psychopathy can be divided into two dimensions. One dimension is primary psychopathy, which refers to a personality trait, and the other is secondary psychopathy, which is anti-social behavior. McHoskey, Worzel, and Szyarto [22] argued that Machiavellianism and primary psychopathy are essentially identical constructs [7]. All the psychopathy items in the Dirty Dozen that imply a lack of remorse, immorality, callousness, and cynicalness are the representative characteristics of primary psychopathy [6].

Therefore, this factor analysis result supports the previous researches. The analysis was conducted using SPSS for exploratory factor analysis, and AMOS for confirmatory factor analysis. The principal axis factoring and the promax rotation were employed because it is not possible for social science research to have no correlation between factors.

The results of the KMO (Keiser-Meyer-Olkin) and the Bartlett test of sphericity were 0.875 (>0.8) and p = 0.000 (<0.05), respectively, indicating no statistical problem. Based on the communality (>0.3) criteria, nine indicators (observed variables) were selected. Finally, indicators 2, 3, 4, 6, and 8 were included in “Machiavellianism-Psychopathy” (Ma-Psy), and indicators 9, 10, 11, and 12 were included in “Narcissism” (Nar). All factor loadings exceeded 0.6, and Cronbach α was 0.908 (Ma-Psy: 0.904; Nar: 0.932), indicating sufficient reliability (internal consistency). Fit and validity tests using AMOS reveal that the CFI and TLI (incremental fit indices) were 0.954 and 0.936, respectively, and the GFI and the RMSEA (absolute fit indices) were 0.908 and 0.112, respectively. These results indicate a good model fit. All factor loadings (estimates) exceeded 0.5 (C.R. > 1.96), and the AVE values were 0.537 for Ma-Psy and 0.672 for Nar. Construct reliability values were found to be 0.852 for Ma-Psy and 0.891 for Nar. Therefore, sufficient convergent validity was achieved. All AVE values exceed the squared correlation between factors (0.227), indicating sufficient discriminant validity.

2.4.2. Employee’s Perception of Ethics and Social Responsibility

For comprehensive understanding of an employee’s perception of ethics and social responsibility, company-PRESOR was analyzed. Original PRESOR research [75] featured three factors containing 13 indicators, but subsequent studies [74,82] report two factors present with differences in the number of indicators for the same factor. On the other hand, company-PRESOR [74] featured only one factor. Thus, factor analysis was conducted based on the original 16 indicators in this analysis. Principal axis factoring and promax rotation were utilized such that two factors appeared and were divided by a reverse-coded item consistent with previous studies. Except for indicator 3 (communality: 0.165), indicators 1, 4, 6, 7, 9, 10, 11, 12, and 15 were loaded on the first factor, and reverse indicators 2, 5, 8, 13, 14, and 16 were loaded on the second factor. The first factor was named “Long-Term, Top Priority” (Top Priority, hereafter), and the second factor, “prima facie Duty” (Duty, hereafter) [82].

The KMO and the Bartlett test of sphericity results were 0.933 and p = 0.000, respectively, indicating no statistical problem. In addition, all pattern matrix loadings exceed 0.5, and Cronbach α was 0.925 (Top Priority: 0.944; Duty: 0.831), indicating sufficient reliability. According to fit indices, the model provided an acceptable fit to the data (CFI = 0.972, TLI = 0.967, GFI = 0.906, RMSEA = 0.057). All factor loadings exceed 0.5 (C.R. > 1.96), and AVE values are 0.575 for Top Priority and 0.348 for Duty. Although the AVE value of Duty is less than 0.5, construct reliability exceeds 0.7 (Top Priority: 0.923; Duty: 0.761), which is acceptable. The squared correlation between factors is 0.446, exceeding the AVE of Duty. However, the discriminant validity can be supported because the factors belong to PRESOR.

2.4.3. CSR Activities

A factor analysis of CSR activities measured by the CSR scale [76] was conducted. The original scale had four factors (society and the environment, employees, customers, and the government), but only two items belonged to the government. In this analysis, only 17 observed variables for 18 items were analyzed because the 8th item had been eliminated (Factor loading of the 8th item was larger than 4 on factor 1 and factor 3 simultaneously in the original research [76]) in the original paper [76]. After testing with principal axis factoring and promax rotation, three factors were established. The first factor, “CSR to employees” (CSR-Em), had six indicators (7, 9, 10, 11, 12, and 13). The second factor, “CSR to society and the environment” (CSR-SE), had three indicators (1, 5, and 6). In addition, the third factor, “CSR to business” (CSR-Bu), had five indicators (14, 15, 16, 17, and 18). Initially, indicators 2 and 4 were included in CSR-Em, but the items were later excluded because they were associated with the “future generation”.

