C
ORPORATE
G
OVERNANCE AND
C
ORPORATE
S
OCIAL
R
ESPONSIBILITY
An examination of the relation between
earnings management, corporate governance and CSR
Judith Harmsma - 10266518 MSc A&C - Accountancy track
University of Amsterdam Amsterdam Business School Faculty of Business & Economics
22nd of June – final version Word count: 14,382
Under supervision of: Dr. Alexandros Sikalidis
2
Statement of originality
This document is written by student Judith Harmsma who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Abstract
There are two main views why firms engage in corporate social responsibility (CSR) activities. One is that management has a moral incentive and the other is that management engages in CSR activities because of opportunistic reasons. This study combines CSR activities with corporate governance to examine what the effect of these variables is on earnings management. It measures corporate governance quality based on a relatively new score system. This score is taken from the ASSET4 database and combines 68 key performance indicator into a single corporate governance score. The results indicate that there is evidence for the opportunistic reason of CSR engagement regarding real earnings management and for the moral incentives regarding accrual-based earnings management. Furthermore, corporate governance seems to be negatively associated with real earnings management and positively with accrual-based earnings management.
This research contributes to the existing literature by combining corporate governance and CSR activities. Previous literature examined corporate governance as a control variable, but not as a variable of interest. To my knowledge, this is one of the few studies that combines CSR activities and corporate governance score. Second, it finds a correlation between CSR and corporate governance, which indicates that the two variables are complementary. This relationship has not been examined thoroughly by previous literature. Third, it employs a new measure for corporate governance, which is backed by a sensitivity analysis. This measure combines information about the structure and function of the board of directors, the compensation policy and shareholder rights into one single corporate governance score.
Keywords: Corporate governance, CSR activities, Corporate Social Responsibility, Real Earnings Management, Accrual-Based Earnings Management
4
Table of contents
1 Introduction 5
2 Literature review & hypothesis development 8
2.1 Earnings management 8
2.1.1 Evidence 8
2.1.2 Incentives 10
2.1.3 Consequences 12
2.2 Corporate governance quality 13
2.2.1 Independent board of directors 15
2.2.2 Audit committee 16
2.2.3 Board diversity 17
2.3 Corporate Social Responsibility 18
2.3.1 CSR framework 19 2.3.2 Motivation to engage in CSR 20 2.4 Hypothesis development 22 3 Methodology 24 3.1 Sample selection 24 3.2 Research design 25
3.2.1 Measuring earnings management 25
3.2.2 Proxies for corporate governance quality 26
3.2.3 CSR activities 26 3.3 Regression model 27 3.3.1 Control variables 29 3.3.2 Empirical models 29 4 Results 31 4.1 Descriptive statistics 31
4.2 The relation between accrual-based earnings management and CSR- and CG score 33
4.3 The relation between real earnings management and CSR- and CG score 36
4.4 Sensitivity analysis 39
5 Conclusion and discussion 43
References 45
5
1 Introduction
This study examines the impact of firms’ engagement in corporate social responsibility activities and the effect of a good corporate governance structure on earnings management. The motivation behind this study is the growing interest in corporate social responsibility activities from society in general as well as from management. In particular, stakeholders are demanding more and more social responsibility from firms (Bondy, 2008) and this increases the pressure for management to act morally. According to Hoi et al. (2013), there are two main reasons for firms to engage in CSR activities. The first reason is the strategic use of CSR to gain support from stakeholders. It is argued that CSR reporting may be a form of public relations to help a firm gain economic value (Holt et al., 2004; Prior et al., 2008). This indicates that the sole purpose of CSR might be to fuel profit instead of participating in CSR from a moral point of view. The second possibility is that firms may actually have a moral reason to engage in CSR. Graafland and Van de Ven (2006) found evidence that firms managed by people with a strong moral view are positively associated with engagement in CSR activities. Furthermore, Kim et al. (2012) found that there is a negative relationship between earnings management and CSR activities.
The effects of a good corporate governance structure are well examined in prior literature. Due to information asymmetry and agency problems, shareholders are unable to perfectly monitor the actions of management (Hart, 1995). This can lead to opportunistic behavior of management. Corporate governance may help to mitigate these agency problems. There are numerous different proxies to measure corporate governance and the results are mixed. For example, most studies find that the independence of the board of directors is beneficial to a firm, because it reduces chances of fraudulent behavior (Beasley, 1996). However, there are also studies that do not find any association between firm performance and corporate governance structure (Hermalin & Weisbach, 1991). Since the results do not complement each other, I will employ a relatively new and unused measure of corporate governance taken from the ASSET4 database. This measure combines data about board structure, board functions, compensation policy and the vision and strategy of the board into one pillar score, which covers all important proxies for corporate governance.
6
The research field of earnings management has been examined extensively. However, it is an important topic that could seriously harm firm value in the long run (Cohen & Zarowin, 2010; Ge & Kim, 2014). CSR activities are a relatively new field of interest in accounting research and literature on the relation with earnings management is mixed. Kim et al. (2012) find that firms engaging in CSR activities are less involved in both real and accrual-based earnings management, which indicates that CSR firms are acting more ethical. However, there is also evidence that CSR might be a deflection strategy and firms are using it to cover up earnings management (Prior et al. 2008). Furthermore, the measure I will employ to proxy for corporate governance has not been used often, which brings me to the following research question: What is the relation between a high corporate governance score and corporate social responsibility activities and earnings management?
To examine this question, I merged the Datastream and Compustat databases. This left me with a sample of 4,350 observations taken from 650 different firms. The sample consists of US firms and covers the period from 2004 to 2013. All observations have data available about CSR activities and received a pillar score for corporate governance in ASSET4.
The results show that there is a significant positive relation between corporate governance and accrual-based earnings management as well as a positive relation between CSR and real earnings manipulation. Furthermore, there is a significant negative relation between CSR and accrual-based earnings management as well as between corporate governance and real earnings management.
