CEO NARCISSISM AND EARNINGS QUALITY
MASTER THESIS University of Groningen Faculty of Economics and Business
MSc Accountancy Word count: 8,708 January 23, 2017 JOSHUA MADURO Supervisor: dr. Y. (Yasemin) Karaibrahimoglu Second assessor: N. (Nazim) Hussain, PhD
ABSTRACT
CEOs make key decisions in firms on behalf of the firm’s shareholders. As part of this role, CEOs have much discretion and are able to influence various firm aspects. This influence can be driven by well-thought decisions in order to take advantage of the situation or simply by who they are as a person. Using a sample of 1,077 firm-year observations from 139 unique CEOs and firms in the S&P 500 Index, I examine the effect of the narcissistic trait of CEOs on the quality of earnings. I find that narcissistic CEOs tend to provide higher earnings quality through accruals, and more specifically with the use of positive accruals. This is in support of the efficient contracting theory, implying that narcissistic CEOs are concerned with maintaining their status, esteem and a positive stakeholders’ perception of their past and current decisions regarding accruals. All in all, narcissistic CEOs provide high quality (positive) accruals to prevent being labeled as opportunists.
Keywords:
TABLE OF CONTENTS
1. INTRODUCTION... 2
2. THEORETICAL BACKGROUND ... 5
2.1 Theory ... 5
2.2 Literature review and hypotheses development... 6
3. METHODOLOGY ... 11
3.1 Sample ... 11
3.2 Research model ... 12
3.3 Measurement of variables ... 12
4. RESULTS ... 19
4.1 Descriptive statistics, correlations and univariate test ... 19
4.2 Main test results ... 23
4.3 Robustness tests ... 25
5. DISCUSSION AND CONCLUSION ... 27
REFERENCES ... 31
APPENDIX A: HECKMAN TEST ... 35
APPENDIX B: PROPENSITY SCORE MATCHING TEST ... 36
1. INTRODUCTION
Chief Executive Officers (CEOs) are in charge of leading firms on behalf of the shareholders, generally in conjunction with the board of directors. For this important role, i.e. the daily key-decision maker of the firm, CEOs receive particularly a lot of attention (Bertrand, 2009). Furthermore, various CEOs have had such an impact inside as well as outside the firm that they have been put in the spotlight, e.g. Henry Ford back in his days, Steve Jobs of Apple, and Mark Zuckerberg of Facebook. Some CEOs receive quite an amount of media attention due to “some idiosyncratic behavior or management practice”, others because they pursue media attention to fulfill the needs of their narcissistic personality (Fetscherin, 2015, p. 6). In other words, the attention CEOs receive, is not only related to the fulfillment of the objectives of the role of CEO (or lack thereof) but is also influenced by such factors as the CEO’s personality.
In the CEO brand literature, CEO personality is an aspect of the CEO’s brand which affects various firm aspects (Fetscherin, 2015). Bendisch et al. (2013) build a model in their paper explaining how the CEO brand is jointly defined by identity and reputation. They emphasize that both aspects are crucial to the CEO brand as the definition of the identity influences the perception of the brand by the stakeholders (reputation), and vice versa. Moreover, Fetscherin (2015) also defined the CEO brand into two main components, i.e. CEO image and CEO reputation. These are further split into four main elements that are considered important to manage and which can be measured directly or indirectly. The four elements are CEO person, CEO personality, CEO prestige and CEO performance, where the first two relate to CEO image and the latter two to CEO reputation. As these four elements are argued to affect financial and non-financial aspects of companies (Fetscherin, 2015), more research on their actual effect on aspects of the firm could enhance our understanding of the CEO brand in more specific situations.
Previous studies conducted on the topic of CEO brand have put much emphasis on the reputation aspect of CEOs in relation to firm aspects (Francis et al., 2008; Jian and Lee, 2011; Kaufman, 2008; Malmendier and Tate, 2009). One aspect gaining attention is CEO personality, and more specifically the narcissistic trait of CEOs. Having CEOs with the narcissistic trait is argued to affect the firm’s reporting quality (Amernic and Craig, 2010) and also firm performance (Chatterjee and Hambrick, 2007). Previous research also identified that narcissistic CEOs tend to manage earnings through real and operational activities (Olsen et al., 2014). Furthermore, a positive relationship between narcissistic trait and fraud was identified
by Rijsenbilt and Commandeur (2013). These papers focus on the effects of the CEO’s personality trait of narcissism related to earnings management to intentionally mislead.
As stated above, previous research focused on how the narcissistic trait influences earnings management, but no research has explicitly been conducted on its effects on the quality of earnings. Earnings quality is found to be crucial in firm valuation and thus the shareholders’ worth (Bao and Bao, 2004; Beaver, 2002). Manipulation of earnings may affect the quality of information available (read: earnings quality) and thus the user’s decisions based on this information (Francis et al., 2004). Francis et al. (2004) argue that earnings quality is determined both by management discretion (and thus possibly driven by their personality) and innate factors of the firm. Therefore, research focusing on whether the narcissistic trait in fact influences management discretion is needed.
In earnings quality research, the quality of accruals is considered as a determinant of earnings quality (Dechow et al., 2010). Prior research of Olsen et al. (2014) attempted to identify what the actual effect of the narcissistic trait is on financial performance measures. They concluded that the personality trait does not affect financial performance measures through the CEO’s discretionary decisions regarding accruals. The authors base their conclusion on an accrual quality measure, which has been commented on by Dechow et al. (2010), stating that the model suffers from Type I error (i.e. false positive). This model treats accruals that represent fundamental performances falsely as abnormal accruals. In order to determine whether the narcissistic trait affects earnings quality, this paper incorporates a distinct accrual quality measure.
Moreover, this paper tries to answer the call of Dichev et al. (2013) to include human character aspects in earnings quality research. They found that the quality of earnings is determined by a firm’s innate factors as well as by managerial discretion, similar to Francis et al. (2004). As noted above, accrual quality is often used as a proxy for earnings quality (Dechow et al., 2010). Hence, the purpose of this paper is to enhance our current understanding of the narcissistic trait as a determinant of earnings quality through the quality of accruals. This paper seeks to provide empirical evidence of the consequences of the narcissistic trait on accrual quality using an alternative measure compared to Olsen et al. (2014). This leads to the following research question for this paper: what is the impact of the degree of CEO narcissism on the firm’s earnings quality through accruals?
Evidence provided by this study further refine our knowledge about the actual impact of the personal trait of narcissism on an aspect of accounting, namely accruals, which is related to the quality of earnings. As noted previously, initial evidence provided by Olsen et al. (2014) shows
that CEO narcissism is associated with earnings-per-share, but in their specific case not because of accrual-based manipulations but through real and operational activities. In this paper an alternative measurement is used for the quality of accruals, based on the Dechow and Dichev (2002) model, to explicitly link the personality trait with the quality of earnings. Additionally, this paper contributes to the literature of CEO brand by providing evidence that narcissistic CEOs are related to higher quality of accruals. This implies that the personality trait is a factor affecting the CEO’s decisions related to the use of accruals. It also implies that firms with narcissistic CEOs report earnings that could be considered as a more reliable basis for firm valuation purposes compared to firms with non-narcissistic CEOs.
The remainder of the paper is organized as follows. In section 2 previous research related to CEO narcissism and earnings quality, with the focus on accruals, is reviewed to formulate hypotheses. Section 3 models the relation and outlines the method used for the testing of the relationship. In section 4 the results are presented and these are discussed in section 5.