The KMO and the Bartlett test of sphericity were 0.894 and p = 0.000, respectively, indicating no statistical problem. All pattern matrix loadings exceed 0.4, and Cronbach α is 0.909 (CSR-Em: 0.896; CSR-SE: 0.855; CSR-Bu: 0.802), indicating sufficient reliability. According to fit indices, the model provides a generally acceptable fit to the data (CFI = 0.911, TLI = 0.890, GFI = 0.864, RMSEA = 0.096). All factor loadings exceed 0.5 (C.R. > 1.96), and AVE values were 0.520 (CSR-Em), 0.511 (CSR-SE), and 0.475 (CSR-Bu). Although the AVE value of CSR-Bu is less than 0.5, construct reliability exceeds 0.7 (CSR-Em: 0.864; CSR-SE: 0.758; CSR-Bu: 0.817), indicating sufficient convergent validity. The squared correlation between CSR-Em and CSR-SE is 0.361, and the one between CSR-SE and CSR-Bu is 0.339. However, the squared correlation between CSR-Em and CSR-Bu is 0.605, exceeding the AVE. However, the factors belong to CSR activities, assuming sufficient discriminant validity.

Table 2. Factor analysis results.

Latent Variables	Negative Traits	Perception	Activities
Sub-constructs
Factor 1	Machiavellianism-Psychopathy (Ma-Psy)	Long-term and top priority (Top Priority)	CSR to employees (CSR-Em)
Factor 2	Narcissism (Nar)	prima facie duty (Duty)	CSR to society and the environment (CSR-SE)
Factor 3			CSR to business (CSR-Bu)

summarizes the aforementioned factor analysis results.

3. Results

3.1. Correlations

Pearson correlation coefficients between variables are presented in

Table 3. Pearson correlations and descriptive statistics ¹.

Variables 2	1.	2.	3.	4.	5.	6.	7.	8.	9.	Mean	Standard Deviation
1. CSR-Em	1									4.048	0.956
2. CSR-SE	0.661 **	1								3.513	1.222
3. CSR-Bu	0.761 **	0.565 **	1							4.770	0.727
4. Ma-Psy	−0.573 **	−0.378 **	−0.490 **	1						2.882	1.113
5. Nar	−0.278 **	−0.157 *	−0.109	0.520 **	1					3.980	1.197
6. Top-priority	0.752 **	0.722 **	0.758 **	−0.544 **	−0.175 *	1				4.096	1.010
7. Duty	0.424 **	0.504 **	0.515 **	−0.529 **	−0.189 **	0.696 **	1			3.971	0.928
8. IPBCS	0.205 **	0.121	0.056	−0.105	−0.151 *	0.020	−0.064	1		5.479	1.580
9. Social Desirability	0.364 **	0.156 *	0.270 **	−0.283 **	−0.204 **	0.299 **	0.195 **	0.025	1	8.976	3.104

¹ ** and * indicate significance at p < 0.01 and p < 0.05, respectively. ² Eight dummy variables (firm size, firm history, industry, CEO tenure, CEO age, CEO gender, employee tenure, and the position within the firm) are not included in this table. This footnote is not written again in other Tables, owing to the same covariates being used.

. Significant relationships are found between dependent and independent variables. A meta-analysis of the previous study on dark triad showed that the correlations of three components are high [6]. This result (Table 3) is similar to the meta-analysis result.

In particular, Ma-Psy and Nar, the main variables of interest, are negatively correlated with the dependent variable. Ma-Psy and Nar are negatively correlated with the mediator, but there is a significant negative relationship only between Nar and the moderator. All these results are consistent with the regression results. Finally, there is a significant correlation between an employee’s perception of ethics and social responsibility (Top-priority and Duty) and CSR activities (CSR-Em, CSR-SE, and CSR-Bu).

3.2. Regression Results 1 (Hypothesis 1)

In order to test Hypothesis 1, regression analyses with an employee’s ethics and social responsibility as dependent variables were performed for Top-priority and Duty, separately. There is no problem with VIF (variance inflation factor) and tolerance, and the assumption of regression errors.

Table 4. Regression analysis results for Hypothesis 1 ¹.

Variables		B (Regression Coefficient)	β (Standardized Beta Coefficient)	t-Value	B	β	t-Value
Dependent Variables		① Top-priority			② Duty
Independent Variables	Ma-Psy	−0.512	−0.501	−5.700 **	−0.529	−0.547	−6.344 **
	Nar	0.095	0.095	1.064	0.107	0.113	1.292 †
Control Variable	Social desirability	0.060	0.189	2.615 **	0.010	0.034	0.474
- Top-priority: R2 = 0.542, Adjusted R2 = 0.379, F-statistic (43, 121) = 3.324 **
- Duty: R2 = 0.559, Adjusted R2 = 0.402, F-statistic (43, 121) = 3.567 **

¹ ** and * indicate significance at p < 0.01 and p < 0.05, respectively, and ^† indicates p < 0.10 (one-tailed test). This footnote is not written again in other tables, owing to the same rules being used.

shows the regression analysis results for Hypothesis 1. First, in regression ① with Top-priority as a dependent variable, the adjusted R² is 0.379 and the F-value 3.324 (p < 0.01). Ma-Psy has a negative effect on Top-priority. t-value for Ma-Psy is −5.700 and it is significant (p < 0.01). The results indicate that the stronger the CEO’s Machiavellianism and psychopathy traits, the less likely his or her employee is to perceive the importance of ethics and social responsibility. In contrast, Narcissistic CEO shows no significance with regard to Top-priority (t-value is 1.064 and p-value > 0.10). Therefore, the results partially support Hypothesis 1.