My study adds to the literature in two ways. First, it uses a relatively new measure of corporate governance to test my hypothesis. Corporate governance is a well examined topic. However, there are numerous measures of corporate governance and the effects of different measures are mixed. In my sensitivity analysis, I also provide evidence that this measure might be a more powerful indicator of good corporate governance compared to the one that is used as a control variable in the paper of Kim et al. (2012). Second, it shows that there is a significant correlation between CSR activities and corporate governance score, which indicates that these two are complementary. Third, this study provides new evidence that support both the notion of strategic use of CSR activities as well as ethical behavior under certain circumstances.
The remainder of this paper is organized as follows. I will start with a literature review in which I will further explain earnings management, corporate governance and corporate social
7
responsibility activities. I will then move forward to my hypothesis development in which I will present the hypotheses which will be examined in this study. Then, the methodology will be explained and the regression models which will be used are discussed. After this, I will present the results of my analysis which will be followed by a sensitivity analysis where I will examine a different measure of corporate governance to assess which of the two measures is more powerful. Finally, the conclusion will be presented as well as suggestions for further research.
8
2 Literature review and hypothesis development
2.1 Earnings management
Earnings management can be divided into two categories: earnings management through accruals and earnings management through real activities manipulation. Accrual-based earnings management involves accounting policies and is affected by the 'iron law'. This law states that accruals will always reverse (Scott, 1997). So if it is used opportunistically to manage earnings upward in one period, it will cause earnings to go down in the next period. On the other hand, real earnings management involves real cash flows. This means that managers manipulate earnings using ‘real’ money. For example, by cutting on research and development expenses, managers can show a higher profit. Scott (1997) states that real earnings management can be costly, because it interferes with the long-term interests of the firm.
The remainder of this section is organized as follows. First, I will provide evidence of earnings management, then incentives for management to engage in earnings management will be discussed. Finally, the possible consequences of earnings management are explained.
2.1.1 Evidence
Ownership of a firm is usually scattered and therefore shareholders are not fully able to monitor the actions that management take and this can lead to an agency problem. Jensen and Meckling (1979) define an agency relationship as a contract between a principal and an agent. Because the principal is not always fully able to observe the actions of the agent, the agent might act in his/her own interest, regardless of whether those actions might hurt the principal. There are two types of agency problems (Eisenhardt, 1989): the first problem is related to a conflict between the goals of the principal and that of the agent. When it is costly to observe the actions of management, it is hard for the principal to verify that the agent has behaved in a desirable way. The second problem arises when the principal and the agent have different stances towards risk sharing. These differences may lead to misalignment between the agent and principal’s interests.
9
A possible outcome of this agency problem is that management may engage in earnings management in order to meet certain objectives (Scott, 1997). In summary, the inability of shareholders to monitor management is used opportunistically by management to engage in earnings management. When managers engage in real earnings management, they value private costs of not engaging in real earnings management to be bigger than the long-term costs that the company will have to bear in the future (Roychowdhury, 2006). Past research has shown that management opportunistically engages in real earnings management (Roychowdhury, 2006; Graham et al. 2005).
Prior studies suggest that managers make a trade-off between the short-term gains of positive earnings and the long-term impact of real earnings management on a firm. Graham et al. (2005) conducted a survey with 400 chief financial officers and found evidence that CFOs believe that earnings are the most important metric in evaluating performance. Therefore, CFOs prefer to have smooth earnings over smooth cash flows. The survey revealed that there is a 'near-obsessive' focus on earnings per share, because earnings per share are relatively simple to understand and can be compared across companies. Graham et al. (2005) revealed that 78 per cent of the interviewees would sacrifice value, and thus engage in real earnings management, in order to achieve a smooth earnings path.
Recently, there has been a shift from accrual-based earnings management to real earnings management due to the implementation of the Sarbanes Oxley Act (SOX). After some big accounting failures, the Act was implemented to restore public trust by making financial information more reliable and accurate (Cohen, Dey & Lys, 2008). The increased scrutiny on firms reduced the discretion that management can exercise in their accounting policies. Cohen, Dey and Lys (2008) argue that the shift towards real earnings management is the result of the increased scrutiny. Real earnings management is harder to detect than accrual-based earnings management. For example, it is difficult to determine whether a cut in research and development expenses is the result of earnings manipulation or the adoption of a different investment strategy. Ewert and Wagenhofer (2005) confirm these results and find evidence that tighter accounting regulations result in a shift from accrual-based to real earnings management. Earnings quality increases when regulations are tighter, and therefore the marginal benefits of engaging in earnings management also increase. Since tighter regulation decrease the discretion management can use, using real earnings manipulation is more effective. Zang (2011) finds that management
10
makes a trade-off between real and accrual based-earnings management based on their relative costs. She attributes the increase in real earnings management post-SOX to the increased scrutiny that firms face. Moreover, she also finds evidence that accrual-based and real earnings management can be seen as substitutes. Contrary to real earnings management, accrual-based manipulation can also be executed after the fiscal year ends. Therefore, Zang (2011) suggests that management decides on the level of accrual-based earnings management based on the level of real earnings manipulation they already engaged in.
Another example that illustrates the trade-off between real activities and accrual-based earnings management is the study of Chi, Lisic and Pevzner (2011). Their study finds that higher audit quality is associated with more real earnings management. The intuition behind this relationship is that higher audit quality reduces the flexibility and discretion that management can apply in their financial statements. Graham et al. (2005) confirm this. In their survey, 55.3 per cent of the questioned CFOs indicated that they are willing to postpone certain projects to show a positive bottom-line earnings number at the end of the year. Cohen et al. (2010) examined the use of advertising expenses to meet earnings benchmarks. Their results are twofold: an increase in advertising expenses can boost sales for more mature firms, while a decrease in advertising expenses leads to an immediate higher earnings level. The latter is usually the case for younger firm. Cohen et al. (2010) found that firms that are in an early stage of their life, cut on advertising expenditures to avoid a loss and to meet the level of the previous year’s earnings, while firms that are in a later stage use advertising expenses to boost revenues and earnings. Dechow and Sloan (1991) find evidence that managers tend to spend less on R&D towards the ending of their career in a company and thus participate in real earnings management.