2. THEORETICAL BACKGROUND
In this section, firstly the agency theory and the economic perspectives are explained to illustrate the framework for the current study. Furthermore, previous research on CEO narcissism, earnings quality, and accrual quality are discussed, preceding the hypotheses formulation.
2.1 Theory
For the current research, an agency approach is taken in defining the relationship between CEO narcissism and accrual quality. More specifically, three economic perspectives, i.e. efficient contracting, rent extraction and matching, are introduced here to outline the basis for the formulation of the direction of the hypotheses at the end of this section.
Agency theory
In a contractual setting where the principal hires the agent to make decisions for him or her, they are both considered to be utility maximizers where it is argued that the interests of both parties do not align perfectly (Jensen and Meckling, 1976). Eisenhardt (1989) further noted that there are two main problems the agency theory tries to resolve. Firstly, their differing interests and goals makes it difficult or even costly for the principal to assess what the agent is doing (information asymmetry). Secondly, the author notes the divergence in risk preferences between the agent and principal. Both problems are arguably present in the case of a narcissistic CEO. Primarily because of this personal trait the CEO’s personal interest (enhancing his or her self-worth) does not align with that of the principal (maximizing shareholder value). Correspondingly, the narcissistic trait is found to lead to more risk-taking by CEOs, for example engaging in earnings management resulting in higher probability of restatements (Schrand and Zechman, 2012).
According to Eisenhardt (1989), agency theory is most appropriate where the problems of contracting are difficult. The personal trait of narcissism and the interests of such CEOs makes contracting difficult, as this characteristic of a person is not clearly visible and detectable ex-ante. Furthermore, when the interests of the agent and principals do not align, an agency problem arises from the separation of ownership and control within a firm leading to agency costs when the self-serving behavior and decisions of the agent negatively impact the principal’s wealth (Ang et al., 2000; Dalton et al., 1998; Jensen and Meckling, 1976; Rashid, 2015).
Economic perspectives
Francis et al. (2008) argued in their paper that the relationship between CEO reputation and earnings quality (in this paper accrual quality serves as a proxy for earnings quality) depends on the economic perspective taken. The same is expected for the relationship between the degree of CEO narcissism and earnings quality through the quality of accruals. The economic perspectives used in their paper are relevant, i.e. efficient contracting, rent extraction, and matching, which will be used as the framework to test for a relationship between CEO narcissism and accrual quality. The efficient contracting perspective is based on the model developed by Fama (1980), in which the past and current performance of the CEO determines how the stakeholders perceive his or her credibility for example and more importantly the CEO’s future wages (Francis et al., 2008).
Under the rent extraction perspective, it is suggested that the CEO uses his or her power to extract rent from the firm (Bebchuk et al., 2002). This rent extraction might affect the effectiveness of the firm performance. However, in order to enhance his or her career the CEO needs to take actions to meet or beat targets set by the shareholders, e.g. earnings management (Francis et al., 2008). This perspective is clearly the opposite of the efficient contracting perspective where the CEO is reluctant to take actions that might affect his or her future opportunities and wages.
The last perspective explained by Francis et al. (2008) is matching, where the board of directors hires CEOs who match certain criteria. These criteria contain CEO-specific and firm-specific factors. Thus, it is possible that a CEO is hired in a firm-specific situation, based on his or her degree of narcissism.
These three perspectives are adopted in this paper to form the basis for the hypotheses development following.
2.2
Literature review and hypotheses development
Numerous aspects of CEOs have been the subjects of previous research. Ali and Zhang (2015) argued and found that CEOs have different incentives in their early tenure as well as in their final year as CEO to manage earnings. The relationships between CEO compensation and various firm aspects have also been extensively researched (Chang et al., 2016; Core et al., 2003; Wade et al., 1997). Others have focused on how CEO reputation affects the firm (Boivie et al., 2016; Cianci and Kaplan, 2010). Another aspect of CEOs that has been researched is their media coverage (Blankespoor and De Haan, 2015; Lee, 2012). These have expanded the
literature on CEOs, but are less relevant for the current research. Prior research on CEO narcissism and accrual quality are discussed next in order to outline the relevant aspects for the formulation of hypotheses.
CEO narcissism
A person’s narcissism characteristic is identified in psychology research to be relatively stable (Campbell et al., 2004; Cramer, 1998) and this characteristic has been formerly defined in terms such as self-serving bias to self-worth enhancement through e.g. manipulation (Emmons, 1987). Nevertheless, narcissistic executives also posit positive characteristics, for instance “authority, self-sufficiency and superiority” (Olsen et al., 2014, p. 245). Firms benefit from the positive aspects of narcissistic CEOs as these executives introduce excellent vision and attract superior followers for the firm as a whole (MacCoby, 2000).
Olsen et al. (2014) state an important distinction between overconfidence and narcissism. Mainly, overconfident individuals put much more emphasis on their past experiences and previous accomplishments when determining their future outcomes. On the other hand, narcissists act in a superior manner irrespective of their past. Campbell et al. (2004) state that a simple way to picture a narcissistic individual is someone overly concerned with improving his or her chances to achieve status and esteem.
CEOs are in fact empowered with accounting to achieve their desired status and esteem. Amernic and Craig (2010) declare that accounting is a “hall of mirrors” for CEOs and thus has the potential to help the extreme forms of narcissism. The authors further argued that accounting is a public activity under which the CEOs will try to express their narcissism in narrative forms.
Patel and Cooper (2014) noted that the research on narcissism in accounting has mostly been based on the “dark side of executive personality,” as Resick et al. (2009, p. 1367) defined the negative aspect of narcissism. In their study, Patel and Cooper (2014) found that narcissistic CEOs affect the firm performance positively as well as negatively. By acting in a manner to elevate themselves these CEOs tumble harder during economic crises but also move on swiftly after one. The authors linked the latter to the high approach (vigorously pursue favorable circumstances) and low avoidance (less emphasis on possible undesirable outcomes) motivation of narcissistic CEOs embedded in their decision-making process.
CEO narcissism, earnings quality, and accrual quality
The quality of earnings is commonly defined as the informativeness of financial reporting concerning the performance of the firm (Dechow et al., 2010; Dichev et al., 2013). Depending on the context this financial reporting is relevant for the users regarding their decisions, and is affected by accrual quality (Dechow and Dichev, 2002; Francis et al., 2004; Francis et al, 2008; Gaio and Raposo, 2011). According to a survey by Dichev et al. (2013), CFOs indicate that earnings have a stewardship role, are used for internal matters in managing the firm, but mostly earnings are used for valuation purposes by investors. Gaio and Raposo (2011) provided empirical evidence for the latter, finding a positive and significant relation between firm valuation and earnings quality. The latter confirms that investors take earnings quality into account when valuing firms in stock markets.
While earnings are mostly used by investors to value firms, the assumption is that they are able to measure the firm’s earnings quality with the information available to them. Because it is found that the market is not fully efficient in processing information and anomalies cannot be known beforehand and incorporated into decisions (Basu, 1977; Malkiel, 2003), it is highly important to identify and highlight determinants of the quality of earnings. Dichev et al. (2013) therefore suggest to include such aspect as the human character of managers in studies regarding earnings quality, as their findings suggest that earnings quality is determined by innate factors of the firm as well as by managerial discretion.
Moreover, Francis et al. (2008) pointed out that much of the heterogeneity existing in earnings quality is not explained by firm-specific characteristics. Although they found that human factors are related to earnings quality, they concluded that in their study the association found between CEO reputation and earnings quality is explained by matching (innate earnings quality) and less by rent extraction by the CEO (discretionary earnings quality). They highlighted an important feature with this, explaining that firms presumably select CEOs by taking into consideration both CEO-specific and firm-specific characteristics. Thus, the latter suggests that the relation between the CEO characteristic and earnings quality is endogenous.
