When Narcissus Became a CEO: CEO Narcissism and I ts Effect on Earnings Management

This study examines the role of CEO narcissism to earnings management. Samples are non-financial companies listed i n Indonesian Stock Exchange. Three indicators are used to reflect the CEO's narcissistic nature: CEO photograph, CEO publicity


Introduction
Public firms are competing to show the best performance through earnings. Higher earnings number is one of the benchmarks for investors. Companies that consistently generate profits can maintain their business continuity and have good prospects.
The benchmarking process through earnings can trigger earnings management. This action is carried out by choosing specific accounting policies to increase and decrease the company's profits as desired. Not infrequently, earnings management is seen as unethical behavior for some people. Earnings management practices are believed to reduce transparency in financial statement information, even misleading, to harm the company, investors, and other stakeholders. In addition, earnings management is often regarded as opportunistic behavior. However, in essence, earnings management activities are related to management's discretion in presenting financial statements (Schipper and Vincent, 2003). They can use their superior information (Healy and Wahlen, 1999) and their discretion for personal gains (McManus, 2016;Harris and Bromiley, 2007).
In the case of Indonesia, earnings management is presented at PT. KAI in 2006. One of top management refused to sign the financial statement because of net income presented higher than it should be (Tempo, 2006). In other cases, the restatement of financial statement of Garuda Indonesia is other earnings management case. Considering both cases, factors of earnings management are crucial to be discussed.
Our main focus is on the role of narcissism. Earnings management activities mostly are dominated by financial motive, include tax avoidance and producing higher stock price. The study to connect personality of the most dominant position behind the financial statement is needed to be explored. Our study believe that narcissism, a person with a narcissistic personality has a strong desire for power (Emmons, 1987), is one of the dominant factors presenting earnings management. Supporting by Amernic and Craig (2010) that discuss those narcissistic CEOs tend to select accounting choices to best represent the company's financial condition.
Narcissistic CEOs inspire and succeed when companies need strong innovation (Zhang et al., 2017). By using income increasing strategy, the CEO seeks to achieve self-improvement in search of social identity and praise. Self-admiration tendencies are stimulating forces to avoid presenting bad news to the market (Olsen et al., 2013). As a result, a narcissistic CEO tends to take actions that strengthen their self-image and maintain their ego ideal to obtain higher compensation. Therefore, it can be concluded that CEO characteristics can influence leadership style, and the effect of his leadership style will flow through management to all parts of the company, thereby affecting organizational achievement (Campbell and Foster, 2007;O'Reilly III et al., 2014).
Connecting the relation between earnings management practices and narcissistic CEO, we are motivated to explore the human side of psychology as a motive for earnings management. This research is interesting, considering that literature on CEO narcissism still has much room to be analyzed using the Indonesian case. Leuz et al. (2003) present evidence that firms in Indonesia are listed to produce a lower disclosure index, more concentrated ownership, and lower investors right. Additionally, they also showed that Indonesia has lower importance on the capital market. Those characteristics positively affect earnings management. Finding other factors is needed.
Our study is expected to offer a portrait with a different point of view on how psychological factors affect earnings management. In addition, research conducted by Buchholz et al. (2020) shows that narcissistic CEOs tend to carry out earnings management.
In presenting the idea, we categorize our section into five parts. The first section introduces the idea. This paper discusses the literature review and methodology in the second and third sections. The fourth section presents the results and discussion. In the last section, conclusion of this study is presented.

Theoretical framework and hypotheses development
The upper echelons theory initiated by Hambrick and Mason (1984) states that top management's characteristics, background, and values can influence a company's performance and strategic decisions. In updating the paper, Hambrick (2007) reveals that the upper echelons theory has two interconnected cores (1) executives usually act on a personal interpretation of the strategic situation at hand, and (2) the personal interpretations made by executives reflect their own experiences, values, and personalities. In this paper, we connect narcissism behavior and earnings management to give evidence that personality traits can stimulate accounting practice. Bassyouny et al (2020) examines the theory using computerized textual analysis. They found that both older, female and financial expert CEOs express a less positive disclosure tone, but more narcissistic CEOs display a more positive tone. The study encourages future work in analyzing narcissism and financial statement.