Second, the adjusted R² is 0.402 and the F-value is 3.567 (p < 0.01) in regression ②. Coefficient of Ma-Psy shows a significant and negative relationship between Ma-Psy and Duty (t-value, −6.344, is significant at p < 0.01). It indicates that the stronger the CEO’s Machiavellianism and psychopathy traits, the less likely his or her employee is to perceive ethics and social responsibility as a prima facie duty. However, Nar shows a positive relationship with Duty (t-value is 1.292 and p-value < 0.10). This contrasting result can be explained as ambivalent narcissism [15,29].

3.3. Regression Results 2 (Hypothesis 2)

In order to test Hypothesis 2, regression analysis with CSR activities as a dependent variable was conducted three times. There seems to be no problem with VIF and tolerance, and with the assumption of regression errors.

Table 5. Regression analysis results for Hypothesis 2.

Variables		B	β	t-Value	B	β	t-Value	B	β	t-Value
Dependent Variables		① CSR-Em			② CSR-SE			③ CSR-Bu
Independent Variables	Ma-Psy	−0.458	−0.457	−5.243 **	−0.311	−0.307	−3.179 **	−0.491	−0.511	−5.502 **
	Nar	0.013	0.013	0.150	0.030	0.030	0.309	0.211	0.225	2.391 **
Control Variable	Social desirability	0.064	0.206	2.871 **	0.018	0.056	0.709	0.053	0.179	2.331 *
- CSR-Em: R2 = 0.549, Adjusted R2 = 0.389, F-statistic (43, 121) = 3.425 **
- CSR-SE: R2 = 0.448, Adjusted R2 = 0.252, F-statistic (43, 121) = 2.282 **
- CSR-Bu: R2 = 0.487, Adjusted R2 = 0.305, F-statistic (43, 121) = 2.673 **

shows regression analysis results for testing Hypothesis 2. First, regression ① with CSR-Em as a dependent variable produces the adjusted R² of 0.389 and the F-value of 3.425 (p < 0.01). Ma-Psy shows negative relationship with CSR-Em (t-value of −5.243, significant at p < 0.01). The result suggests that the stronger the CEO’s Machiavellianism and psychopathy traits, the less likely his or her organization is to execute CSR activities to employees. In contrast, Narcissistic CEO shows no significance with regard to CSR-Em (t-value is 0.150 and p-value > 0.10).

Second, results of regression ② with CSR-SE are similar to those of regression ①, with the adjusted R² of 0.252 and the F-value of 2.282 (p < 0.01). Also, Ma-Psy and CSR-SE show a negative significant relationship with the t-value of −3.179 (p < 0.01). The result indicates that the stronger the CEO’s Machiavellianism and psychopathy traits, the less likely his or her organization executes CSR activities for society and the environment. However, Nar reveals no statistically significant effects on CSR-SE (t-value is 0.309 and p-value > 0.10).

Third, in the regression ③ with CSR-Bu as a dependent variable, adjusted R² is 0.487 and the F-value is 2.673 (p < 0.01). Ma-Psy shows a negative effect on CSR-Bu with t-value of −5.502 (p < 0.01). Thus, the stronger the CEO’s Machiavellianism and psychopathy traits, the less likely his or her organization is to execute CSR activities related to business. This results are the same as those from regression ① and ②, with exception of positive relationship between Nar and CSR-Bu (t-value is 2.391 and p-value < 0.01). It is inferred that the exception with Nar is due to an ambivalent propensity of narcissism as in Table 4, where Duty is a dependent variable.

As mentioned above, although the narcissistic trait is part of the dark triad, it also contains the attribute of charismatic disposition. Thus, organizational innovation to respond rapidly and proactively to environmental changes may be led effectively by the narcissistic leader’s self-absorption and self-superiority [19,30]. CSR-Bu, a latent variable in this research consisting of protecting consumer rights, product liability, customer satisfaction and active compliance is implemented by the leader’s proactive will. The narcissistic trait shows a positive influence on CSR-Bu in this analysis.

3.4. Regression Results 3 (Hypothesis 3)

In order to test the single mediator model (Hypothesis 3), Sobel’s test [83] as well as the method of Baron and Kenny [84] were used. The analysis for the mediating effect of the employee’s perception of ethics and social responsibility is conducted six times (2 mediators $\times$ 3 dependent variables). The results indicate no problem with VIF (variance inflation factor), tolerance, and the assumption of regression errors.