Furthermore, Dechow et al. (2010) noted that managers of the firm have several and often competing objectives and only one earnings number to report. Consequently, these CEOs are prone to influence the earnings with their accounting choices, e.g. through accruals. Other studies focused on earnings management and more specifically on the reasons why CEOs manage earnings. These ranges from incentives from the threat of violating debt covenants (Daniel et al., 2008) to incentives based on compensation (Ali and Zhang, 2015; Cheng and
Warfield, 2005). Ali and Zhang (2015) also argue that “high ability” CEOs tend to manage earnings to avoid reporting meager outcome and being labeled as “low ability” CEOs. By doing so they avoid being punished with a lower compensation or having negative consequences for their career. They stated further that CEOs managing earnings are at risk of being considered as acting opportunistically by stakeholders. Olsen et al. (2014) elaborated on this matter in relation to narcissistic CEOs, stating that these CEOs desire the positive outcome of their actions when it comes to earnings management. Nonetheless, they stated further that it can also lead to restatements and thus damaging their reputation. Schrand and Zechman (2012) found evidence confirming that narcissistic CEOs participate in earnings management which leads to higher probability of restatements. The latter conforms with Armenic and Craig (2010) who argued that accounting provides narcissistic CEOs with the ability to influence the financial information they provide to achieve their desired outcome of appraisal.
As earnings are considered a primary source of firm-specific information (Francis et al., 2004), its quality is crucial for the users. More specifically, accrual quality is found to affect the information risk carried by shareholders (Francis et al., 2005). This information risk weighs heavily on the portfolio of shareholders as it is an undiversifiable risk (Easley and O’hara, 2004). Bushman and Smith (2001) stated that financial accounting information not only has the possibility of affecting the decisions of investors, but it mostly helps to reduce information asymmetry. More generally, the actual effects of earnings management depend, as Nelson and Skinner (2013) claimed, on the intention of the management but also on the context in which the users make their decisions. In the agency setting, this implicates that accrual quality could be affected by earnings management. More specifically, in the sense that it increases information risk as the agent acts in his or her self-interest and not in attempting to reduce information asymmetry.
Summarized, the self-serving behavior of narcissistic CEOs influences the performance of a firm and empowered with accounting these CEOs can influence the information provided to shareholders (Amernic and Craig, 2010; Patel and Cooper, 2014). Also, narcissistic CEOs, having a high approach and a low avoidance motivation (Patel and Cooper, 2014), are expected to pursue their self-interest of obtaining status and esteem. The quality of accruals could be affected by their accounting and strategic choices, as CEOs are found to manage earnings in terms of the discretionary accruals but also through real and operational activities by spending less on R&D, advertising, and selling, general and administrative expenses for example (Ali and Zhang, 2015). Consequently, the information risk carried by the shareholders emerges from the information asymmetry, which on the long run affects the valuation of the shares they
hold in the firm (Ang et al., 2000; Bushman and Smith, 2001; Francis et al., 2005; Gaio and Raposo, 2011; Xie et al., 2003). The latter should be considered as the agency costs of the agency problem outlined for the current research.
This previous discussion could be linked to the rent extraction perspective introduced above, as the narcissistic CEO pursue his or her self-serving goals in ways that impact the shareholder’s wealth. In fact, to report a better performance the CEO is expected to manage earnings through accruals to meet or beat targets. The earnings management through accruals under this perspective results in lower quality of accruals.
Hypothesis 1. Under the rent extraction perspective, CEO narcissism is negatively associated with the quality of accruals.
On the contrary to the discussion above, based on the efficient contracting perspective a CEO could be concerned with not attaining his or her desired status or esteem if he or she misleads the shareholders. The CEO, aware of the possible consequences of his or her actions, will provide a higher quality of accruals in order not to negatively affect the quality of earnings. By doing so he or she avoids being labeled as acting opportunistically, which would affect his or her achievement of the desired status or esteem.
Hypothesis 2. Under the efficient contracting perspective, CEO narcissism is positively associated with the quality of accruals.
The selection of CEOs may be influenced by the quality of accruals prior to the CEO’s appointment. Patel and Cooper (2014) argue that firms benefit from narcissistic CEOs during difficult periods. Thus, the board of directors may consider and hire narcissistic CEOs when the firm suffers from poor quality of accruals, resulting either from past earnings management or due to innate factors. This implies that narcissistic CEOs are associated negatively with accrual quality, yet for different reasons than the rent extraction perspective.
Hypothesis 3. Under the matching perspective, CEO narcissism is negatively associated with the quality of accruals.
3. METHODOLOGY
This section describes how the sample was identified and defines the variables, including control variables, for testing the relationship between CEO narcissism and accrual quality.
3.1
Sample
The initial sample included all CEOs of S&P 500 firms available on ExecuComp database and consists of 4,755 CEO-years observations from 920 CEOs and 500 firms over the 2005-2014 period. Further, 867 CEO-years observations from financial firms (SIC 6000-6799) and 72 CEO-years observations from firms with ambiguity regarding CEO information were excluded.
Following other papers (Chatterjee and Hambrick, 2007; Olsen et al., 2014), the sample was further refined by imposing a filter on the identified CEOs for every firm-year. Rather than requiring minimum four or more total tenure years for each CEO in the sample, to prevent limiting the observations significantly, the requirement is applied in a different manner than Chatterjee and Hambrick (2007) and Olsen et al. (2014). The requirement imposed in this paper is that for every CEO and firm-year there are seven or more years of data between the years 2005-2014. Of these seven years, the first observation per CEO and firm-year is labeled as year one of the CEO’s tenure. The requirement functions to be able to measure CEO narcissism in years two and three (t+1 and t+2) of the CEO tenure in the sample and to have at least six observations for calculating the accrual quality measure. The accrual quality measure is computed in years two and onwards (t+n where n > 1) for the CEO tenure in the sample. Time t is the first year the CEO served on the function in the sample, or for some CEOs it is the actual first year of their tenure. The first year per CEO-firm is omitted as it possibly includes inconsistencies related to the change of CEO. Contrary to Chatterjee and Hambrick (2007) and Olsen et al. (2014), this paper does not incorporate a lagged design that ensures no circular or recursive relationship exist between the CEO narcissism measure and the earnings quality measure. The lagged design was not incorporated in this paper in order not to limit the sample significantly.
The sample was further adjusted to exclude firm-years with missing financial information in Compustat necessary for the accrual quality measure. Since the focal years for the accrual quality measure is year two and extending throughout the CEO’s tenure years in the dataset, the sample for the accrual quality measure includes the CEO’s tenure years two and up during the period of 2005-2014. The latter with a minimum of six yearly observations per CEO-firm.
After imposing these filters the accrual quality measure spans 2006-2014, and the final sample consists of 1,077 firm-year observations involving 139 CEOs and 139 unique firms. This is an average of 119 yearly observations during the period of 2006-2014 for the sample regarding the accrual quality measure, ranging from 86 observations in 2006 and 139 observations in 2010. Olsen et al. (2014), had 1,118 firm-year observations with 283 CEOs to measure the effect of the narcissistic trait. By not using the lagged design, while this paper has a sample with fewer CEOs, the amount of firm-year observations is close to that of Olsen et al. (2014).