Earnings management
Earnings management occurs when managers use a variety of considerations in presenting a financial statement that can mislead stakeholders about the company's performance (Schipper, 1989;Healy and Wahlen, 1999). In practice, managers can manage earnings using various methods to influence operating, financing, and investment decisions directly. The form of managing earnings that managers can do is related to accounting decisions or policy choices.
Using positive accounting theory ideas, Zimmerman (1986, 1990) define three hypotheses that underlie the selection of accounting methods for management. One of them is the debt covenant hypothesis. According to the debt covenant hypothesis, in a state of ceteris paribus, the closer the maturity of the debt to be paid, the more likely it is for managers to choose favorable accounting policies by changing profit recognition from what should be recorded in the future period to this period. That way, the company's leverage ratio becomes smaller.

CEO narcissism
In essence, narcissism lies at the heart of leadership (Vries. 2004;Kernberg, 1975;Kohut, 2009). A person with a narcissistic personality has a strong desire for power (Emmons, 1987). Usually, narcissistic CEOs tend to engage in risky initiatives because they need constant attention to affirm their increased positive self-view. On the other hand, a narcissistic person usually gets an important position in the company because his innate personality, which refers to achievement, charisma, and self-confidence (Campbell and Campbell, 2009), helps them excel in a competitive work environment. Patel and Cooper (2013) discuss while narcissistic CEOs are less likely to protect against potential shocks, they are adept at helping firms recover from such shocks. Supporting the idea that narcissism gives benefit to CEOs, O'Reilly et al. (2014) finds that more narcissistic CEOs who have been with their firm longer receive more total direct compensation (salary, bonus, and stock options), have more money in their total shareholdings, and have larger discrepancies between their own (higher) compensation and the other members of their team. Maintaining the compensation, Olsen et al. (2014) show that firms with narcissistic CEOs have higher earnings-pershare and share price than those with nonnarcissistic CEOs.

Influence of CEO narcissism on earnings management
In general, someone with extreme narcissism tends to weaken corporate governance (Grant and McGhee, 2013). It has the potential to weaken the control of the board of commissioners (Zhu and Chen, 2015) to achieve power and self-interest, namely obtaining higher status or compensation. Narcissistic CEO produce higher earnings management because the CEO tends to manage earnings to fulfil their ego (Kontessa, Brahmana and Tong, 2021). Additionally, Bassyouny et al (2020) also present findings that narcissistic CEO display more positive narrative disclosure tone to present optimistic signal to the markets. Narrative plays important role in sending information from managers to external parties that build confidence to make collaboration to the companies in the future.
As a result, highly narcissistic CEOs are stimulated to apply the accounting choices that are most beneficial to the company in order to create a favorable financial situation. Therefore, narcissistic CEOs stimulate earnings management based on the above framework, the following hypotheses can be formulated: H1: CEO narcissism has a significant positive effect on earnings management