2.1.2 Incentives
Missing the zero earnings benchmark or the forecast of analysts has serious implications for a firm. Several studies show the importance for a firm to meet or beat a certain earnings benchmark and the consequences if they fail to meet the benchmark (Matsumoto, 2002; Skinner and Sloan, 2002). Market reactions are more severe when firms announce bad news compared to
11
the release of good news, especially for firms that reported growing earnings in the past periods (Skinner and Sloan, 2002). Therefore, management has an incentive to just meet or beat the earnings benchmark. This extreme reaction can be explained as evidence that the market expects firms to engage in earnings management in order to meet the benchmark (Graham et al., 2005). If a firm is not able to find money to meet the benchmark, it is an indication that something is wrong. Management therefore makes the decision to sacrifice some economic value in order to avoid the costs of the overreaction of the market.
If a firm is just under the zero-earnings benchmark, it has an incentive to engage in earnings management in order to meet the target. Several studies find evidence that this is the case. Burgstahler and Dichev (1997) find that 30 to 44 percent of the firms that are just below the zero pre-managed earnings engage in earnings management to report positive earnings. More specifically, management uses cash flows from operation to meet the target and thus prefers to engage in real earnings manipulation rather than accrual-based earnings management. Degeorge et al. (1999) nuance this finding by stating that management only engages in earnings management when earnings are falling just under zero. Furthermore, Degeorge et al. (1999) hierarchically order the thresholds that incentivize management to engage in earnings management. Firms view it as most important to report a profit, then to report quarterly profits and finally to meet the analysts benchmark. Therefore, the zero earnings benchmark is viewed as most important. Burgstahler and Eames (2006) report that firms use both accruals and cash flows to manage earnings upwards.
A horizon problem occurs when the expected tenure of the manager is shorter than the firm’s investment horizon (Baber et al., 1998). This causes management to be more likely to accept a project that has a small net present value, but does increase earnings in the short run. An example of the horizon problem is the case where a CEO retires within a year (Butler & Newman, 1989). The firm is not able to sanction the CEO if he or she underperforms, because the CEO will leave anyway. Therefore, the manager might focus only on the short-term personal gain of a project without taking into account the possible negative consequences for the firm in the long run. Dechow and Sloan (1991) found evidence that managers who are closer to retirement are more likely to cut on R&D expenditures and thus engage in real earnings management. This indicates that earnings-based incentives are associated with a short-term performance focus.
12
Another incentive for management to engage in earnings management is the structure of the compensation plan. Cheng and Warfield (2005) conducted a study to examine the relation between equity incentives and earnings management. They found that CEOs are more likely to sell shares when they have high equity incentives. Moreover, managers that have high equity incentives are likely to meet or just beat the analysts’ forecast. The study also shows that managers with high equity incentives do not report large earnings surprises. This indicates that management participates in earnings management in order to report a profit or to smooth earnings.
2.1.3 Consequences
Research about the consequences of earnings management in general is extensive. However, there is much less information about the consequences of real earnings management. Although the intuition behind the consequences of real earnings management implicates that it might be harmful to a firm, the empirical studies regarding subsequent performance are mixed. In general, there are two general views regarding real earnings management.
The first view states that real earnings management can be used as a signalling device. Firms participating in real earnings management to just meet the earnings benchmark have better subsequent performance (Gunny, 2010). Meeting the benchmarks enhances the reputation of a firm and increases the relation with clients, debtors and shareholders. Furthermore, Gunny (2010) finds that firms that engage in real earnings management to meet the benchmark perform better than firms that do not meet the benchmark. This implicates that real earnings management is used as a signalling device to indicate superior future performance. Taylor and Xu (2010) find similar results and state that firms only occasionally participate in real earnings management. Contrary to Gunny (2010) they only find evidence that firms engaging in real earnings management do not underperform compared to non-earnings managing firms. There is no evidence that firms that participate in real earnings management perform better. Their study therefore does not support the signalling theory.
There is also evidence that managers do use real earnings management in an opportunistic way. Cohen and Zarowin (2010) examined the use of real versus accrual-based
13
earnings management prior to seasoned equity offerings (SEO). An SEO involves the issuance of new equity of a firm that is already listed. The stakes are high, because the higher the stock price, the more a firm earns. Firms that perform in real earnings management outperform their industry peers during the SEO, but underperform in the long run. This therefore indicates that real earnings management is used in an opportunistic way. Cohen and Zarowin (2010) contribute to the literature by examining the relative impact of accrual-based versus real earnings manipulation and find that the latter has a more severe impact on future performance. Zang (2010) confirms these results and finds that firms that engage in real earnings management actually underperform in the long run compared to firms that do not engage in earnings manipulation. Zhao, Chen, Zhang and Davis (2012) nuance the findings that firms engaging in real earnings manipulation over- or underperform their peers. They find evidence that both the signalling and the opportunistic behaviour theory are represented. Firms that only engage in earnings management to beat the target are performing well in subsequent periods, while firms that do engage in real earnings management, but fail to meet the target, underperform in subsequent periods.
Since real earnings management involves the manipulation of cash flows, it is also impacting the credit worthiness of firms. Ge and Kim (2014) examined the relation between real earnings management and the issuance of new corporate bonds. They show that overproduction is associated with a lower credit rating and that both overproduction and sales manipulation are associated with higher bond yield spread. This indicates that credit rating agencies and issuers of bonds perceive real earnings management as a risk-indicator and demand a risk premium. Real earnings management thus increases the interest costs and therefore might decrease the firm value.
2.2 Corporate governance quality
Management and ownership is usually separated, so in order to make management act in the interest of financiers, both have to sign a contract (Shleifer & Vishny, 1997). An ideal contract is complete, which means that it consists of all possible scenarios and the way management should act if these take place. The second option is to make a manager sign a contract, which states that
14
the investors hold all residual control (Shleifer & Vishny, 1997). That would make investors eligible to any right to decide how to the firm invests the money. However, there are a few problems with both these approaches. In a perfect world, without any agency problems, corporate governance is not necessary. However, when there are agency problems and contracts are not complete, corporate governance structure does play a role in reducing opportunistic behaviour. Hart (1995) defines corporate governance structures as a mechanism for making decisions that are not in the initial contract between agent and principal. It provides the right to decide how assets should be used. In public companies, ownership is scattered between shareholders. None of these shareholders have an incentive to monitor a firm, because there is a free-ride problem: monitoring is costly and it is a public good, because annual reports are free to everyone if they are free for one person. So if one shareholders monitors the firm, each shareholder benefits if the monitoring leads to improved performance. This lack of monitoring leads to agency problems and may result in managers acting in their own interest. Corporate governance is therefore a mechanism to ensure that shareholders and investors get their investment back (Shleifer & Vishny, 1997). Figure 1 shows the relation between the board of directors and the balance sheet. Management is in charge of deciding how to invest and spend capital, while the external part of the figure shows how firms raise capital.
