Do Narcissistic Managers Prefer Incentive Systems Based on Financial Instruments? An Analysis Based on Choice Experiments

The purpose of this work is to analyse the preferences of Chief Executive Officers (CEOs) in relation to the different components of incentive systems: financial vs. non-financial. The incentive systems could be an instrument for the sustainable development of Firms. Upper Echelons Theory establishes that the traits of executives affect the decision-making processes, and among these traits, narcissism is a potentially influential factor in these processes. Therefore, the extent to which the level of narcissism influences the choice of one instrument or another is also analysed. For this purpose, a choice experiment has been carried out to analyse the preferences of CEOs. The questionnaire developed incorporates both the choices about different systems and the NPI-16 test that allows individuals to be classified according to their narcissistic nature. The main results show that, in general, there is a stronger preference for non-financial instruments than for financial instruments in the design of incentive systems. However, narcissistic CEOs show a clear inclination towards financial incentives that bring them benefits rather than provide incentives.


Introduction
The main objective of management control systems (MCS), as practices designed and implemented in organisations, is to ensure that individuals belonging to that organisation direct their efforts towards achieving the objectives established in the strategy and, consequently, perform better [1,2]. In the configuration of the systems to achieve such behaviour, the design and implementation of incentive systems capable of encouraging the coordination and commitment of individuals and [3,4] motivating and directing their efforts is especially relevant [5,6]. Their effect on both employee and organisational performance [7][8][9] has been empirically demonstrated, and this has led to interest in researching them [5,[10][11][12]. They are defined as the set of monetary and non-monetary rewards granted to employees in an organisation in exchange for work performed [13]. They are also understood as attraction and stimulation mechanisms with the aim of making workers work harder [14].
In addition, there are models that group together the different types of incentives [15], which include base salary, performance, and career management; salary in the event of contingencies; and good working environment, perks, and work-life balance. These are designed according to the norms and cultures of each country [16], so it is easy to find specific features that differentiate them. Although incentives can be classified as intrinsic or extrinsic, direct or indirect, and financial or non-financial [17][18][19][20], in this study we focus on the distinction between financial and non-financial ones. Financial incentives include, for example, wages and salaries, performance pay, maturity curve, merit pay increases, compensation for experience and knowledge, pensions, insurance, and share purchase options [21][22][23][24][25]. Non-financial incentives can include promotion, holiday, training, sporting events and social activities, time off, special distinctions and privileges, staff canteen, medical assistance, transport, flexible working hours, etc. [22,[26][27][28][29].
The selection of non-financial attributes can contribute to a more sustainable development of the firms. For example, by considering circumstances not only related to profit, fairer and more decent work can be promoted in line with goal number 8 of sustainable development. In this sense, several authors have highlighted the role of employees in the sustainability development of a firm [30,31].
The main agent in this process of designing incentive systems tends to be the Chief Executive Officer (CEO), given that they have the leading responsibility for defining the company's strategy and objectives and for guiding the actions of the middle managers who implement these guidelines. The effect of their dominant position in the company, especially in those where they have greater discretion, has aroused considerable interest among the scientific community. The decisions and behaviour of senior executives have been found to be influenced by their experiences, preferences, and attitudes [32,33]. Specifically, the personality traits of CEOs have become a relevant topic for research into decision-making processes [34][35][36][37]. However, as set out by Schaeffer and Dossi (2014, p. 219), "theoretical knowledge and empirical findings on the importance of senior executives in the design, perception, and use of management control systems are still scarce". Moreover, Abernethy et al. [38] also point out that it is interesting to study the effect of the behavioural trends and personal traits of a leader on the design and application of the control mechanisms to influence the behaviour of their subordinates.
This study focusses on analysing the preferences of CEOs when designing incentive systems, as it has been found that the traits of CEOs help explain organisational results [32].
Among these traits, narcissism has emerged as an influential factor in decisionmaking [39,40]. Due to its very nature, compared to other personality traits, it can especially drive CEOs to take measures that defy convention, as a way to attract attention and applause, thus affecting the performance of organisations. This study is also motivated by the apparently contradictory results found in this line of research to date [41][42][43], as well as by the call for research made by Chatterjee and Hambrick [34] in relation to studying the effects of narcissistic leaders on people and organisations. The work therefore tries to determine whether there are differences between CEOs with different levels of narcissism in terms of their incentive system preferences; more specifically in the establishment of systems based on financial attributes compared to non-financial attributes.
The work is structured as follows: the first section contains a discussion of the influence of the traits of CEOs on the company in light of the Upper Echelons Theory. The hypotheses being studied are then developed in Section 2. Section 3 contains the methodology used to test the hypotheses. The results are discussed in Section 4, before the work ends with the main conclusions reached.