Research method
Our study uses listed firms' data from 2018 until 2020, except financial sector companies listed. In selecting the sample, this study used a purposive sampling technique which has the following research criteria: 1. Companies issue financial statements and annual reports that have been completely audited during the study period. 2. The financial reporting and annual reporting periods end on December 31. 3. Financial statements are presented in rupiah. 4. Companies that did not experience CEO turnover during the study period. 5. All non-missing variables are used.
From the results of sample selection, we get 906 firms' years of listed firms in Indonesia. After deleting firms with (1) negative equities, and (2) ROA and DAR with more than 100%, we process 840 firms' years.
In measuring earnings management, this paper employs discretionary accruals using Dechow et al. (1995). In this paper, the measurement of CEO Narcissism is calculated using indicators developed by Raskin and Terry (1988). They built the psychometric Narcissistic Personality Inventory (NPI) scale to test narcissism as a measurable personality dimension. However, NPI Score data is challenging to obtain because it requires direct interaction with the CEO. Therefore, this study uses an alternative that is often used in narcissism research studies, namely the unobtrusive measure to determine how high the level of CEO narcissism is. In unobtrusive data collection, the subjects, in this case, the CEO, were not aware that they were being studied, so this study was not affected by their responses or behavior.
The basis for selecting the CEO Narcissism indicator follows two main criteria set out by Chatterjee and Hambrick (2007), namely (1) CEO's volition which means it is based on the CEO's original personality without being influenced by external pressures, and (2) each type of indicator needs to represent one or more personality traits. However, the CEO narcissism indicator used in this study was also adjusted to the data available in Indonesia where the data is possible to find. Below is presented table 1 which contains three indicators to determine the score of CEO Narcissism.  Chatterjee and Hambrick (2007) verified that CEOs pay close attention to the information in financial reports, and in particular, they have the final authority over how their portrait presented. It shows that the CEO's volition criteria have been met in selecting this indicator. CEO's photo can be seen as evidence that they are the most important person in the company (Rijsenbilt and Commandeur, 2013) and shows an attitude of superiority and high status (Lin et al., 2020).
Specifically, the study rates each CEO's photo as follows: 1. The annual report does not contain a photo of the CEO; 2. CEO pictured with other executives; 3. The CEO has photographed alone, and his/her photo occupies less than half the page; 4. The CEO is photographed alone and fills at least half the page; and 5. The CEO was photographed alone, and his/her photo-filled the entire page.
The second indicator, namely the number of CEO publications, searches for news trends about CEOs through google trends starting from the 2018-2020 period. CEO publicity can be calculated by dividing the number of CEO names mentioned in the mass media by company publicity. Finally, this study considers CEO's social media. Their narcissistic tendencies are closely related to activity on social media. Andreassen et al. (2017) stated that social media opens up opportunities for CEOs to show their ambition and success through likes and positive comments from a broader mass. Inspired by previous research studies (Davenport et al., 2014;Bergman et al., 2011), the researcher adopted, modified, and created a new indicator of CEO narcissism, namely the number of social media accounts owned by CEOs with a scale ranging from one (1) up to (3), namely: 1. CEO only has one social media account or none at all; 2. CEO has two social media accounts; and 3. The CEO has three or more social media accounts.
Our study uses the principal component analysis (PCA) method to determine the CEO narcissism score. PCA reduces dimensions in data sets by transforming large data sets into small data sets without losing valuable information contained in the data and maintaining data integrity. In other words, PCA reduces the data dimensions from the three CEO narcissism indicators to one overall CEO narcissism variable. According to Rijsenbilt (2011), some criteria are able to be used by (1) using the Kaiser criterion (Kaiser, 1960) where this rule only maintains factors with eigenvalues more excellent than one, (2) looking at the scree plot (Cattell, 1966) to find behavioral changes in the variance plot where the eigenvalues decrease subtly and appear flat to the right of the plot, (3) looking at the cumulative variance table, and (4) the extent to which the solution can be interpreted.
The following is presented in table 2, which summarizes the operationalization of the variables in this study. Return on asset ratio as a proxy for profitability Cash Flow (CFO) Standardize of operating cash flow

Data analysis method
This research method uses panel data regression in analyzing the data and interpreting the research results. Before performing panel data regression, it is important to carry out classical assumption tests to ensure and validate that the data is free from regression problems while at the same time detecting deviations in the regression model. The regression model can be formulated through the following equation: DAC %,' = β 0 + β 1 DCPh i,t + β 2 DCP i,t + β 3 SIZE i,t + β 4 DAR i,t +BIG4 i,t + β 6 AUDIT i,t +β 7 ROA i,t +β 8 CFO i,t +ε i,t DCP is a dummy variable of CEO publicity. DCPh represents dummy of CEO photograph. DCP and DCPh are 1 when we find the photograph and publicity in google trend, otherwise zero.