15 2.2.1 Independent board of directors
The board of directors oversees management and is in charge of monitoring and advising management. It can also hire and fire the management team and decide on its compensation (Gillan, 2006; Jensen, 1993). This makes the board of directors an important part of corporate governance. Baysinger and Butler (1985) divide board members into three categories: executive, monitoring and instrumental board members. The executive directors are closely aligned with top management, because the firm formally employs them. Monitoring directors are outside directors, who are not related to management by blood or marriage or who represent an organization that is not highly affiliated with the firm (Baysinger & Butler, 1985). Furthermore, members who are part of the instrumental component are placed on the board for their functionality. These members have valuable knowledge about managing a firm and are consulted to improve decision-making. Baysinger and Butler (1985) state that the optimal board consists of a mix of different types of directors. However, they also state that firms with a higher portion of independent directors are more likely to perform better in the long run.
Literature on the number of outside directors as a proxy of corporate governance is extensive. Outside directors are defined as not being employed by the firm, either currently or in the past (Rosenstein & Wyatt, 1990). Furthermore, the only connection between the director and the firm is the seat on the board. Beasley (1996) examined the relation between board composition and financial statement fraud and found that fraudulent firms had a significantly larger number of inside directors on their board compared to firms that did not engage in fraudulent practices. Xie, Davidson and DaDalt (2003) examined the relation between corporate governance and accrual-based earnings management and found that the level of outside directors is negatively associated with earnings management. Additionally, Core et al. (1999) found evidence that CEO compensation is negatively associated with the number of outside directors, thus indicating that outside directors are more independent from management and are a factor in mitigating opportunistic behaviour of management.
Not all studies confirm the finding that outside directors are associated with higher firm performance. Hermalin and Weisbach (1991) find that there is no difference in firm performance regardless of the level of outside and inside directors. They explain this by stating that the CEO is in charge of the appointment of directors, thus implicating that outside directors are not
16
entirely independent. It might be possible to mitigate this problem by only defining outside members as independent when they are hired before the CEO was appointed.
Another crucial element concerning to the independence of the board of the directors is the appointment of the chairman (Gillan, 2006). If one individual is both the chairman of the board of directors and the CEO of a firm, this might affect the decision to fire an underperforming CEO (Goyal & Park, 2002). CEO turnover is less associated with firm performance in firms where the CEO is also the chairman, compared to firms where these functions are separated. That suggests that the board is less likely to replace a CEO who is underperforming if he or she also holds the position of chairman.
A board interlock happens when an individual has a seat in more than one board of directors (Fich & White, 2005). Moreover, a reciprocal interlock occurs when the CEO of firm A is a board member in firm B, while the CEO of firm B is a board member of firm A. Fich and White (2005) find that reciprocal interlocks are associated with a higher level of CEO involvement when directors are appointed. This indicates that CEOs use reciprocal interlock opportunistically for their own gain, instead of acting in the interest of shareholders.
2.2.2 Audit committee characteristics
All listed firms in the US are required to have an audit committee. Audit committees have the responsibility to provide oversight on matters related to the credibility of financial reporting. It can therefore be considered to be the ultimate monitor of the financial statement process. It is expected to increase the quality of financial reporting and this makes it an important body in the corporate governance structure (Mangena & Pike, 2005).
All big US stock markets require that the audit committee of firms include at least one member with a financial background (Agrawal & Chadha, 2005). It is assumed that members who have no financial or economic experience are less likely to report problems in financial reporting. Agrawal and Chadha (2005) find that audit committees or boards that contain one or more financial experts are significantly less likely to have restatements in future periods. Mangena and Pike (2005) confirm this finding and state that interim disclosure is related to audit committee financial expertise. Interim disclosures are important in mitigating the information
17
asymmetry between managers and investors and thus improve financial reporting. Bédard et al. (2004) find evidence that aggressive earnings management is negatively associated with the presence of financial expertise in the audit committee. Additionally, Defond et al. (2005) find a positive relation between cumulative abnormal returns and the appointment of a financial expert on the audit committee. This suggests that markets value the presence of a financial expert in the audit committee, because this might increase the credibility of the financial statements.
The Sarbanes Oxley Acts requires all members of the audit committee to be independent (Bédard et al., 2004). The intuition is that independent members of the audit committee are better monitors of the firm’s financial accounting process (Klein, 2002). Just like outside directors of the board, audit committee members are viewed as being independent when they are not currently employed by the firm or in the past. Bédard et al. (2004) find that the requirement of SOX to have 100 percent independent audit committee members is significantly negatively related to the likelihood of aggressive earnings management, while there is no significant effect if the committee consists of 55 to 99 percent independent members. In addition, Carcello and Neal (2000) found that the number of affiliated audit committee members is negatively associated with the issuance of a going concern opinion, indicating that less independent audit committee members are associated with financial distress in firms
2.2.3 Board diversity
The rationale behind this component of corporate governance is that more diverse boards are more likely to make good decisions in complex cases due to the different points of view (Ray, 2005). Decisions will be made after discussions, instead of on the basis of a homogenous viewpoint. Heterogeneous boards can also contribute to social and ethical responsibilities of a company, because diversity helps the board focus on a broader perspective, instead of just the financial aspects of a strategy. Supervision of the board may therefore be improved if a certain level of conflict arises that enhances discussion and provides space to discuss different ideas (Erhardt et al., 2003).