3.2
Research model
Current research attempts to determine the effect of the degree of CEO narcissism on the firm’s accrual quality that serves as a proxy for earnings quality. The degree of CEO narcissism is measured in years two and three of the CEO tenure in the sample and considered as a stable personality. The accrual quality is measured for every CEO-firm observation on a yearly basis from year two onwards of the CEO’s tenure in the sample. Therefore, the model is estimated using panel data regression with fixed effects, with robust standard errors clustered by CEO following Olsen et al. (2014). I use the research model in equation 1.
𝐴𝑄𝑖,𝑡+𝑛 = 𝛽0+ 𝛽1𝐶𝐸𝑂𝑛𝑎𝑟𝑐𝑖𝑠𝑠𝑖𝑠𝑚𝑖 + 𝛽2𝐶𝐸𝑂𝑎𝑔𝑒𝑖,𝑡+𝑛−1+ 𝛽3𝑆𝑡𝑎𝑟𝑡𝑡𝑒𝑛𝑢𝑟𝑒𝑖,𝑡 + 𝛽4𝐶𝐸𝑂𝑡𝑒𝑛𝑢𝑟𝑒𝑖,𝑡+𝑛−1+ 𝛽5𝐶𝐸𝑂𝑐ℎ𝑎𝑖𝑟𝑖,𝑡+𝑛−1+ 𝛽6𝐿𝑜𝑔𝑇𝐴𝑖,𝑡+𝑛 + 𝛽7𝐶𝐹𝑂𝑉𝑎𝑟𝑖,𝑡+𝑛+ 𝛽8𝑆𝑎𝑙𝑒𝑠𝑉𝑎𝑟𝑖,𝑡+𝑛+ 𝛽9𝐿𝑒𝑛𝑔𝑡ℎ𝑂𝐶𝑖,𝑡+𝑛 + 𝛽10𝑁𝑒𝑔𝐸𝐴𝑅𝑁𝑖,𝑡+𝑛+ 𝛽11𝑅𝑂𝐴𝑖,𝑡+𝑛−1+ 𝛽12𝐴𝑄𝑖,𝑡+𝑛−1 + 𝛽13𝑅𝑂𝐴𝑖,𝑡−1+ 𝐶𝐸𝑂𝑎𝑝𝑝𝑜𝑖𝑛𝑡𝑖 + 𝜙𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝐷𝑢𝑚𝑚𝑖𝑒𝑠 + 𝜙𝑌𝑒𝑎𝑟𝐷𝑢𝑚𝑚𝑖𝑒𝑠 + 𝜀
3.3
Measurement of variables
Accrual qualityPrior research used accrual quality as a single measure to proxy earnings quality or several other proxies (for a summary see Dechow et al., 2010). Francis et al. (2004) defined two categories of earning attributes in their paper, i.e. accounting-based and market-based attributes, with seven earning attributes in total. The seven attributes are accrual quality, persistence, predictability, smoothness, value relevance, timeliness, and conservatism. The purpose of earnings based on the accounting-based attributes is defined as the appropriate allocation of cash flows through accruals to their concerning period. Whereas the market-based
attributes consider earnings as the reflection of economic income, with stock returns as a proxy. The authors further found in their study of the effects of the earning attributes on the cost of equity that the accounting-based attributes, and especially accrual quality, have an extensive effect.
Therefore, to use an alternative measure of accrual quality to test for the effects of the degree CEO narcissism, compared to Olsen et al. (2014), this paper uses the model of Dechow and Dichev (2002) for measuring the accrual quality. The same model was used in the paper of Francis et al. (2004). This paper focuses solely on the accrual quality instead of incorporating the other accounting-based attributes as to isolate the accruals effects of the narcissistic trait.
The model for accrual quality measure from Dechow and Dichev (2002) relates the working capital accruals with the prior, current, and future cash flow from operations, where a poor match indicates low accrual quality:
𝑊𝐶𝐴𝑖,𝑡 = 𝛽0,𝑖+ 𝛽1,𝑖𝐶𝐹𝑂𝑖,𝑡−1+ 𝛽2,𝑖𝐶𝐹𝑂𝑖,𝑡+ 𝛽3,𝑖𝐶𝐹𝑂𝑖,𝑡+1+ 𝜈𝑖,𝑡 (2)
where:
WCAi,t is firm i’s working capital accruals between year t-1 and t, and CFOi,t is the firm’s cash flow from operations, where all variables are scaled by average assets for year t.
The working capital accruals is calculated as follows:
𝑊𝐶𝐴𝑖,𝑡 = Δ𝐴𝑅𝑖,𝑡+ Δ𝐼𝑛𝑣𝑒𝑛𝑡𝑜𝑟𝑦𝑖,𝑡− Δ𝐴𝑃𝑖,𝑡− ∆𝑇𝑃𝑖,𝑡+ Δ𝑂𝑡ℎ𝑒𝑟 𝐴𝑠𝑠𝑒𝑡𝑠 (𝑛𝑒𝑡)𝑖,𝑡 (3)
where WCAi,t is the change between year t-1 and year t of firm i’s accounts receivables, inventory, accounts payable, taxes payable, and other (net) assets, respectively. The data for equation (3) was gathered from Compustat as items of the cash flow statement.
Cash flow from operations (CFOi,t) was obtained from Compustat and is the net cash flow from operational activities for year t-1, t, and t+1.
Equation (2) is calculated for each firm i and for the period t+n in the sample. The accrual quality measure is then calculated as the absolute values of the residuals (|𝜈𝑖,𝑡|) per CEO-firm on a yearly basis:
Consistent with Dechow and Dichev (2002), the higher the outcome of the accrual quality measure, the lower the accrual quality is taken to be, as this is an indication that less of the variation in current accruals is explained by operating cash flows. Earnings quality is considered to be negatively affected by the measure of accrual quality, as the possible manipulation resulting in the poor accrual quality leads to a mismatch between the accruals and the cash flows.
CEO narcissism
The measurement of the narcissistic personality trait in psychology is usually performed by the Narcissism Personality Inventory (NPI), which is an instrument using a forty-item forced-choice method (Raskin and Terry, 1988). In research regarding CEO narcissism this is not feasible as the questions of the instrument are sensitive in nature where the responses from CEOs would be limited (Cycyota and Harrison, 2006) and possibly influenced by social desirability bias (Chatterjee and Hambrick, 2007). This paper follows the unobtrusive measure of CEO narcissism used by Olsen et al. (2014). Their measure is a modified composite measure based on the unobtrusive measure developed by Chatterjee and Hambrick (2007). The measure of Olsen et al. (2014) consists of the composite index of three items attempting to capture the CEO’s narcissistic personality trait, namely the prominence of the CEO’s photograph in the firm’s annual report and the relative cash and noncash pay of the CEO to that of the second-highest paid executive. The measure of Olsen et al. (2014) differs from that of Chatterjee and Hambrick (2007) in that it does not include either the prominence of the CEO in the firm’s press release or the CEO’s use of first-person pronouns in interviews (Olsen and Stekelberg, 2016). Chatterjee and Hambrick (2011) documented the measure of first-person pronouns as weakening the composite measure after the passage of the Sarbanes-Oxley Act in 2002.
Annual reports, a yearly opportunity for the firm to disclose the firm’s performance and to depict the firm’s future potentials, also provide the opportunity for CEOs to promote their image. CEO photographs are common in annual reports, but the size of the photographs differ across annual reports (Chatterjee and Hambrick, 2007). Since narcissistic CEOs seek admiration and recognition from others and have inflated self-concepts, it could be expected that they would also influence the size of their photograph accordingly to their desire to stand out (Chatterjee and Hambrick, 2007; Olsen et al., 2014; Olsen and Stekelberg, 2016).