Results and discussion Descriptive statistics and correlation analysis
Descriptive statistics describe a general description of the characteristics of the data presented in Table 3. The descriptive statistical tests show that the dependent variable, namely discretionary accruals (DAC), has a value range from -0.388 to 1.606. The minimum value indicates the practice of earnings management by reporting earnings that are lower than they should be (income decreasing), while the maximum value indicates the practice of earnings management by reporting earnings that are higher than it should be (income increasing).
The descriptive statistics on the DAC variable also show an average value of 0.007 and a standard deviation of 0.124. In addition, the average value smaller than the standard deviation value indicates that the average value cannot represent all DAC variable data because the data tends to be heterogeneous and the level of deviation of the observed value from the average tends to be large.
Then, the independent variable in this study, CEO Narcissism, has a range of values ranging from -2.428 to 4.769. The descriptive statistics on the NARCISS variable also show an average value of 0.021 and a standard deviation of 1.059. Higher value represents more CEO narcissism.
Specifically, this paper uses Pearson correlation analysis or Pearson Product Moment (PPM) correlation. Pearson correlation values ranging between -1 and +1 indicate that the strength of the relationship between variables is very strong, and vice versa. In this case, the positive and negative signs indicate the relationship between variables. The following is presented in Table 4, which shows the correlation test results between variables.  Table 5 show the panel data regression test results using fixed effect model. The baseline model 1 in Table 5 shows that the coefficient of determination adjusted R2 is 0.042. NARCISS has an insignificant effect on the dependent variable DAC. One-tailed test, ***, **, * represent the significant level at 1%, 5%, and 10% respectively. Source: Processed by authors Using alternative measurement (Model 2), CEO Publicity (DCP) positively affects discretionary accrual. The more popular CEO tend to present income-increasing strategies. After controlling Covid-19 year (DCOVID) (Model 3), we also find the same results. We use 2020 as a Covid-19 year.

Panel data regression analysis
The supplementary analysis is expected to provide additional insights (see Table 5 model 2 and  3). Changes in the CEO Narcissism indicator measurement can be seen in Table 6. Table 5 model 2 shows that CEO publicity positively stimulate earnings management. With more publicity, firms tend to be aggressive in presenting earnings. DCP is significant at 10%. using additional variables, we found different findings, the interaction between narcissism indicators and profitability. Table 5 model 3 give evidence that DCP affect the relation between ROA and earnings management. DCP is able to change the sign of ROA coefficient. It is significant at 1% level. The situation can be interpreted that higher ROA tends to use income increasing strategy, but for firms with CEO publicity, higher ROA reduce earnings management. It can happen because of the trust from the market, which believes that famous CEOs prefer to avoid earnings management practices when they have higher ROA. In this case, the determinant of narcissism is external to the company, namely public trust in CEOs who have high levels of popularity. Table 6. Summary of changes in measurement of CEO narcissism indicators Initial indicator Modification indicator CEO photograph (CPh) To assess how big the CEO's photo is, this study determines a scale range from 1 to 5, explaining the criteria for a scale of 1-5.
The measurement scale on the CPh indicator is changed to a dummy variable with the following conditions: 0: a scale that is worth 1-3 1: a scale that is rated 4-5 CEO Publicity (CP) A ratio with the formula measures CEO publicity.
The ratio to determine the CP indicator is changed to a dummy variable with the following conditions: 0: ratio is worth 0 1: ratio value > 0