Legislation also tries to enhance board diversity. An example of this is the recent adoption of the women quotas in Germany that forced companies to allocate 30 per cent of all
18
seats in the board to be occupied by women (The Guardian, 2014). Empirical studies found evidence for the benefits that gender diversity can bring to a company. Carter et al. (2003) examined the impact of board diversity on firm value and found that there is a positive relationship between the number of women and minorities in the board and firm value, indicating that a more diverse board may lead to better financial performance. Additionally, Gul et al. (2011) found evidence that stock prices of companies with a gender-diverse board of directors are more informative. Gender diversification enhances the monitoring function of a company, which leads to an increase in transparency and therefore improves the quality of financial statements. Adams and Ferreira (2009) found that gender diversification enhances the monitoring role for three reasons. First of all, their study showed that female directors have a higher attendance rate than male directors. Second, the male attendance rate is positively associated with the number of female directors. And finally, women are more likely to join monitoring committees, such as the audit committee. These results indicate that the monitoring quality of the board is positively associated with the number of female directors.
Although a lot of studies found positive effects related to the number of female directors, there are also studies that found no comparable results. Chapple and Humphrey (2014) found no significant relation between the number of female directors and firm performance.
Furthermore, companies add female directors to the board to respond to outside pressure. Farrell and Hersch (2005) state that firms, once the board reaches the minimum level of diversity, stop considering ways to further increase board diversity. Therefore, the demand to add female directors to the board seem to be endogenous, instead of exogenous. Additionally, Farrell and Hersch (2005) did not find any abnormal market reaction with the announcement of the addition of a woman to the board, which also adds to the believe that engagement in board diversity may be endogenous in nature.
2.3 Corporate Social Responsibility
Corporate social responsibility (CSR) is gaining interest in larger firms, as well as smaller firms. Hoi, Wu and Zhang (2013, p. 2028) define CSR as ‘the shared belief within the organization about the “right” course of action that takes into account the economic, social, environmental,
19
and other externalized impacts of the company’s activities.’ Although this definition suggests that firms engage in CSR for ethical reasons, prior research has shown that this is not always the case. In general, there are two main theories regarding motivation for engagement in CSR activities: firms may engage in CSR for opportunistic reasons (Bondy, 2008) or from a more ethical point of view (Kim et al., 2012).
This section will start with an explanation of the existing frameworks regarding corporate social responsibility, followed by a description of the literature regarding the motivation for managers to engage in CSR.
2.3.1 CSR frameworks
There are several frameworks that try to describe different domains of CSR activities. Carroll (1979) developed a model to conceptualize social responsibilities. His definition is as follows: “The social responsibility of businesses encompasses the economic, legal, ethical and discretionary expectations that society has of organizations at a given point in time.” Limitations of this framework are that it does not take into account the overlapping nature of activities and that there is a hierarchical order that ranks different activities (Carroll & Schwarz, 2003).
To address the shortcomings of Carroll’s (1979) framework, Schwartz and Carroll (2003) further develop CSR into three domains: economic, legal and ethical. This is visualized in figure 2. The purely economic domain includes all activities that generate a direct or indirect economic
20
benefit. An activity is expected to be economically beneficial when 1) the profits are maximized and 2) when share value is maximized. Corporate social activities can help to gain economic benefits in several ways, for example by providing a positive public image of the firm, which might increase the likelihood that consumers consider buying products from the firm. Purely economic activities are considered to be unethical and can even be illegal. Furthermore, they only take into account the economic benefits associated with the activities and ignore all other responsibilities.
The legal component of the framework contains the reaction of the firm to mandatory laws and the expectations of society. Compliance with the law can be passive, restrictive or opportunistic in nature. Opportunistic compliance happens when companies do act within the boundaries of the law, but do not obey the spirit of the law (Schwartz & Carroll, 2003). An example of this is finding loopholes in legislation and taking advantage by performing certain activities. It is also possible to opportunistically comply by choosing to perform certain activities in areas where the legal standards are weaker. Purely legal activities only take place because of the legal system.
Furthermore, the ethical component of the diagram consists of the ethical responsibilities of the firm, which are expected by stakeholders and society (Schwartz & Carroll, 2003). Purely ethical behaviour does not evoke any economic benefits. In practice, purely ethical activities are not very common, because even ethical activities will probably lead to economic benefits in the long-term. An ideal situation arises when an activity is economic, legal and ethical. Firms should try to operate in the middle part of the diagram whenever possible, because it addresses profitability, while acting within legal boundaries and obeying ethical standards (Schwartz & Carroll, 2003).
2.3.2 Motivation to engage in CSR
The benefits of communicating CSR activities to consumers are numerous. It helps building a credible and positive image of the company and therefore might boost profits (Du, Bhattacharya & Sen, 2010). Furthermore, a positive image of the company may help to attract new employees. Since high quality personnel are able to choose between different companies, reporting on CSR
21
activities can be a competitive advantage in there as well (Greening & Turban, 2000). The rationale behind this is that CSR conscious firms are more likely to be ethical and therefore treat their employees better.
As corporate social responsibility is gaining more power, society increasingly expects businesses to engage in CSR activities (Bondy, 2008). As has been discussed before, the benefits of CSR activities are numerous and this could implicate that firms may use CSR opportunistically. Perks et al. (2013) impose that CSR can be used to deflect attention from controversial corporate social irresponsible behaviour. The legitimacy theory states that organizations try to create the impression that they are socially responsible by disclosing information about their responsible behaviour. Communication of information about their actions regarding social responsibilities are used to gain support or approval from stakeholders or to hide less responsible actions. Another important aspect of CSR reporting is the degree of usefulness that customers perceive. Holt et al. (2004) argue that CSR reporting may be just another form of public relations in to help gain economic value. They give an example of a company that reduces package material and advertises that it is doing this because of environmental reasons. However, the real reason for reducing packing material may be that it reduces costs and thus increases profit. Therefore, Holt et al. (2004) argue that CSR activities are only a way to increase profits, consistent with the thought that the sole goal of a firm is to deliver value for the shareholders. Following Schwartz and Carroll’s framework (2004), this finding is also consistent with the idea that the activities are purely economic and do not take any other responsibilities in account. Prior et al. (2008) found a positive relationship between earnings management and CSR. They state that managerial entrenchment is the main reason for managers to engage in CSR. By satisfying stakeholders’ interests, management secures its position, because these stakeholders may protect management against disciplinary initiatives from shareholders.