Table 6. Regression analysis results for Hypothesis 3 (CSR-Em).

Analysis	Variables	B		S (Standard Error)		β		t-Value
1	Dependent Var.	CSR-Em
	Ma-Psy	−0.458		0.087		−0.457		−5.243 **
	Nar	0.013		0.087		0.013		0.150
	Variables	B	S	β	t-Value	B	S	β	t-Value
2	Dependent Var.	① Top-priority				② Duty
	Ma-Psy	−0.512	0.090	−0.501	−5.700 **	−0.529	0.083	−0.547	−6.344 **
	Nar	0.095	0.089	0.095	1.064	0.107	0.083	0.113	1.292 †
3	Dependent Var.	CSR-Em
	Ma-Psy	−0.160	0.079	−0.160	−2.022 *	−0.392	0.101	−0.391	−3.898 **
	Nar	−0.042	0.070	−0.043	−0.602	0.000	0.087	0.000	−0.005
	Top-priority	0.582	0.071	−0.593	8.179 **
	Duty					0.125	0.095	0.121	1.316 †
	Social desirability	0.029	0.018	0.094	1.580 †	0.063	0.022	0.202	2.820 **
- Top-priority: R2 = 0.710, Adjusted R2 = 0.604, F-statistic (44, 120) = 6.690 **
- Duty: R2 = 0.555, Adjusted R2 = 0.392, F-statistic (44, 120) = 3.407 **

shows regression results for testing Hypothesis 3(CSR-Em). In analysis 3, the model F value shows a significance (p < 0.01) and adjusted R² values (0.604 for Top-priority and 0.392 for Duty) are larger than those in Table 5. First, in regression ① with Top-priority as a mediator, Top-priority positively affects CSR-Em (t-value of 8.179, significant at p < 0.01) Also, the direct effect of Ma-Psy on CSR-Em is negative and statistically significant at p < 0.01 (t-value is −2.022). Second, in regression ② with Duty as a mediator, Duty also positively influences CSR-Em (t-value is 1.316, p < 0.10). Similarly, Ma-Psy negatively influences CSR-Em, referred to as the direct effect of Ma-Psy on CSR-Em (t-value is −3.898, p-value < 0.01).

The mediating effect of Top-priority can be tested in accordance with Baron and Kenny [84]. The absolute value of the regression coefficient (B = −0.160) of Ma-Psy in analysis 3 is smaller than that (B = −0.458) in analysis 1, and the direct effect of Ma-Psy is significant. Consequently, Top-priority mediates partially the relationship between Ma-Psy and CSR-Em. To verify statistical significance of the partial mediation, Sobel’s z-test is conducted. z value of −4.6736 (p < 0.01) indicates the significance of the mediating effect of Top-priority.

$$z=\frac{\left(-0.512\times 0.582\right)}{\sqrt{-{0.512}^2\times {0.071}^2+{0.582}^2\times {0.09}^2}}=-4.6736(\mathrm{for}\mathrm{Top}-\mathrm{priority}),$$

(1)

Conversely, the absolute value of the regression coefficient (B = −0.392) of Ma-Psy in analysis 3 with Duty is smaller than that (B = −0.458) in analysis 1, and the direct effect of Ma-Psy is significant. Consequently, Duty mediates partially the relationship between Ma-Psy and CSR-Em. However, the result of Sobel’s test does not support the mediating effect of Duty.

$$z=\frac{\left(-0.529\times 0.125\right)}{\sqrt{-{0.529}^2\times {0.095}^2+{0.125}^2\times {0.083}^2}}=-1.2886(\mathrm{for}\mathrm{Duty}),$$

(2)

Table 7. Regression analysis results for Hypothesis 3 (CSR-SE).

Analysis	Variables	B		S		β		t-Value
1	Dependent Var.	CSR-SE
	Ma-Psy	−0.311		0.090		−0.307		−3.179 **
	Nar	0.030		0.116		0.030		0.309
	Variables	B	S	β	t-Value	B	S	β	t-Value
2	Dependent Var.	③ Top-priority				④ prima facie Duty
	Ma-Psy	−0.512	0.090	−0.501	−5.700 **	−0.529	0.083	−0.547	−6.344 **
	Nar	0.095	0.089	0.095	1.064	0.107	0.083	0.113	1.292 †
3	Dependent Var.	CSR-SE
	Ma-Psy	0.031	0.087	0.030	0.354	−0.166	0.110	−0.164	−1.511 †
	Nar	−0.033	0.077	−0.033	−0.430	0.001	0.095	0.001	0.009
	Top-priority	0.667	0.078	0.673	8.504 **
	Duty					0.272	0.104	0.260	2.621 **
	Social desirability	−0.022	0.020	−0.071	−10.097	0.015	0.024	0.048	0.612
- Top-priority: R2 = 0.655, Adjusted R2 = 0.529, F-statistic (44, 120) = 5.188 **
- Duty: R2 = 0.478, Adjusted R2 = 0.286, F-statistic (44, 120) = 2.494 **