A CEO's Influence on the Organisation, and Narcissism
Upper Echelons Theory (UET) establishes that the traits of senior executives influence a company's decision-making, at both an individual and strategic level [44]. These traits include cultural background, personality, preferences, attitudes, etc. [32]. It is assumed that CEOs, as representatives of the highest organisational levels, make decisions based on their personal interpretations of events or situations that occur [45] and that these decisions affect the company's results. Researchers using UET as the basis of their studies have found relationships between organisational results and certain traits of CEOs, based on both personality and demographics [33,37,46]. An example is the five personality  [36,47,48]; core self-evaluation [37]; locus of control [49,50], need for achievement [51]; risk tolerance [52]; and narcissism [53,54].
Determining and understanding which personality traits of CEOs affect decisionmaking and leadership behaviour, and therefore the operation and performance of the organisation, is undoubtedly of value, as shown in the literature [37,55,56].
Among the personality traits of CEOs studied, narcissism has attracted special attention from researchers [57]. Although narcissism has been seen as dysfunctional behaviour since it first began to be studied [58][59][60][61][62][63], being considered a personality disorder in the clinical and psychiatric literature, through the works of Emmons (1987) [62] and Raskin and Terry (1988) [64] it has also been considered to be a personality dimension that individuals may have to a greater or lesser degree [65]. In this study, we focus on this latter aspect.
According to Campbell et al. (2011, p. 4) [66] "narcissism is a relatively stable individual difference consisting of grandiosity, pride, and self-esteem". A narcissistic individual "is characterised by positivity, 'specialness' and uniqueness, vanity, a sense of entitlement, and a desire for power and esteem"; their interpersonal relationships "contain low levels of empathy and emotional intimacy (...) there are (often numerous) superficial relationships that can range from exciting and engaging to manipulative and exploitative"; and finally, "they have strategies for maintaining inflated self-views (for example, they seek out opportunities for attention and admiration, brag, steal credit from others, and play games in a relationship)".
If things work out as they wish, their self-esteem is higher and their positivity increases [67], as would happen for any individual. However, if they do not work out, the situation could cause their behaviour to become more radical, displaying aggressiveness, anxiety, and depression [68].
According to various authors [69,70], it is considered a characteristic of people who achieve high levels of responsibility in organisations. As stated by Kets de Vries and Miller (1984, p. 32) [50] "it is to be expected that many narcissistic people, with their need for power, prestige, and glamour, will end up seeking leadership positions. Their sense of drama, their ability to manipulate others, their ability to establish quick and superficial relationships, is useful to them". CEOs, given their position of power in the company, are treated as special people, leading to higher self-esteem [71] that results in the satisfaction of their needs and the possibility that they may have a certain degree of narcissism [42].
Other authors state that a narcissistic CEO may develop two types of behaviour, known in the literature as the dark side [41,57,66] and the bright side [57,66,72]. Different studies have detected examples of behaviour on the negative side of narcissism, such as, lack of empathy and sensitivity to criticism [66], inability to learn from mistakes and the destruction of the resources necessary for long-term success [73], loss of relationships, exploitation of subordinates [74], destruction of the superior-subordinate relationship [41], having low-quality relationships [75], risky behaviour due to overvaluing profits [76], making unnecessary and risky decisions [41], use of unethical methods [42,75], and destruction of the good atmosphere in the organisation [41].
On the other hand, the positive or bright side of a narcissistic leader is revealed as their ability to inspire their followers to achieve targets, contributing to them being seen as effective leaders [57]. In this regard, there are studies that have revealed that this type of leader is considered more charismatic [77]. This positive side is also seen as improving their good image as a leader [41], the good perception of their performance by others [77], the recognition and admiration of their followers and subordinates for their vision of the future [41], the perception of a positive relationship with the professional success of their followers [43], their passion and innovation [66], and their ability to act in turbulent times or times of crisis at the company [41,78].
Faced with this duality, Kollmann et al. [39] recognise that narcissism is an even more complex trait than previously imagined. For years, there has been a demand for more research to determine the effects of narcissism, [65,66,79,80], whose results must be taken into account in the configuration of the company's managerial organisational chart in order to guarantee the implementation of suitable practices within the management control system.