This paper uses the exact measure utilized and validated by Olsen et al, (2014) and Olsen and Stekelberg (2016). The CEOs photograph in the firm’s annual report is rated from 1 to 5 in the following order:
1. The annual report does not contain a photograph of the CEO. 2. The CEO was photographed with other executives.
3. The CEO was photographed alone and the photograph occupies less than half of a page. 4. The CEO was photographed alone and the photograph occupies at least half of a page,
and the photograph shares the page with text.
5. The CEO was photographed alone and the photograph occupies the entire page. The ratings for the second and third year of the CEO’s tenure years available in the dataset are then summed up and averaged over the two years. The average rating is the input in the composite measure of the CEO’s narcissism based on the prominence of his or her photograph. Besides the prominence of their photograph in the firm’s annual report, narcissistic CEOs are found to obtain larger cash compensation and their pay varying greatly from the other executives in the firm (O’Reilly et al., 2014). This could be driven by the CEO influencing his or her pay and that of the other executives to receive greater compensation reflecting his or her greater self-perceived value (Chatterjee and Hambrick, 2007). Following Chatterjee and Hambrick (2007) and Olsen et al. (2014), the relative cash and noncash pay of the CEO to that of the second-highest paid executive are calculated as follows:
- Relative cash pay: the salary and bonus (i.e. cash compensation) of the CEO divided by that of the second-highest paid executive.
- Relative noncash pay: the total compensation minus the cash compensation of the CEO divided by that of the second-highest paid executive.
Both the relative cash and noncash pay are also averaged over the second and third year of the CEO’s tenure.
Table 1 presents the descriptive statistics of the CEO narcissism measure and the frequency of the average score for the photograph size. A factor analysis (untabulated) was performed to determine the factor weightings used in the composite measure of CEO narcissism, following Olsen et al. (2014). The factor loadings are .216 for CEO photograph size, .714 for relative cash pay, and .705 for relative noncash pay.
TABLE 1
Measure CEO Narcissism Panel A: Descriptive statistics
n Mean Std. Dev. Min. Max.
Size CEO photographa 139 2.356 .935 1 5
Relative cash payb 139 1.633 .807 .000 8.219
Relative noncash payc 139 2.551 1.471 .001 9.781
Narcissism scored 139 .006 .970 -2.128 6.681
a The size of the photograph is rated based on a scale of 1 to 5: (1) No photograph of the CEO; (2) Photograph
was of the CEO with other executives; (3) Photograph of the CEO alone and less than half the page; (4) Photograph of the CEO alone and more than half the page sharing it with some text; (5) Photograph of CEO alone taking the whole page.
b Relative cash pay is the CEO’s cash pay relative to that of the second-highest paid executive. c Relative noncash pay is the CEO’s noncash pay relative to that of the second-highest paid executive.
d Is the calculated narcissism score using the factor weightings of the size of the CEO photograph, relative cash
and noncash pay resulted from a factor analysis.
Panel B: Overview CEO photograph size
Average of CEO photograph
size Frequency
Percentage of total CEOs in sample 1 32 23% 1.5 2 1% 2 32 23% 2.5 5 4% 3 53 38% 3.5 9 6% 4 4 3% 4.5 1 1% 5 1 1% Total CEOs 139
Control variables
This paper includes control variables at two levels to control for several other factors that might affect the accrual quality: at the CEO- and the firm-level.
Regarding CEO-level controls, following Chatterjee and Hambrick (2007) and Olsen et al. (2014) this paper controls for overall trends by controlling for the year the CEO started his or her tenure (StartTenurei,t). Similarly, controls for CEO characteristics such as age (CEOage i,t+n-1) and tenure (CEOtenurei,t+n-1) are also included, as these possibly affect the tendency for the CEO to take actions affecting the accrual quality (e.g. Ali and Zhang, 2015). Furthermore, to control for structural power (Finkelstein, 1992), this paper also includes a control variable for whether the CEO is also chairman of the board (CEOchairi,t+n-1). Lastly, this paper controls for the inclusion of CEOs of whom the first year of his or her tenure is either included or excluded resulting from the sample filter. CEOs whose first year(s) are not included in the final sample possibly served as CEO for a long period prior to the starting of the sample period. The latter could affect the results, as Ali and Zhang (2015) found that CEOs overstate earnings more greatly in the early years of their tenure. An indicator variable takes the value of 1 if the year of appointment is included in the sample, or zero if the first tenure year of the CEO in the sample is not the appointment year.
Regarding firm-level controls, following Dechow and Dichev (2002), I use controls for firm size measured as the log of the average total assets of firm i (LogTAi,t+n), cash flow variability as the standard deviation of firm i’s cash flows for the years in the sample scaled by total assets (CFOVari,t+n), for sales variability as the standard deviation of firm i’s sales scaled by total assets (SalesVari,t+n), length of operating cycle as the log of the sum of the firm i’s days accounts receivables and days inventory (LengthOCi,t+n), and for firm i’s proportion of years with negative earnings during the sample period (NegEARNi,t+n). Dechow and Dichev (2002) posit and found that the firm size is negatively related to the accrual quality measure, and the length of operating cycle, loss incidence, sales variability and cash flow from operations variability are positively related.
This paper further controls for the firm’s prior year performance (ROAi,t+n-1) similar to Olsen et al. (2014), and also includes the firm’s prior year accrual quality (AQi,t+n-1) as control variable. Furthermore, the firm’s situation before the CEO took the function is controlled for the possibility that the firm has innate performance tendencies, by including control for the performance prior to the CEO’s tenure start (ROAi,t-1). Finally, industry dummies using
two-digit SIC codes and year dummies are included to control for the possibility of omitted factors in the industry and years affecting the accrual quality.
All variables are defined in Table 2 below.
TABLE 2
Variable descriptions
Variable Description
AQi,t+n Accrual quality measure as the absolute residuals of the regression of the
working capital accruals on cash flow from operations from prior, current, and following year, based on the model of Dechow and Dichev (2002).
AQnegpos Accrual measure as the positive or negative accruals per firm-year observation.
CEOnarcissismi A composite measure of the score per CEO calculated as the factor weightings
of the propensity of the CEO photograph, relative cash pay, and relative noncash pay.
CEOagei,t+n-1 CEO’s prior year age.
Starttenurei,t The year the CEO was appointed.
CEOtenurei,t+n-1 The CEO’s (lagged) tenure length.
CEOchairi,t+n-1 An indicator variable of 1 if the CEO was the chairman in prior year, 0
otherwise.
LogTAi,t+n Logarithm of total assets.
CFOVari,t+n Variability of the cash flow from operations.
SalesVari,t+n Variability of sales revenue.
LengthOCi,t+n Length of the operating cycle.
NegEARNi,t+n The firm’s proportion of years with negative earnings.
ROAi,t+n-1 Return-on-assets of prior year.
AQi,t+n-1 Accrual quality of prior year.
ROAi,t-1 Return-on-assets of the year prior to the CEO’s appointment.
CEOappointi An indicator taking the value of 1 if the CEO’s appointment year is included in
4. RESULTS
In this section an overview of the variables, the descriptive statistics, the correlations among variables, and the results of the panel regression as well as robustness tests are presented.