shows the regression analysis results for testing Hypothesis 3 (CSR-SE). In analysis 3, the model F value shows significance (p < 0.01) and adjusted R² values (Top-priority is 0.529 and Duty is 0.286) are larger than those in Table 5. First, in regression ③ with Top-priority as a mediator, Top-priority positively influences CSR-SE (t-value of 8.504, p < 0.01). However, the result shows no significant relationship between Ma-Psy and CSR-SE (t-value is −0.354). Second, in regression ④ with Duty as a mediator, Duty also positively influences CSR-SE (t-value is 2.621, p < 0.01), and Ma-Psy has a negative effect on CSR-SE (t-value is −1.511, p-value < 0.10).

The absolute value of the regression coefficient (B = 0.031) of Ma-Psy with Top-priority in the analysis 3 is smaller than that (B = −0.311) in analysis 1, and the direct effect is not significant [84]. Consequently, Top-priority completely mediates the relationship between Ma-Psy and CSR-SE. The result of the Sobel’s test provides further support.

$$z=\frac{\left(-0.512\times 0.667\right)}{\sqrt{-{0.512}^2\times {0.078}^2+{0.667}^2\times {0.09}^2}}=-4.7365(\mathrm{for}\mathrm{Top}-\mathrm{priority}),$$

(3)

The absolute value of the regression coefficient (B = −0.166) of Ma-Psy in analysis 3 with Duty is smaller than that (B = −0.311) in analysis 1, and the direct effect is significant. Therefore, Duty mediates partially the relationship between Ma-Psy and CSR-SE. The result of the Sobel’s test further supports the mediating effect.

$$z=\frac{\left(-0.529\times 0.272\right)}{\sqrt{-{0.529}^2\times {0.104}^2+{0.272}^2\times {0.083}^2}}=-2.4196(\mathrm{for}\mathrm{Duty}),$$

(4)

Table 8. Regression analysis results for Hypothesis 3 (CSR-Bu).

Analysis	Variables	B		S		β		t-Value
1	Dependent Var.	CSR-Bu
	Ma-Psy	−0.491		0.089		−0.511		−5.502 **
	Nar	0.211		0.088		0.225		2.391 **
	Variables	B	S	β	t-Value	B	S	β	t-Value
2	Dependent Var.	⑤ Top-priority				⑥ Duty
	Ma-Psy	−0.512	0.090	−0.501	−5.700 **	−0.529	0.083	−0.547	−6.344 **
	Nar	0.095	0.089	0.095	1.064	0.107	0.083	0.113	1.292 †
3	Dependent Var.	CSR-Bu
	Ma-Psy	−0.182	0.080	−0.190	−2.274 *	−0.285	0.096	0.296	−2.955 **
	Nar	0.154	0.071	0.164	2.177 *	0.170	0.083	0.180	2.039 *
	Top-priority	0.602	0.072	0.642	8.359 **
	Duty					0.390	0.091	0.393	4.288 **
	Social desirability	0.017	0.019	0.057	0.908	0.049	0.021	0.165	2.306 *
- Top-priority: R2 = 0.676, Adjusted R2 = 0.557, F-statistic (44, 120) = 5.686 **
- Duty: R2 = 0.555, Adjusted R2 = 0.392, F-statistic (44, 120) = 3.405 **

shows regression analysis results for testing Hypothesis 3(CSR-Bu). In analysis 3, the model F value shows a significance (p < 0.01) and adjusted R² values (Top-priority is 0.557 and Duty is 0.392) are larger than those in Table 5. First, in regression ⑤ with Top-priority as a mediator, Top-priority positively affects CSR-Bu (t-value is 8.359, p < 0.01). The direct effect of Ma-Psy is significant (t-value is −2.274, p-value < 0.05), and negatively influences CSR-Bu. Furthermore, the positive direct effect of Nar on CSR-Bu is significant (t-value is 2.177, p-value < 0.05), which is different from the results of CSR-Em and CSR-SE. Second, in regression ⑥ with Duty as a mediator, Duty also positively effects CSR-Bu (t-value is 4.288, p < 0.01), and Ma-Psy has a negative direct effect on CSR-Bu (t-value is −2.955, p-value < 0.01). In addition, Nar has a positive direct effect on CSR-Bu (t-value is 2.039, p-value < 0.05).