Not all research finds that managers only engage in CSR activities for opportunistic reasons. Jones (1995) argues that firms have a moral reason to engage in CSR activities. According to the theory of the firm, a company can be viewed as a nexus of contracts. Efficient contracting will reduce agency problems and commitment problems, such as opportunism. If firms solve commitment problems efficiently, they gain a competitive advantage over firms that do not solve these problems. Jones (1995) argues that commitment problems are easier to resolve with ethical solutions, rather than attempts to restrain opportunism. Therefore, firms will have a
22
competitive advantage if they contract with their stakeholders on mutual trust and cooperation. Corporate social performance activities can be viewed as attempts to strengthen the relationship between firms and stakeholders. Graafland and Van de Ven (2006) find that firms managed by people with a strong moral view on CSR are significantly more involved in CSR activities. This implies that firms do engage in CSR activities for moral reasons, instead of strategic reasons. Furthermore, Kim et al. (2012) examined the relation between earnings management and the level of CSR activities performed by the firm and found that there is a negative relation, thus indicating that CSR firms are more moral.
2.4 Hypothesis development
As has been discussed in the previous section, literature on the relation between corporate governance and earnings management is quite extensive. However, there are numerous proxies to measure corporate governance and there are multiple issues with that. For example, the database of Compustat changed the way it measures certain factors in 2006, which makes it hard to compare data from different time periods. Therefore, I will use a relatively new and unused measure for corporate governance, which is taken from the ASSET4 database. This measure combines board structure, board function, compensation policy and the vision and strategy of management into one pillar score which varies between zero and one hundred.
Although the measurement is different from other literature, I still expect to find the same results and thereby strengthen the assumption about the relation between corporate governance and earnings management. Corporate governance is well-known for its mitigating effect on earnings management and therefore I propose the following hypotheses which predict that earnings management is negatively associated with a good corporate governance score. To assess the power of this measure, I will also include a sensitivity test to compare the corporate governance score with another score taken from the KLD database. This variable is also used by Kim et al. (2012) as a control variable.
H1a: firms with a high corporate governance score are less likely to be involved in accrual-based earnings management
23
H1b: firms with a high corporate governance score are less likely to engage in real earnings management
Literature on the amount of CSR activities in relation to earnings management is not as extensive as research on corporate governance. However, there are a few studies that examined this and their results are mixed. One way to view engagement in CSR is as an entrenchment mechanism which will help management to obtain and maintain support from stakeholders (Prior et al., 2008). This way, management prevents the shareholders from firing them when they have engaged in earnings management. Therefore, I will propose the following hypotheses:
H2a: firms that score high on CSR activities are more likely to be involved in accrual-based earnings management
H2b: firms that score high on CSR activities are more likely to be involved in real earnings management
Figure 3 is a graphical representation of the relations I am going to examine in the following sections. As is shown in the figure, I expect a positive relationship between CSR activities and earnings management and a negative relationship between corporate governance and earnings management.
24
3 Methodology
This section will explain how the research is conducted. It will start with an explanation of the sample selection. After that, I will introduce proxies for measuring earnings management, corporate governance and corporate social responsibility. Then I will introduce my regression models as well as the control variables I am going to use. Finally, the empirical models are presented.
3.1 Sample selection
To conduct the regressions I will use a sample of US firms. I extracted data from three different databases: Compustat, ASSET4 and the KLD-database. The ASSET4 database was the constraining factor in the period selection, since there is no data available before 2004. Therefore, the time period is from 2004 until 2013. I merged the databases and deleted missing values as well as firms with a standard industry classification (SIC) code between 6000 and 6999, which indicate financial institutions, following Kim et al. (2012). The rationale behind is that the accruals of financial institutions are uncomparable with those of other industries. Following Roychowdhury (2006), as long as selling, general and administrative expenses are available, advertising costs and R&D expenses are set to zero. If SG&A is not available, the observation is deleted. This way, I ended up with a sample of 4350 observations taken from a total of 635 firms. These observations contain information regarding accrual-based earnings management, real earnings management, CSR activities, corporate governance score and four different control variables.
3.2 Research design
25
As has been mentioned before, there are two ways to engage in earnings management: either through accruals or by manipulating real activities. Total accruals are the sum of nondiscretionary and discretionary accruals. Discretionary accruals are those that management can influence by participating in earnings management. The model that will be used to detect discretionary accruals is the Modified Jones Model, which is further discussed in the next section. Dechow, Sloan and Sweeney (1995) examine different models and find that the Modified Jones Model is the most powerful in estimating accrual-based earnings management. Following Jones (1991), I will calculate total accruals as the change in working capital before income taxes minus total depreciation expenses.
There are three ways for firms to engage in real earnings management (Roychowdhury, 2006). The first way is to boost sales in order to report a higher profit at the end of the year. Higher sales can be achieved by offering price discounts to clients or to offer them more lenient credit terms. Price discounts decrease the margin between the sales and production costs, but do generate temporarily higher earnings. Lenient credit terms are, according to Roychowdhury (2006) essentially price discounts, which last over the life of the sales. This results in an abnormal cash flow from operations. The second way to manage earnings upward is through overproduction (Roychowdhury, 2006). By increasing the number of units produced, the fixed overhead costs are spread over more units. This decreases the costs of goods sold and therefore increases bottom-line profit. However, the firm has to incur holding costs and is exposed to the risk that the inventory becomes obsolete. This causes production costs to be abnormally high. The final way to participate in real earnings management is by cutting on discretionary expenses. These expenses include research and development (R&D) costs, advertising expenses and maintenance costs (Roychowdhury, 2006). These expenses are usually immediately expensed, so cutting on them means a lower cash outflow in the current period. Therefore, it is tempting to postpone the activities until the new period begins.