Development of the Hypothesis
In this study, the preferences of individuals holding the position of CEO are analysed in decision-making related to the incentive system. Specifically, the effects of the level of narcissism on the aforementioned decision-making are observed. As indicated above, narcissism may have positive and negative effects, so hypotheses cannot be made about a specific positive or negative relationship.
Narcissistic CEOs generally do not take into account factors that motivate other people. They do not appreciate or encourage employees who show initiative, they may cause real or perceived injustices and confront individuals, they do not give their employees the opportunity to voice complaints or opinions, and they may make their staff work too hard, regardless of the personal cost [81].
It is possible that narcissistic CEOs, who are characterised by low emotional stability and a need for external recognition, do not take into account those around them [82], and they also may end up demonstrating exploitative behaviour [55,[83][84][85], which could be hidden by convincing their subordinates to make sacrifices in pursuit of big goals [81]. This would mean employee recognition and reward processes being used to create a greater benefit for the CEOs. As a result, the design of these systems could be affected by the narcissistic tendency of a CEO [82].
This type of leader also has a positive effect in terms of stabilising the work atmosphere and is a force for cohesion and coordination. Their constant and fearless pursuit of success builds courage and encourages their subordinates when faced with challenges [41]. In contrast to the situation above, a narcissistic CEO might be motivated to be fair, even generous, as long as they can gain benefits and status from this. At the same time, if they identify with or feel responsible for the subordinates for whom they are responsible, they may be concerned about the treatment they receive [81]. Given that a narcissistic CEO needs followers and acclaim, they may design an incentive system that is beneficial for employees in order to generate loyalty and achieve their desired goal.
As a result, the questions to be researched in this work are, firstly, the preferences among CEOs for financial or non-financial instruments (H1); secondly, whether a CEO with narcissistic traits behaves differently when deciding on the incentive systems compared to those who do not possess these traits (H2); and thirdly, whether this type of leader shows a tendency to establish work incentive instruments in line with aspects less oriented towards the employee and more related to their own interests (H3). The hypotheses are presented as follows: Hypothesis (H1). CEOs have preferences between financial and non-financial instruments in the design of an incentive system.

Hypothesis (H2). The CEO's level of narcissism influences the choice of incentive system instruments.
Hypothesis (H3). The most narcissistic CEOs choose incentives that best serve their own interests.

Data
The data used in this study have been obtained through a questionnaire that includes both the choice experiment and the personality tests. This allows us to group together the preferences and differentiate them according to the degree of narcissism of the respondents. This questionnaire, initially tested on six pilot participants to detect possible flaws in its design, was completed during February 2020 by 91 students taking accounting and management subjects in the Degree in Business Administration and Management (https://docs.google.com/forms/d/1zKFHnJqf9JT-ia1i8AzPWkh4HQNm1 BLwIiX1mnFdS1Q/edit). The proportion of men is 51.16% of the total sample. The age of the participants is between 20 and 23 years old.
As mentioned by various authors [22,[86][87][88], the main reasons for selecting students from this field were as follows: firstly, there is extensive literature in favour of using accounting and management students as substitutes for their professional counterparts [22,87,[89][90][91], and, secondly, it has been proven that, with regard to management decisions, there are no major differences between experienced and inexperienced professionals.