The absolute value of the regression coefficient (B = −0.182) of Ma-Psy with Top-priority in analysis 3 is smaller than that (B = −0.491) in analysis 1, and the direct effect of Ma-Psy is significant [84]. Consequently, Top-priority partially mediates the relationship between Ma-Psy and CSR-Bu. The result of the Sobel’s test supports the mediating effect. However, the mediating effect of Top-priority between Nar and CSR-Bu is not present, as the t-value (1.064) of the regression coefficient of Nar in analysis 2 is not significant

$$z=\frac{\left(-0.512\times 0.602\right)}{\sqrt{-{0.512}^2\times {0.072}^2+{0.602}^2\times {0.09}^2}}=-40.7034(\mathrm{for}\mathrm{Top}-\mathrm{priority}),$$

(5)

The absolute value of the regression coefficient (B = −0.285) of Ma-Psy in analysis 3 with Duty is smaller than in analysis 1 (B = −0.491), and there is a significant direct effect of Ma-Psy on CSR-Bu. Therefore, Duty partially mediates the relationship between Ma-Psy and CSR-Bu. The result of the Sobel’s test also supports the mediating effect. Similarly, Duty could partially mediate the relationship between Nar and CSR-Bu, according to Baron and Kenny [84]. The result of the Sobel’s test, however, does not support the mediating effect of Duty

$$z=\frac{\left(-0.529\times 0.390\right)}{\sqrt{-{0.529}^2\times {0.091}^2+{0.390}^2\times {0.083}^2}}=-30.5564(\mathrm{for}\mathrm{Duty}),$$

(6)

3.5. Regression Results 4 (Hypothesis 4)

In order to test the moderating model (Hypothesis 4), all regression analyses were conducted with individual performance-based compensation system as a moderator after mean-centering to remove multi-collinearity.

Table 9. Regression analysis results for Hypothesis 4 (CSR-Em).

Variables		B	β	t-Value	B	β	t-Value
Dependent variable		① CSR-Em
Independent variables	Ma-Psy	−0.217	−0.215	−2.855 **	−0.431	−0.428	−4.346 **
	Nar	0.009	0.009	0.131	0.046	0.046	0.533
	Top-priority	0.578	0.586	8.431 **
	Duty				0.152	0.146	1.622 †
Covariate	SD	0.021	0.067	1.195	0.055	0.177	2.521 **
Moderator	IPBCS	0.084	0.137	2.468 **	0.083	0.137	1.904 *
Interaction term	Top-priority × IPBCS	−0.067	−0.111	−2.028 *
	Duty × IPBCS				−0.068	−0.101	−1.443 †
- Top-priority: R2 = 0.747, Adjusted R2 = 0.649, F-statistic (46, 118) = 7.579 **
- Duty: R2 = 0.586, Adjusted R2 = 0.424, F-statistic (46, 118) = 3.627 **

shows regression analysis results for Hypothesis 4 (CSR-Em). The F value indicate the significance of model and adjusted R² values (Top-priority is 0.649 and Duty is 0.424) are larger than those in Table 6. Individual performance-based compensation system (IPBCS) moderates negatively both the positive relationship between Top-priority and CSR-Em (t-value is −2.028, p-value < 0.05), as well as the positive relationship between Duty and CSR-Em (t-value is −1.443, p-value < 0.10). These results imply that the positive effects of Top-priority (or Duty) on CSR-Em decrease when IPBCS is in place.

As shown in

Table 10. Regression analysis results for Hypothesis 4 (CSR-SE).

Variables		B	β	t-Value	B	β	T-Value
Dependent variable		② CSR-SE
Independent variables	Ma-Psy	0.034	0.034	0.392	−0.176	−0.174	−1.606 *
	Nar	−0.010	−0.010	−0.132	0.016	0.016	0.163
	Top-priority	0.677	0.683	8.577 **
	Duty				0.281	0.269	2.710 **
Covariate	SD	−0.024	−0.077	−1.198	0.011	0.034	0.446
Moderator	IPBCS	0.082	0.134	2.095 *	0.084	0.137	1.729 *
Interaction term	Top-priority × IPBCS	0.005	0.009	0.144
	Duty × IPBCS				−0.069	−0.101	−1.314 †
- Top-priority: R2 = 0.668, Adjusted R2 = 0.538, F-statistic (46, 118) = 5.157 **
- Duty: R2 = 0.497, Adjusted R2 = 0.300, F-statistic (46, 118) = 2.530 **

, adjusted R² values (0.538 and 0.300) are greater than those of Table 7. For CSR-SE, individual performance-based compensation system (IPBCS) negatively moderates the positive relationship between Duty and CSR-SE (t-value is −1.314, p-value < 0.10). However, IPBCS does not moderate the relationship between Top-priority and CSR-SE. These results imply that adoption of IPBCS decreases the positive effect of Duty on CSR-SE, while it does not affect the relationship between Top-priority and CSR-SE.

In

Table 11. Regression analysis results for Hypothesis 4 (CSR-Bu).