4.1
Descriptive statistics, correlations and univariate test
An overview of the variables is presented in Table 2. The descriptive statistics of the variables are presented alongside the univariate test between narcissistic and non-narcissistic CEOs in Table 3. In the full sample, the average CEO started their tenure in 2001 and had an average tenure length of 4.465 years. The mean difference between absolute accruals for narcissistic CEOs and non-narcissistic CEOs is significant t(635)=3.593; p=.000, indicating that firms with narcissistic CEOs have a higher quality of accruals. For an average of 65.5% of the observation years in the full sample, the CEO was also the chairman of the board of directors. The mean difference between narcissistic and non-narcissistic CEOs for the chairmanship is highly significant t(1071)=-4.189; p=.000, with the narcissistic CEOs in the sample having the function of chairmanship in more observation years than the non-narcissistic CEOs. Sales variability has a sample mean of .243, with the mean difference between narcissistic and non-narcissistic CEOs highly significant t(757)=3.161; p=.002, where firms with narcissistic CEOs have a lower variability in sales. The mean difference between narcissistic and non-narcissistic CEOs for the firm size, length of the operating cycle and proportion of years with negative earnings are significant at the .05 level, whereas CEO age is slightly significant at the .10 level.
Table 4 presents the correlations among the variables. The degree of the CEO narcissism is significantly negative correlated with the accrual quality measurement (r=-.127; p<.01).
TABLE 3
Descriptive statistics with univariate test
Variable Full sample (n = 1,077) Narcissistic CEOs (n = 531) Non-narcissistic CEOs (n = 546) Mean difference Mean Std.
dev. Min. Max. Mean
Std.
dev. Min. Max Mean
Std.
dev. Min. Max Coefficient p-value
AQabs .009 .019 .000 .378 .007 .007 .000 .069 .011 .025 .000 .378 .004 *** .000 AQnegpos .000 .021 -.378 .310 -.000 .010 -.069 .041 .000 .028 -.378 .310 .000 1.000 CEOagei,t+n-1 55.589 6.523 37 82 55.942 5.559 42 75 55.245 7.329 37 82 -.696 * .079 Starttenurei,t 2001 6 1978 2008 2001 6 1978 2008 2001 6 1979 2008 .393 .291 CEOtenurei,t+n-1 4.465 2.367 1 9 4.394 2.340 1 9 4.535 2.393 1 9 .141 .328 CEOchairi,t+n-1 .655 .476 0 1 .716 .452 0 1 .595 .491 0 1 -.120 *** .000 LogTAi,t+n 3.959 .489 2.489 5.154 3.929 .474 2.863 5.056 3.989 .501 2.489 5.154 .060 ** .043 CFOVari,t+n .046 .045 .005 .491 .046 .041 .007 .294 .047 .047 .005 .491 .001 .828 SalesVari,t+n .243 .367 .008 6.415 .207 .209 .008 1.727 .277 .470 .009 6.415 .070 *** .002 LengthOCi,t+n 1.988 .315 0 2.800 2.008 .231 1.120 2.731 1.968 .378 0 2.800 -.040 ** .037 NegEARNi,t+n .030 .077 0 .714 .025 .057 0 .286 .036 .093 0 .714 .011 ** .007 ROAi,t+n-1 .087 .076 -.853 .503 .084 .072 -.853 .503 .090 .079 -.474 .449 .006 .229 AQi,t+n-1 .010 .020 .000 .435 .008 .008 .000 .084 .012 .025 .000 .435 .004 *** .001 ROAi,t-1 .081 .066 -.223 .309 .080 .057 -.050 .309 .081 .074 -.223 .255 .001 .821 CEOappointi .317 .465 0 1 .294 .456 0 1 .339 .474 0 1 .045 .112
*, **, *** Significant at the .10, .05, and .01 level, respectively. p-values are two-tailed.
TABLE 4
Variable correlations Panel A: Correlations variables AQi,t+n to LogTAi,t+n
Variable AQabs AQnegpos CEOnarcissismi CEOagei,t+n-1 Starttenurei,t CEOtenurei,t+n-1 CEOchairi,t+n-1 LogTAi,t+n
AQabs 1.000 AQnegpos -.061 1.000 CEOnarcissismi -.127*** -.000 1.000 CEOagei,t+n-1 -.092*** .015 .118*** 1.000 Starttenurei,t -.050 .000 -.151*** -.358*** 1.000 CEOtenurei,t+n-1 .028 .038 -.007 .327*** -.079*** 1.000 CEOchairi,t+n-1 -.005 .002 .194*** .309*** -.190*** .183*** 1.000 LogTAi,t+n -.153*** .012 -.035 .138*** .103*** .147*** .144*** 1.000 CFOVari,t+n .160*** -.030 -.113*** -.191*** -.187*** -.250*** -.137*** -.445*** SalesVari,t+n .172*** -.029 -.163*** -.210*** -.146*** -.175*** -.045 -.310*** LengthOCi,t+n -.330*** -.031 .007 .031 .197*** .023 -.091*** -.010 NegEARNi,t+n -.004 .000 -.158*** .026 -.030 -.019 -.108*** .045 ROAi,t+n-1 .087*** .054 -.077** -.073** -.032 -.030 -.077** -.261*** AQi,t+n-1 .447*** .121*** -.124*** -.100*** -.055* .004 .005 -.179*** ROAi,t-1 .121*** -.000 -.083*** -.130*** -.006 .006 -.059* -.282*** CEOappointi -.058* -.000 .030 -.191*** .511*** -.072** -.119*** .120***
TABLE 4 (continued)
Panel B: Correlations variables CFOVari,t+n to CEOappointi
Variable CFOVari,t+n SalesVari,t+n LengthOCi,t+n NegEARNi,t+n ROAi,t+n-1 AQi,t+n-1 ROAi,t-1 CEOappointi
CFOVari,t+n 1.000 SalesVari,t+n .687*** 1.000 LengthOCi,t+n -.123*** -.282*** 1.000 NegEARNi,t+n .023 .034 .038 1.000 ROAi,t+n-1 .227*** .112*** .092*** -.309*** 1.000 AQi,t+n-1 .217*** .213*** -.297*** -.012 .143*** 1.000 ROAi,t-1 .188*** .175*** -.020 -.143*** .462*** .186*** 1.000 CEOappointi -.162*** -.166*** .143*** .002 -.006 -.046 .066** 1.000
4.2
Main test results
The results of the panel regression testing for the relationship between the degree of CEO narcissism and the level of accrual quality are presented in Table 5. The results are presented for the absolute accruals as well as for the negative and positive accruals separately.
After controlling for factors related to the CEO and the firm, the degree of CEO narcissism is significantly negative related to the level of absolute accruals (β=-.001; p=.009). This result indicates that the higher (lower) the degree of CEO narcissism, the lower (higher) the outcome of the accrual quality measure, and recall that a lower (higher) outcome of the accrual quality measure implies higher (lower) quality of accruals. The latter provides evidence that a statistically significant portion of the variation in current accruals is explained by operating cash flows (Dechow and Dichev, 2002), implying that the accruals are of high quality. This is in support of Hypothesis 2, providing evidence in support of the efficient contracting hypothesis.
Of the control variables, the variability of the cash flow from operations (β=.063; p=.002) has a positive coefficient and is significant, in accordance with Dechow and Dichev (2002). Furthermore, the year the CEO was appointed (β=-.000; p=.042), the return-on-assets in the year prior to the CEO was appointed (β=.022; p=.012), and the length of the CEO’s tenure (β=-.001; p=.091) are also significantly related to the level of accrual quality.