3.2.2 Proxy for corporate governance quality
Contrary to Kim et al. (2012), I will use a comprehensive score taken from the ASSET4 database that represents the corporate governance structure. This database is part of Thomson Reuters and
26
is accessible through Datastream. I used the corporate governance pillar score to measure the strength of the corporate governance structure of the firms in the sample. The database uses 68 key performance indicators (KPI). These KPIs are divided into four different fields. The first field scores the board of directors and the board functions. Examples of data are scores regarding the independence of the board and the number of board meetings. The second field is regarding the board structure. This section scores, for example, the diversity ratio of the board and whether the chairman of the board is an ex-CEO. The third pillar is about the compensation policy of the board of directors and scores firms based on facts like the appearance of a compensation committee and whether long term objectives are associated with compensation structure. The final pillar is about shareholder rights. This section scores the firms based on, for example, the rights of shareholders to call meetings and whether there is a confidential voting policy.
As a sensitivity analysis, I will use the KLD corporate governance score to measure the strength of corporate governance. This is similar to the corporate governance score Kim et al. (2012) use as a control variable.
3.2.3 CSR activities
Most studies that examine CSR activities use the KLD database. This database rates the environmental, social and governance performance of firms. It contains information about seven qualitative issues regarding strength and concern, as well as information about controversial business issues. Following Kim et al. (2012), the CSR score is calculated as the total number of strengths minus total number of concerns regarding business activities. Kim et al. (2012) exclude the data about companies that operate in controversial areas, such as the tobacco or gambling industry, which I will also do. Furthermore, because corporate governance is studied separately in my research, I will eliminate this dimension only assess the six remaining dimensions. The first issue regards communities. This dimension rates the involvement and impact that a firm has on a community. The second dimension is diversity. Firms that score high in this dimension are diverse and have at least four seats available on the board for women or minorities. Regarding the third issue, employee relations, a good retirement benefit plan can be seen as a strength, while poor union relations are judged as a concern. Furthermore, the dimension about
27
environment is about the commitment of a firm to sustainability and the reduction of environmental issues. The human rights dimension is a concern if the company operates in countries which are controversial, such as South Africa. The product dimension is regarding, for example, the availability of the product to the economically disadvantaged.
3.3 Regression models
I will use the model of Roychowdhury (2006) to measure real earnings management activities. As has been discussed before, real earnings (RAM) can be manipulated through overproduction, cutting on discretionary expenses and boosting sales. Following Roychowdhury (2006) and Dechow et al. (1998), normal cash flows are expressed as follows:
CFO/At-1 = α0 + α1(1/At-1) + β1(St/At-1) + β2(ΔSt/At-1) + ε
Where:
At-1 = total assets in period t-1 St = total revenue in period t
ΔSt = change in revenue between period t and t-1
The error term represents the abnormal level of cash flows and is defined as the actual cash flow minus the estimated cash flow from the model above.
Following Roychowdhury (2006), production costs are defined as the sum of cost of goods sold and the change in inventory. Normal production costs are therefore defined as follows and the error term represents the abnormal production costs:
PRODt /At-1 = α0 + α1(1/At-1) + β1(St/At-1) + β2(ΔSt/At-1) + β3(ΔSt-1/At-1) + ε
Finally, expected discretionary expenses are the sum of the research & development expenses, advertising expenses and the selling, general and administrative expenses. The error term represents the abnormal level of discretionary expenses.
28 DISEXPt/At-1 = α0 + α1(1/At-1) + β1(S1/At-1) + ε
Following Kim et al. (2012), the combined measure to capture all three forms of earnings manipulation is defined as follows:
COMB_RAM = AB_CFO – AB_PROD + AB_EXP
Where:
AB_CFO = abnormal level of cash flow from operations AB_PROD = abnormal level of production costs
AB_EXP = abnormal level of discretionary expenses
I will use the modified Jones model of DeFond and Subramanyam (1998) to estimate discretionary accruals (DA):
DAt = TAt – [β1(1/At-1) + β2(ΔREVt - ΔRECt ) + β3PPEt+ β4ROAt-1]
Where:
TAt = total accruals in period t
ΔREVt = change in revenue between period t and t-1
ΔRECt = change in accounts receivable between period t and t-1 PPEt = property, plant and equipment in period t
ROAt-1 = return on assets in period t
To examine the relative impact of corporate governance and CSR on earnings management, I developed the following models. First, I will examine the impact of CSR and corporate governance separately. Then, I will run a model that includes both variables to examine whether the explanatory power of the model is higher compared to the regressions based on the separate dimensions.
29 3.3.1 Control variables
I will use four control variables in my study. The first variable is the market-to-book ratio (MB). This ratio is the market value of equity, which can be calculated as the number of outstanding shares times the stock price, divided by the book value of equity. The second control variable is SIZE. Both control for growth opportunities and firm size and therefore reduce systematic variation in abnormal cash flows (Roychowdhury, 2006).
Following Roychowdhury (2006) and Kim et al. (2012), I will also include a control variable to control for performance (ADJ_ROA). This variable is measured as net income before extraordinary items scaled by total lagged assets. Previous literature has shown that abnormal accruals are positively correlated with firm performance and therefore it is necessary to reduce the measurement errors (Roychowdhury, 2006).
Lastly, again following Kim et al. (2012), I will include a control variable that measures the level of long term debt divided by total lagged assets (LEV) to control for earnings management associated with leverage related incentives.
3.3.2 Empirical models
The above mentioned measures for earnings management and the control variables result in the following empirical models. Either discretionary accruals or real earnings management proxies are the dependent variable, and CG and CSR score are together with the control variables the independent variables.