Choice Experiment
Since the earliest use of choice experiment models in the field of transportation and marketing, multiple applications have been developed in field of behavioural economics, finance, and environmental assessment; in the field of health economics; and in the agrifood area [92][93][94][95]. However, the use of this methodology in the accounting area is still incipient, and it is only recent studies that propose its use in this field [22,87].
In the accounting field, two papers applied this methodology [22,87]; however, in the field of finance, several papers are implementing this approach, for example, in the analysis of risk aversion and willingness to pay to avoid renewable energy externalities [96], the use of discrete choice to analyse trading bankruptcy [97], or the flexibility in agricultural investment decision [98]. In general terms, many papers have been published since the survey's publication review in the Handbook of Behavioural Finance [99].
We believe that, in order to analyse the preferences of CEOs with regard to a company's remuneration system, the application of this type of model is appropriate, as it allows us to analyse the possible decisions that will be made by CEOs [22,87]. The followed steps to apply the choice experiment methodology are the indicated ones in the standard approach [100,101].
Specifically, the attributes shown in Table 1 have been used to configure the incentive system. The full set of comparisons, given the attributes selected and the number of levels, amounts to 1024 (4 × 4 × 2 × 2 × 2 × 2 × 2), which in practice is an impossible number of comparisons for an individual to respond to. The Stata "Dcreate" module was used to carry out the design. This allows a fractional design to be generated to reduce the number of comparisons with a level of efficiency [102]. Specifically, the module uses Federov's modified algorithm to create an efficient design [95]. Finally, eight choice sets were generated and incorporated into the survey (D-efficiency: 0.4489). Each choice set consists of two alternatives and the additional option of neither of the above. As a result, the experiment is based on choosing eight comparisons that each contain three options. Table 2 shows an example of a choice set. Select the option modified algorithm to create an efficient design [95]. Finally, eight choice sets were generated and incorporated into the survey (D-efficiency: 0.4489). Each choice set consists of two alternatives and the additional option of neither of the above. As a result, the experiment is based on choosing eight comparisons that each contain three options. Table 2 shows an example of a choice set.

Logit Model
The model used to assess the preferences of CEOs for remuneration incentive systems is conditional logit. Conditional logit is based on the random utility model [100,101,103,104], which assumes that the utility function of each individual is the sum of two components, a deterministic part that can be obtained as a function of the factors that affect the utility of the individuals and another random part, not directly observed, which is considered stochastic. Thus, given a sample of N individuals with the option of choosing between J alternatives on T occasions, the utility of individual n, derived from the choice of alternative j on occasion t, is as follows: where is the vector of specific coefficients of each individual, is the vector of the observable attributes of individual n and alternative j on choice occasion t, and is the random term that we assume to be an independently and identically distributed extreme value. Therefore, the probability of choosing alternative j in choice t is given by the following expression: The above expression is the conditional logit formula [10]. The econometric specification used in this work takes the functional form of derived from individual n for alternative j in choice set t, which can be defined as follows: where 0 is associated with the current situation (ACS), that is, not choosing either of the two proposed incentive systems, and is the marginal utility associated with each attribute provided by the specific incentive system. modified algorithm to create an efficient design [95]. Finally, eight choice sets were erated and incorporated into the survey (D-efficiency: 0.4489). Each choice set cons two alternatives and the additional option of neither of the above. As a result, the e ment is based on choosing eight comparisons that each contain three options. Ta shows an example of a choice set.

Logit Model
The model used to assess the preferences of CEOs for remuneration incentiv tems is conditional logit. Conditional logit is based on the random utility m [100,101,103,104], which assumes that the utility function of each individual is the s two components, a deterministic part that can be obtained as a function of the factor affect the utility of the individuals and another random part, not directly observed, w is considered stochastic. Thus, given a sample of N individuals with the option of cho between J alternatives on T occasions, the utility of individual n, derived from the c of alternative j on occasion t, is as follows: where is the vector of specific coefficients of each individual, is the vector observable attributes of individual n and alternative j on choice occasion t, and the random term that we assume to be an independently and identically distribute treme value. Therefore, the probability of choosing alternative j in choice t is given b following expression: The above expression is the conditional logit formula [10]. The econometric specification used in this work takes the functional form of rived from individual n for alternative j in choice set t, which can be defined as follo = 0 + 1 + 2 + 3 + 4 + 5 + 6 + 7 + 8 + where 0 is associated with the current situation (ACS), that is, not choosing either two proposed incentive systems, and is the marginal utility associated with ea tribute provided by the specific incentive system. modified algorithm to create an efficient design [95]. Finally, erated and incorporated into the survey (D-efficiency: 0.4489) two alternatives and the additional option of neither of the ab ment is based on choosing eight comparisons that each con shows an example of a choice set.