The main test was repeated twice using the negative (540 observations) and positive (537 observations) accruals separately as the dependent variable, with results showing that the degree of narcissism drives only the positive accruals significantly (β=-.001; p=.022). The control variables remain in the same direction for the subsample of positive accruals, but at a lower significance. The CEO’s tenure is no longer significant at the .10 level compared to the full sample, and the prior year’s accruals become slightly significant.
TABLE 5
CEO Narcissism and Accrual Quality: Panel Regression
Variable Absolute accruals Negative accruals Positive accruals Coefficient p-value Coefficient p-value Coefficient p-value
CEOnarcissismi -.001 *** .009 .001 .159 -.001 ** .022 CEOagei,t+n-1 -.000 .212 .000 .439 -.000 .259 Starttenurei,t -.000 ** .042 .000 * .097 -.000 * .077 CEOtenurei,t+n-1 -.001 * .091 .001 .126 -.001 .187 CEOchairi,t+n-1 -.001 .599 .001 .743 -.001 .476 LogTAi,t+n -.000 .865 .001 .663 -.000 .750 CFOVari,t+n .063 *** .002 -.064 * .066 .091 ** .048 SalesVari,t+n -.008 .128 .009 .113 -.015 .245 LengthOCi,t+n .002 .450 -.001 .775 .002 .600 NegEARNi,t+n .010 .194 -.002 .803 .012 .143 ROAi,t+n-1 .005 .552 .002 .832 .001 .846 AQi,t+n-1 -.024 .619 -.470 ** .035 -.228 * .093 ROAi,t-1 .022 ** .012 -.008 .600 .035 *** .008 CEOappointi -.001 .361 .001 .207 -.002 .319
Industry Dummies Included Included Included
Year Dummies Included Included Included
CEO-firm observations 1,077 540 537 CEOs 139 139 139 Overall R2 .492 .524 .599
*, **, *** Significant at the .10, .05, and .01 level, respectively. p-values are two-tailed.
Variables are defined in Table 2.
4.3
Robustness tests
Endogeneity could possibly be an issue in the research model introduced in section 3.2, as narcissistic CEOs might be attracted to and/or hired in certain situations where they could potentially influence the quality of accruals in one’s favor. This concern is addressed with a robustness test using propensity score matching [PSM] (Rosenbaum and Rubin, 1983) which matches CEO-firm observations for narcissistic CEOs and non-narcissistic CEOs. To separate narcissistic and non-narcissistic CEO, the paper uses a narcissism indicator that takes the value of 1 if the CEO’s narcissism score is larger than the average and zero otherwise, similar to Olsen et al. (2014). Another concern of the model is that sample selection bias might affect the results, as the dataset is unbalanced due to varying observations per CEO-firm. Following Chatterjee and Hambrick (2007), this paper addresses this possible bias with a robustness test, by using the procedures of the Heckman selection model (Heckman, 1976). The latter is used to calculate the inverse Mills ratio which then serves as a control variable in the main panel regression. The tables with the results of the robustness tests are included in the Appendices, for the results of the Heckman test refer to Appendix A, and for the results of the PSM refer to Appendix B.
Sample selection bias test
A concern of the model used for current paper is that the results might be affected by sample selection bias. In the sample used for the main test, there are CEOs with different tenure length which may affect the results. It is namely possible that the subsample of narcissistic CEOs could be affecting the results differently than that of non-narcissistic CEOs. To address this concern, the procedure of the Heckman method (Heckman, 1976) was followed in order to calculate the inverse Mills ratio which was included in the main regression serving as a control for the sample selection bias.
The results (see Appendix A) indicate that after including the inverse Mills ratio in the main panel regression, the variable of interest (degree of narcissism) remains significant, although now at the .05 level (β=-.001; p=.013). Also, the inverse Mills ratio as control variable does not appear to be statistically significant. These indicate there is no evidence of a sample selection bias driving the main results.
Propensity Score matching
Regarding the endogeneity concern, firstly the narcissistic CEOs are separated from the non-narcissistic CEOs using an indicator variable. This indicator variable takes the value of 1 if the CEO has a narcissism score higher than the average, zero otherwise. This separation resulted in 71 of the 139 CEOs in the full sample being categorized as narcissistic CEOs.
The following step is to match narcissistic CEOs with non-narcissistic CEOs based on CEO- and firm-specific characteristics. The CEOs were matched using CEO- and firm-specific factors that might either enhance or restrain the degree of the narcissistic trait, i.e. the CEO’s tenure start, tenure length, age, chairmanship, the proportion years with negative earnings in the sample, return-on-assets and sales variability of the firm. The variables used for matching are as described in Table 2.
Using PSM (Rosenbaum and Rubin, 1983) this paper tests whether narcissistic CEOs are more prone to be related to the level of accrual quality compared to non-narcissistic CEOs. After matching the narcissistic and non-narcissistic CEOs, the sample size is reduced to 686 firm-years observations, with 343 matched pairs. The propensity scores for the matched pairs were then included in the panel regression as a weight. The results of the PSM test show (see Appendix B) that the main result for the variable of interest, i.e. degree of narcissism (β=-.001; p=.044), is robust and that the model does not suffer from endogeneity. It is important to mention that the effect of the narcissism score on the accruals quality has a lower significance in the test, yet it remains significant at the .05 level. These results therefore show no support for Hypothesis 3, indicating that matching does not occur for narcissistic CEOs.
OLS results
Almost the same results were obtained using an OLS regression (see Appendix C), with the variable of interest (degree of narcissism) becoming less significant, yet still significant at the .05 level (β=-.001; p=.033). Compared to the main panel regression, the sales variability becomes highly significant (β=-.008; p=.000). Thus, the results are robust disregard the method used.
5. DISCUSSION AND CONCLUSION
This section includes a discussion of the results and the implications for the literature and practice. Also, the limitations of the paper alongside aspects for future research on the topic are outlined.
Based on the ‘personality aspect’, as defined in the CEO brand literature, this paper proposed that the CEO’s narcissistic trait affect the quality of earnings through accruals. This paper proposes that the relationship could be explained either by the rent extraction perspective, stemming from the agency theory, or the efficient contracting perspective. Furthermore, Patel and Cooper (2014) noted that firms benefit from narcissistic CEOs during difficult periods, because of their high approach and low avoidance motivation in their decision-making process. Therefore, a third possible explanation was included for the effect of the narcissistic trait on the quality of earnings through accruals, namely a matching perspective. Based on the matching perspective, narcissistic CEOs are expected to be linked to lower quality of accruals for different reasons compared to the rent extraction perspective. It is thus expected that narcissistic CEOs were appointed to firms with low accrual quality resulting either from past earnings management or due to innate firm factors.
Using panel regression, this paper found that the degree of the narcissistic trait is significantly related to the quality of accruals. More specifically, it is found that a higher (lower) degree of narcissism is related to a higher (lower) accrual quality. This result supports the efficient contracting perspective, indicating that narcissistic CEOs are in fact concerned with the consequences of their actions related to the use of accruals. They are aware that the quality of accruals affect the stakeholders’ perception of their ability and credibility and that misuse of accruals might damage their status and esteem (Fama, 1980; Francis et al., 2008).
Moreover, as the year the CEO started his/her tenure is found to be significant and 2001 being the average CEO starting year in the sample, might suggest that these CEOs are aware of the scrutiny after the passage of the Sarbanes-Oxley Act (SOX). As narcissistic CEOs desire appraisal and the achievement of status and esteem, they make sure to provide higher quality accruals. By doing so, they avoid being labeled as opportunistic by shareholders and other stakeholders in general. The latter is plausible as various researchers argued that there has been a shift from earnings management from discretionary accruals to real and operational activities after the passage of SOX (Chang et al., 2009; Cohen et al., 2008). Therefore, it is important to note that the result of this study showing that the quality of accruals for firms with narcissistic
CEOs is higher than that of firms with non-narcissistic CEOs, does not necessarily mean that narcissistic CEOs do not participate in earnings management.