Overall result analysis
In this study, the hypothesis is formulated to obtain empirical evidence whether CEO Narcissism has a positive and significant effect on earnings management. Based on the panel data regression test results shown through the t-test in Table 5 (model 1), CEO Narcissism has a negative and insignificant effect on earnings management. This study is consistent with the research results conducted by Kim (2018) and Savitri and Siswantoro (2021). Olsen et al. (2013) argue that CEOs with high levels of narcissism will face the cost of losing their reputation or self-esteem if fraud or manipulation of accounting policies is found where the narcissistic CEO does not like this effect. Furthermore, Olsen et al. (2013) concluded that narcissistic CEOs prefer to influence company performance through real management decisions rather than taking earnings management actions.
After enhancing the analysis, we also find that CEO publicity stimulate income increasing. This finding is supported with Kontesa et al. (2020) and Buchholz et al. (2020). They provide evidence that CEOs who are narcissistic are more likely to engage in earnings management than CEOs who are less narcissistic. It can happen because the characteristics of narcissism in the CEO who maintains pride and ego, also influence its strategic decisions. As a result, CEOs (agents) tend to place their reputation above shareholders' welfare (principal). Therefore, it can be concluded that the research results of Kontesa et al. (2020) and Buchholz et al. (2020) are in line with and contribute to confirming the correctness of the upper echelons theory and agency theory.
Supplementary analysis conducted in this paper obtain more evidence. CEO Publicity had a positive and significant effect on earnings management. It may be because the CEO's narcissistic need for public recognition causes CEOs to prefer more exposure (Rijsenbilt and Commandeur, 2013;Kim, 2018). The high number of publications that include the CEO's name in the news provides facts and information about the organization and its leaders, reinforcing a strong CEO image (Chen and Meindl, 1991). This situation stimulates the CEO to decide to carry out earnings management to not reduce his excellent reputation.
The extension of test also finds the debt-to-asset ratio's effect in strengthening the relationship between CEO narcissism and earnings management. The higher the financial risk, the narcissistic CEO will be more stimulated to display a positive financial condition with earnings management to maintain his reputation in the public's eyes.
However, after further investigation, the researchers found an interesting fact. Wink (1991) first conceptualized the nature of narcissism into two forms, namely grandiose narcissism and vulnerable narcissism.
Someone who displays the characteristics of grandiose narcissism often shows dominance in every action (Miller et al., 2011) and has arrogant self-confidence (Reich, 1970). On the other hand, vulnerable narcissism shows behavior that is always suspicious or afraid so that they are described as "hypersensitive" or "hypervigilant" (Ronningstam, 2009). Vulnerable narcissism exhibits behaviors similar to grandiose narcissism, such as the expectation of receiving special treatment from others, but with different motivations. Vulnerable narcissists believe that they deserve special treatment. They feel vulnerable, whereas grandiose narcissists expect attention from others because they believe they are better (Miller et al., 2011).
We believe our findings contribute to earnings management studies, especially the role of CEO publicity to stimulate more aggressive accounting by presenting more discretionary accrual. We also find that leverage positively affects earnings management when high CEO publicity. Conversely, the profitability negatively influences earnings management when the CEO publicity is more present than it should be. By adding narcissism factors, as an important factor in motivating incomeincreasing strategy, our paper will be helpful for investors and creditors.

Conclusions
This research study aims to obtain empirical evidence that narcissistic CEOs tend to practice earnings management than CEOs who are less narcissistic. This study of all listed firms other than the financial industry. Based on the regression test results, it was found that CEO Narcissism has no significant effect on earnings management where this result is in line with the research of Kim (2018) and Savitri and Siswantoro (2021). However, the results of this study contradict the research conducted by Kontesa et al. (2020) and Buchholz et al. (2020). It can be said that the results of this study contradict agency theory and upper echelons theory. Therefore, researchers conducted further testing through alternative models to explore in-depth about this.
In the supplementary analysis, different results were found with the initial study results. After three indicators of CEO Narcissism in the break-down and made more specific, CEO of publicity has a positive and significant impact on earnings management. Furthermore, the researcher also wants to see the effect of profitability in moderating the relationship between CEO Narcissism and earnings management in alternative model II. The results show the ROA strengthens the relationship between CEO Narcissism and earnings management. Overall, it can be concluded that in this study, a narcissistic CEO will be encouraged to perform earnings management if it is measured by the number of publications received by the CEO. The higher the exposure in the mass media about them, the higher the earnings management carried out by the narcissistic CEO.
This research study uses a panel data approach so that it is possible to assess changes in the level of CEO narcissism from time to time (time series) in all companies other than the financial sector (crosssection). That way, the research results are believed to provide more reliable conclusions because they are free from bias. We also control the regression using 2020 as Covid-19 situation variable. The relation between narcissism and earnings management is not affected by Covid-19 pandemic.
Future studies can be developed after considering the limitation of this paper. The main issue is limited data and access to CEO in exposing indicators of narcissism. This study also is not able to distinguish the behavior of narcissism using a two-dimensional approach, grandiose narcissism, and vulnerable narcissism. Separating the characteristics of narcissists who tend to be confident and narcissists who tend to be hypersensitive is important because it allows CEO narcissism to have a positive or negative relationship with earnings management. Examining using different techniques of earnings management is also needed to identify accounting techniques that frequently used by CEO. Future research should differentiate the impact of narcissism to accrual earnings management and real earnings management.