Logit Model
The model used to assess the preferences of CEOs for r tems is conditional logit. Conditional logit is based on t [100,101,103,104], which assumes that the utility function of ea two components, a deterministic part that can be obtained as a affect the utility of the individuals and another random part, n is considered stochastic. Thus, given a sample of N individuals between J alternatives on T occasions, the utility of individual of alternative j on occasion t, is as follows: where is the vector of specific coefficients of each individu observable attributes of individual n and alternative j on cho the random term that we assume to be an independently and treme value. Therefore, the probability of choosing alternative following expression: The above expression is the conditional logit formula [10 The econometric specification used in this work takes the rived from individual n for alternative j in choice set t, which = 0 + 1 + 2 + 3 + 4 + 5 where 0 is associated with the current situation (ACS), that i two proposed incentive systems, and is the marginal util tribute provided by the specific incentive system.

Logit Model
The model used to assess the preferences of CEOs for remuneration incentive systems is conditional logit. Conditional logit is based on the random utility model [100,101,103,104], which assumes that the utility function of each individual is the sum of two components, a deterministic part that can be obtained as a function of the factors that affect the utility of the individuals and another random part, not directly observed, which is considered stochastic. Thus, given a sample of N individuals with the option of choosing between J alternatives on T occasions, the utility of individual n, derived from the choice of alternative j on occasion t, is as follows: where β n is the vector of specific coefficients of each individual, x njt is the vector of the observable attributes of individual n and alternative j on choice occasion t, and ε njt is the random term that we assume to be an independently and identically distributed extreme value. Therefore, the probability of choosing alternative j in choice t is given by the following expression: The above expression is the conditional logit formula [10]. The econometric specification used in this work takes the functional form of U njt derived from individual n for alternative j in choice set t, which can be defined as follows: where β 0 is associated with the current situation (ACS), that is, not choosing either of the two proposed incentive systems, and β k is the marginal utility associated with each attribute provided by the specific incentive system.

Level of Narcissism
The level of narcissism is measured using the NPI-16 inventory, which has four items for each of the four sub-dimensions conceptualised by [62]: leadership/authority, selfabsorption/self-admiration, superiority/arrogance, and exploitativeness/entitlement. The NPI is currently the main tool for measuring narcissism [42].
The test has 16 standard questions of NPI. Each question takes the value 1 when the survey respondents select a response that belongs to the narcissistic personality according to NPI instructions. Then, we calculate the sum of all answers with value 1 and obtain the mean value of the sample. The instruction of the test recommends the use of the mean value. Individuals have a high or low level of narcissism depending on whether they score above or below average. Table 3 are the results obtained after applying the logit model to the full sample. Specifically, column 1 shows the name of the attribute and its levels, column 2 contains the value of the regression coefficients, followed by the standard error (column 3), the z value (column 4), and finally, the significance shown in the p-value. As can be observed, all coefficients have a positive utility, that is, they indicate that they have a positive influence in the preferences of the CEO's point of view within the overall incentive system, except for fixed remuneration. However, it should be noted that RM and RV are not significant at any of their levels, which shows that financial indicators do not report a significant utility.