Variables		B	β	t-Value	B	β	t-Value
Dependent variable		③ CSR-Bu
Independent variables	Ma-Psy	−0.209	−0.217	−2.587 **	−0.304	−0.317	−3.150 **
	Nar	0.149	0.159	2.097	0.164	0.174	1.960
	Top-priority	0.578	0.615	7.946 **
	Duty				0.380	0.383	4.161 **
Covariate	SD	0.015	0.050	0.800	0.046	0.155	2.168 *
Moderator	IPBCS	−0.002	−0.003	−0.048	0.010	0.018	0.240
Interaction term	Top-priority × IPBCS	−0.069	−0.120	−1.974 *
	Duty × IPBCS				−0.081	−0.126	−1.761 *
- Top-priority: R2 = 0.686, Adjusted R2 = 0.564, F-statistic (46, 118) = 5.610 **
- prima facie Duty: R2 = 0.567, Adjusted R2 = 0.398, F-statistic (46, 118) = 3.355 **

, the F value indicate the significance of model and adjusted R² values (0.564 and 0.398) are larger than those in Table 8. The results show that the IPBCS negatively moderates the positive relationship between Top-priority (and Duty) and CSR-Bu (t-value is −1.974, p-value < 0.05 and t-value is −1.761, p-value < 0.05). These results imply that implementation of IPBCS decreases the positive effect of Top-priority (or Duty) on CSR-Bu.

3.6. The Summary of the Results

Table 12. The summary of the test.

Hypotheses	Independent Variables	Dependent Variables	Mediators	Moderator	Results
1	Ma-Psy	Top-priority			($-)$
	Ma-Psy	Duty			($-)$
	Nar	Duty			($+)$
2	Ma-Psy	CSR-Em			($-)$
	Ma-Psy	CSR-SE			($-)$
	Ma-Psy	CSR-Bu			($-)$
	Nar	CSR-Bu			($+)$
3	Ma-Psy	CSR-Em	Top-priority		Partial
	Ma-Psy	CSR-SE	Top-priority		Complete
	Ma-Psy	CSR-SE	Duty		Partial
	Ma-Psy	CSR-Bu	Top-priority		Partial
	Ma-Psy	CSR-Bu	Duty		Partial
4	Top-priority	CSR-Em		IPBCS	($-)$
	Duty	CSR-Em		IPBCS	($-)$
	Duty	CSR-SE		IPBCS	($-)$
	Top-priority	CSR-Bu		IPBCS	($-)$
	Duty	CSR-Bu		IPBCS	($-)$

summarizes regression analysis results. A CEO with Ma-Psy traits is likely to influence negatively the employee’s perception of ethics and social responsibility and place less importance on CSR activities (Hypothesis 1 and 2), consistent with previous literature. In addition, these results verify the negative relationship between Machiavellianism and psychopathy with CSR in the business context of South Korea. Furthermore, these results show that Machiavellianism is identical to psychopathy and that an employee’s perception of CSR mediates the relationship between a CEO’s Ma-Psy trait and CSR activities (Hypothesis 3).

A CEO with the Nar trait is likely to have a positive effect on the perception of Duty, but this trait is not related to the perception of Top Priority. A narcissistic CEO places importance on being viewed as a “good person” and being admired by others. In addition, the propensity for pursuing honor may lead to attempts to exude a good appearance through utilization of “shallow needs” instead of a “fundamental approach”. As a result, an employee with a Nar CEO may perceive Duty while he or she does not perceive Top Priority. Consistent with previous studies that found Nar to be a positive factor for firms within an innovation-oriented context, Nar is found to have a positive effect on the perception of importance of CSR activities. The same rationale applies to the relationship between Nar and CSR-Bu.

In addition, the relationship between the employee’s perception of ethics and social responsibility and CSR activities is positive for both Top Priority and Duty. Thus, higher perception of Top Priority or Duty leads to the more active engagement in CSR activities. Furthermore, IPBCS negatively moderates the relationship between these perceptions and CSR activities, except for the relationship between Top-priority and CSR-SE. The results are consistent with previous evidence that IPBCS is not necessarily a positive contributor for achieving high performance.

4. Discussion

4.1. Discussion and Conclusions

This study examines the relationship between a CEO’s negative traits and CSR, which serves an important role in developing social capital. The CEO’s negative managerial behavior and socially undesirable decision making influenced by the dark triad traits results in a negative effect on the employee’s perception and performance of CSR activities. The study also finds that an individual performance-based compensation system plays as a negative moderator, although the perception of ethics and social responsibility is highly regarded in the organization. It is useful to consider Machiavellianism, psychopathy, and narcissistic traits in CEOs when we examine the differences in CSR activities across organizations. The business environment has changed rapidly in recent years, and the CEO’s managerial decision affects not only the organization, but also society.