Additionally, the main panel regression was conducted separately for the quality of negative and positive accruals. The results indicate that the degree of the narcissistic trait, in the setting of this paper, affects only the quality of positive accruals significantly. This indicates that the use of positive accruals by narcissistic CEOs results in a better indication of the firm’s future earnings compared to firms with non-narcissistic CEOs (Dechow and Ge, 2006). Dechow and Ge (2006) posit that positive accruals are an indication of investment, generation of sales and expansion of business. They further state that these firms have negative transitory cash where accrual adjustments are more likely to improve the earnings persistence. Thus, it can be concluded that narcissistic CEOs, in comparison with non-narcissistic CEOs, provide higher accrual quality using positive accruals. This leads to a better indication of the firm’s future earnings with an improved earnings persistence for firms with narcissistic CEOs.
On the other hand, no evidence was provided for a relation between the degree of CEO narcissism and negative accruals. As stated above, the passage of SOX has its effects on the quality of accruals. The scrutiny after the passage of SOX could be an explanation why narcissistic CEOs use positive accruals to provide a better indication of the firm’s future earnings. Similarly, the passage of SOX could explain why narcissistic CEOs avoid transitory earnings with the use of negative accruals, which leads to low earnings persistence according to Dechow and Ge (2006).
The findings provide several implications for the literature and practice. For the literature related to earnings quality, the findings suggest that such factor as the narcissistic trait indeed affect the quality of accruals which in turn affects the quality of earnings. Thus, besides firm-specific factors and other CEO-firm-specific factors affecting earnings quality, evidence is provided showing that the degree of narcissism, a personality trait, also affects the quality of earnings. An important difference of this paper with Olsen et al. (2014) is that their paper is focused solely on earnings management, whereas this paper puts emphasis on the quality of earnings. Furthermore, in this paper a distinction is provided for positive and negative accruals. This paper’s findings, that narcissistic CEOs avoid negative accruals and use positive accruals to provide a better quality of earnings, could be an explanation why Olsen et al. (2014) did not find narcissistic CEOs to manage earnings through accruals. All in all, this paper further refines our understanding of the effect of the narcissistic trait on the quality of earnings, by providing evidence that the degree of CEO narcissism is related to a high quality of positive accruals. In
quality positive accruals. Investors could use these earnings as a basis to value the firm more accurately as it is argued by Dechow and Ge (2006) that positive accruals improve the earnings persistence.
For the CEO brand literature, this paper further provides evidence that also the narcissistic trait, part of the CEO personality aspect, influences the decisions and thus actions of the CEOs in order to build their brand. In contrast with Francis et al. (2008) and their finding that CEO reputation affects earnings quality based on the matching perspective, this paper provides evidence that the effect of the narcissistic trait on the quality of earnings is explained by the efficient contracting perspective. The contrasting findings suggest that competing economic perspectives could explain the effect of different personality aspects on the quality of earnings.
Although this paper attempted to provide an extensive analysis by decomposing the sample into narcissistic and non-narcissistic CEOs, analyzing the effects on both negative and positive accruals separately, like any other there are limitations to the research conducted. Regarding the generalizability of the results, the limited sample must be taken into consideration. Namely, for the selection of the sample, the CEO-firm combination required at least six yearly observations to be included in the sample, indicating that only surviving CEO-firm relationships are represented in the sample. Also, the firms in the sample were selected from the firms listed in the S&P 500 Index, which includes large companies and thus does not represent firms with different sizes, especially small- and medium-sized firms.
Another potential limitation is that, diverging from Chatterjee and Hambrick (2007) and Olsen et al. (2014), this paper does not incorporate a lag in the design to measure the degree of narcissism before the focal years for the accrual quality, due to sample limitation if doing so. Even though the degree of narcissism is maintained constant to test for its effect on the quality of accruals, this was calculated in the CEO’s tenure year two and three, whereas the focal years of the accrual quality start from year two through the length of the CEO’s tenure. This might introduce a reverse and circular relationship between the variables for year two and three of the CEO’s tenure (Chatterjee and Hambrick, 2007; Olsen et al., 2014).
The limitations described create opportunity for future research. Future research could not only make use of a larger sample and extended period to incorporate a lagged design, but also could include firms of different sizes to determine whether the effects of the narcissistic trait differs for firms with different sizes. Moreover, future research could use a more comprehensive earnings quality measurement, such as that of Francis et al. (2004) to identify whether, in the presence of a narcissistic CEO, individual earning attributes are affected differently. Lastly, future research could make a comparison of the effects of the degree of
narcissism on the quality of earnings before and after the passage of SOX. This could provide evidence whether such legislations either enhance or deter the effects of the personality trait.
REFERENCES
Ali, A., and Zhang, W. (2015). CEO tenure and earnings management. Journal of Accounting and Economics. 59: 60-79.
Amernic, J., and Craig, R. (2010). Accounting as a facilitator of extreme narcissism. Journal of Business Ethics. 96(1): 79-93.
Ang, J., Cole, R., and Lin, J. (2000). Agency costs and ownership structure. The Journal of Finance. 55(1): 81-106.
Bao, B., and Bao, D. (2004). Income smoothing, earnings quality and firm valuation. Journal of Business Finance & Accounting. 31(9-10): 1525-1557.
Basu, S. (1977). Investment performance of common stocks in relation to their price-earnings ratios: A test of the efficient market hypothesis. The Journal of Finance. 32(3): 663-682. Beaver, W. (2002). Perspectives on recent capital market research. The Accounting Review.
77(2): 453-474.
Bebchuk, L., Fried, J., and Walker, D. (2002). Managerial power and rent extraction in the design of executive compensation. University of Chicago Law Review. 69(3): 751-846. Bendisch, F., Larsen, G., and Trueman, M. (2013). Fame and fortune: A conceptual model of
CEO brands. European Journal of Marketing. 47(3): 596-614.
Bertrand, M. (2009). CEOs. Annual Review of Economics. 1(1): 121-150.
Blankespoor, E., and De Haan, E. (2015). CEO Visibility: Are Media Stars Born or Made? Unpublished working paper, Rock Center for Corporate Governance at Stanford University, Stanford. Available at SSRN: http://ssrn.com/abstract=2577994
Boivie, S., Graffin, S., and Gentry, R. (2016). Understanding the direction, magnitude, and joint effects of reputation when multiple actors’ reputations collide. Academy of Management Journal. 59(1): 188-206.
Bushman, R., and Smith, A. (2001). Financial accounting information and corporate governance. Journal of Accounting and Economics. 32(1): 237-333.
Campbell, W., Goodie, A., and Foster, J. (2004). Narcissism, confidence, and risk attitude. Journal of Behavioral Decision Making. 17: 297-311.
Chang, H., Fernando, G., and Liao, W. (2009). Sarbanes–Oxley Act, perceived earnings quality and cost of capital. Review of Accounting and Finance. 8(3): 216-231.
Chang, W., Hayes, R., and Hillegeist, S. (2016). Financial distress risk and new CEO compensation. Management Science. 62(2): 479-501.
Chatterjee, A., and Hambrick, D. (2007). It’s all about me: Narcissistic chief executive officers and their effects on company strategy and performance. Administrative Science Quarterly. 52: 351-386.