Shown below in
Moreover, in terms of the attributes we can call non-financial, RE, SERV, and HOR show significant and positive utilities, and RE and HOR are positioned as the most important attributes when establishing incentive systems. In order of importance, HOR (z = 2.82) is in first position, followed by RE (z = 2.18) and finally, SERV (z = 1.87), with non-financial instruments being preferred to financial instruments.
In order to analyse whether the degree of narcissism influences preferences in the establishment of incentive systems, the above regression is carried out for those with low levels of narcissism (Table 4) and those with high levels of narcissism (Table 5). Overall, 41.17% of the total sample present a certain level of narcissism.  Specifically, the group formed by individuals with a low score in narcissism shows a clear tendency towards the establishment of non-financial instruments. Proof of this is the significant positive result for the different non-financial attributes. Specifically, the attribute on which they place most importance is HOR (z = 3.81), followed by RE (z = 2.72), SERV and RESP (z = 2.71), VAC (z = 2.51), and PE (z = 2.25). In contrast, the financial attributes are not significant.
In contrast, the results change when repeated for the group characterised by a high degree of narcissism (Table 4). This group places greater importance on financial instruments compared to non-financial instruments. Thus, RV is significant and positive at all levels, which indicates that variable remuneration is shown as a motivating indicator in the system, the 10% level being the most relevant for the whole sample (z = 2.91). All of the non-financial instruments have negative utilities, although none of them are significant.

Discussion
The establishment of incentive systems that fall into one of the two broad groups known as financial and non-financial incentives has been analysed in numerous studies on management [10,26,[104][105][106] and in works that only analyse financial incentives [12,21,107].
Specifically, the results obtained in this work highlight that CEOs are more interested in establishing non-financial measures than financial measures. This result is in keeping with studies that reveal that the establishment of incentives in line with financial aspects may have repercussions over a longer period of time and higher costs compared to nonfinancial aspects [48,104]. Moreover, the results obtained in the study by [26], analysing the satisfaction of workers in the food sector, show a greater predisposition to establish systems with financial incentives, which is a supplementary vision to that obtained in this study, as it is carried out from the perspective of the workers and not the CEOs. However, splitting CEOs by level of narcissism reveals differing results with regard to incentive instruments. This supports the argument that a CEO's personality has a notable influence on decision-making and, more specifically, on incentive systems, as discussed in several studies [11,74,81,83].
More specifically, a narcissist makes choices on incentive systems that firstly bring a clear benefit for themselves, and then provide an incentive. A clear preference for profit-based variable remuneration systems can be observed. They also tend not to choose instruments that deliberately provide incentives or a clear benefit for workers if they do not receive anything in exchange (rewards, extra services, etc.). This behaviour is typical of narcissists compared to non-narcissists, who are more prone to taking into account the welfare of the workers with regard to their motivation. This is reflected in the choice of more cost-free instruments for the worker, in which the company does not receive anything in return and has not previously received anything (leave, extra services, recognition, etc.).
Consequently, this work reinforces the research that examines how the personality traits of CEOs influence their business actions [34][35][36][37].

Conclusions
This work aims to analyse the preferences of CEOs in relation to incentive systems. Specifically, both the financial and non-financial characteristics of incentive systems are analysed. In order to analyse whether this personality trait influences their preferences for incentive systems, an analysis was also carried out based on the degree of narcissism shown by the CEO. This study contributes to the literature as follows: firstly, through its application of choice experiment models to work incentive systems with both financial and non-financial attributes, and secondly, through including a CEO personality trait in the analysis.
The main results obtained show that in terms of financial attributes, there is a preference for the establishment of variable remuneration systems compared to fixed remuneration systems. Moreover, there is a greater tendency to establish systems based on non-financial remuneration, such as the establishment of flexible working hours, recognition for work performed or offering services such as childcare, a staff canteen, training, etc. However, it is important to note that the results differ when carrying out the analysis distinguishing individuals according to their level of narcissism. Individuals with high levels of narcissism prefer to establish systems with performance-based variable remuneration, whereas those who score low in narcissism favour systems based on nonfinancial instruments.
This study reveals how the traits of a CEO can influence decision-making and, more specifically, how a narcissist prefers to implement measures that are designed more for their own benefit than to reward their staff. This behaviour could be justified by motivation for the objective of the narcissist: which is looking for their own benefit. All of this is of particular importance in understanding business decision-making processes and their consequences. The main limitation of the study is based on the selection of the sample used; however, the work presents the future line of investigating how different personalities influence business choices.