The study contributes to the extant literature by providing academic and practical implications. Academic implications based on the results of this study are as follows: First, the CEO’s negative traits influencing his or her decision making role in an organization are crucial. In a structured organization where conflict between the individual’s desire and organizational benefits increases, there is the need to pay close attention to Machiavellianism when a performance-based compensation system is operated. Sutton [85] also described the problem in his book “Good Boss, Bad Boss” in which the boss is an “asshole”. Bad bosses are selfish, lack empathy, focus on results, and show the tendency to promote “solo star” employees who generally devote themselves to personal performance. These characteristics place the organization at risk especially when they are combined with individual performance-based compensation system. The prevalence of psychopathy in top management would be higher in the business world than in the general public [8]. Further, according to a recent study [86], psychopathy results in higher workload, conflict, and organizational constraints, and these all effect job satisfaction negatively. CSR to employees (CSR-Em) which is one of variables in this analysis is closely related with them. Boddy et al. [46] reported that the psychopaths were seen as being organizational stars and as deserving of awards by others, but they simultaneously showed extreme behavior (bullying, intimidation and coercion). More in-depth research on the Machiavellianism and psychopathy traits needs to be done in the future.

Second, the study finds that the narcissism trait has a positive effect only on ‘CSR activities related to business’ which are composed of protecting consumer rights, product liability, customer satisfaction and active compliance. This result follows from the attributes of narcissism which reveals the intention of being recognized as ‘special person’ and considers visible performance important. Mass media highlights CEO’s negative behaviors, thereby influencing organizational leaders who want to become "star CEOs". Business outcomes are especially important to narcissistic CEOs. They are likely to pay little attention to CSR activities related to employees, environment and society, which they consider not related to business outcomes. As a result, it is even more needed to investigate the relationship between narcissistic leadership and the economic and social performance of an organization considering the ambiguity of narcissism.

Third, many studies experience a difficulty with defining CSR performance. Due to the narrow or biased operational definitions and diverse measurements [87], uncertainty exists concerning the reliability of measurement and application of previous findings. In this regard, the present study considers comprehensive aspects of CSR through the survey of perceptions and CSR activities. Many previous studies have used survey methods without testing reliability and validity and employed external indicators which do not correspond with CSR concepts. The research method and scales used in this study will be a useful reference for future research.

Practical implications based on the results of this study are as follows: Owner-managers are common in South Korean companies and non-executive directors and board members often are not independent of each other. In such contexts, power is centralized in a single individual. Although the patriarchal decision-making structure has advantages in terms of capability of innovation and growth through rapid decision making and aggressive investment, it also has disadvantages in that there is a lack of controls in place that can prevent errors in decision making. Therefore, these results are meaningful to a society such as South Korea because CSR is not only a type of donation, charity, or environmental conservation, but also a form of strategic management. If a CEO features dark triad traits in a patriarchal company, the negative impacts on society will be more severe. However, it is not fair to deprive the CEO’s status, or to impose limitations on their behavior resulting from the potential presence of dark triad traits without anti-social behavior. Therefore, extensive discussion on the personnel management system is needed in order to diminish undesirable impacts of adverse decisions.

The negative aspects of market-based compensation in this study will be a vital clue to redesigning compensation systems. The fact that individual performance-based compensation moderates CSR negatively means that the compensation system gives influence not only within the organization but also society. An issue of the Royal Bank in 2014 having recruited psychopaths as employees was raised by Gavin Palmer (shareholder activist). Palmer related psychopaths in banks to the ethical misbehaviour of those banks [7]. Therefore, a potential negative attribute within leadership can be managed through human resources management. The debate on a rational governance structure should also be expanded to prevent arbitrary decision making resulting from negative traits of CEOs.

4.2. Future Research

This study contributes to the understanding of the negative traits of CEOs and their effects of CSR activities. To broaden our knowledge of CEO behavior and CSR perception and activities, we suggest further research. First, it is important to investigate the CEO’s disposition, particularly his or her negative traits, but it is difficult to survey CEOs directly. The present study controlled for social desirability and employed a peer evaluation method to deal with the problem. Although this approach has no problem in peer-peer agreement and self-peer agreement [68,70,72,73], future research should consider a case study or in-depth qualitative interview method. The scale for the dark triad used in this study was the Dirty Dozen scale [66]. Although this scale has been verified in previous literature, its brief measures may be limited in identifying detailed attributes [88]. In this regard, future research should utilize other scales like SD3 [89] or qualitative methodology.

Second, there is a need to identify relevant management approaches in designing compensation systems and managing leaders’ negative traits. A systematic approach for organizations and efforts identifying positive drivers of corporate social performance are required. Especially, based on previous studies of the narcissistic trait, future research should analyze business contexts influenced by narcissistic leadership that facilitate perceptions of ethics and social responsibility within CSR activities.

Third, the analysis includes only South Korea, but future research should consider a wider range of business systems by considering additional countries. In the same vein, this study has a limitation on the size of samples; thus, further research should use larger data, and should reveal the relationship between negative traits and corporate social irresponsibility. Finally, cross-sector research is needed to determine if there is variation between CEOs in, for example, financial and banking services versus retail and distribution.