DA = CSR_SCORE + COMB_RAM + MB + SIZE + ADJ_ROA + LEV + ε DA = CG_SCORE + COMB_RAM + MB + SIZE + ADJ_ROA + LEV + ε
DA = CSR_SCORE + CG_SCORE + COMB_RAM + MB + SIZE + ADJ_ROA + LEV + ε
COMB_RAM = CSR_SCORE + DA + MB + SIZE + ADJ_ROA + LEV + ε COMB_RAM = CG_SCORE + DA + MB + SIZE + ADJ_ROA + LEV + ε
30
COMB_RAM = CSR_SCORE + CG_SCORE + DA + MB + SIZE + ADJ_RA + LEV + ε
Where:
DA = discretionary accruals, measured with the modified Jones model;
COMB_RAM = real earnings management, measured as the sum of AB_CFO, AB_PROD and AB_EXP;
AB_CFO = the abnormal level of cash flow from operations, which indicates sales boosting;
AB_PROD = the abnormal level of production costs, measured as the sum of costs of goods sold and changes in inventory;
AB_EXP = the level of abnormal discretionary expenses, which is the sum of research & development costs, advertising expenses and selling, general and administrative expenses;
CSR_SCORE = the sum of the strengths and concerns in the KLD rating; CG_SCORE = the score taken from the ASSET4 database;
MB = market-to-book equity ratio, which is the market value of equity divided by the book value of equity;
SIZE = the natural logarithm of the market value of equity (outstanding shares times the stock price);
ADJ_ROA = income before extraordinary items divided by total lagged assets; LEV = the level of long term debt scaled by total assets.
31
4 Results
This section will start with descriptive statistics, where I provide the initial results, as well as the mean and other metrics of the variables I use. Then, I will move on to the relation between accrual-based earnings management and corporate governance and CSR. After that, I will do the same for real earnings management and finally I perform a sensitivity analysis to show a different measure for corporate governance quality.
4.1 Descriptive statistics
Table 1 presents the descriptive statistics of the dependent variables, the variables I am interested in and the control variables. Panel A shows the mean values and quartiles of the variables, while panel B covers the correlation among the variables. Panel A provides initial evidence that the sample firms engage in real earnings management by boosting sales (AB_CFO) and by overproducing (AB_PROD), with a mean of -0.0439 and -0.3260 respectively. Furthermore, the descriptive statistics show evidence that firms engage in upward accrual-based earnings management with a mean of 0.4620.
The variables of interest show that the average score in the KLD database is 0.8870 with a mean of zero and a 75th percentile of 2. The corporate governance score shows that half of the sample firms have a score higher than 78.8650 on a scale of 1 to 100, which indicates that the firms in general have a strong corporate governance structure. Regarding the control variables, the descriptive statistics show that the firms are more profitable than their peers (Kim et al., 2012), with a mean adjusted return on assets (ADJ_ROA) of 0.0809. Compared to Kim et al. (2012), the distribution of the variables is slightly different. The mean of the discretionary accruals is 0.462, while Kim et al. (2012) show a mean of 0.200. Furthermore, the real earnings management measure COMB_RAM shows a mean of 0.3078, while Kim et al. (2012) report 0.2380. Additionally, the sample of Kim et al. (2012) shows a mean of -0.055 regarding the CSR score, while my sample is 0.8870. This indicates that the firms in my sample are less associated with CSR concerns.
32
Panel B shows the correlation between the different variables. CSR score is positively correlated with abnormal discretionary expenses (AB_EXP) and the combined measure of real earnings management (COMB_RAM) at a significance level of 0.01 and is negatively correlated with abnormal production (AB_PROD) and discretionary accruals (DA) on a significance level of 0.01 and 0.05 respectively. Furthermore, corporate governance score is negatively correlated with COMB_RAM and AB_CFO on a significance level of 0.01 and 0.05 respectively. Corporate governance score is positively correlated with the use of discretionary accruals on a significance level of 0.05. Additionally, there is a correlation found between CSR activities and corporate governance score, which indicates that CSR and corporate governance might be complementary. Another interesting result is that discretionary accruals and real earnings management are negatively correlated. This indicates that these two forms of earnings management are substitutes, which is consistent with previous literature (Ewert & Wagenhofer, 2005; Cohen, Dey & Lys, 2008).
Table 1
Descriptive Statistics of Selected Variables
Panel A: Full Sample
n Mean Median Std. Dev.
25th Percentile 75th Percentile Dependent Variables DA 4349 0.4620 0.3272 1.1542 0.1637 0.5572 AB_CFO 4349 -0.0439 0.0826 1.3117 -0.0219 0.1576 AB_PROD 4348 -0.3260 -0.2319 0.6365 -0.4066 -0.0992 AB_EXP 4349 0.0259 -0.0451 0.4516 -0.1434 0.1116 COMB_RAM 4348 0.3078 0.2733 1.0440 0.0321 0.6000 Variables of Interest KLD_SCORE 4350 0.8870 0.0000 3.5446 -1.000 2.000 CG_SCORE 4350 75.6446 78.8650 15.6234 68.4575 86.6800 Control variables MB 4338 3.6179 2.6647 26.1675 1.6609 4.2563 SIZE 4338 8.8841 8.7239 1.2214 8.0425 9.5877 ADJ_ROA 4349 0.0809 0.0679 0.2215 0.0283 0.1153 LEV 4330 0.2090 0.1862 0.1795 0.0886 0.2973
Panel B: Correlations among CSR Score, CG Score, Earnings Management Proxies and Control Variables 1 2 3 4 5 6 7 8 9 10 11 1. KLD_SCORE 1.000 2. CG_SCORE .271** 1.000 3. DA -.036* .037* 1.000 4. AB_CFO .025 -.032* -.621** 1.000 5. AB_PROD -.064** .002 -.652** .576** 1.000 6. AB_EXP .151** .000 .545** -.588** -.728** 1.000 7. COMB_RAM .136** -.041** -.147** .651** -.201** .137** 1.000 8. MB .028 -.011 -.006 .001 -.028 .025 .029 1.000 9. SIZE .312** .246** .093** -.098** -.108** .138** .002 .036* 1.000 10. ADJ_ROA .032* .050** .559** -.627** -.642** .491** -.185** .016 .245** 1.000 11. LEV -.073** -.082** .026 .025 .100* -.095* -.071** -.008 -.132** .123** 1.000
4.2 The relation between Accrual-Based Earnings Management and CSR- and CG-Score
Table 2 contains the results of a multivariate regression analysis of the effect of several variables on accrual-based earnings management. Panel A consists of the beta-coefficients and accompanying t-statistics of accrual based earnings management as a dependent variable, the KLD score as an independent variable and several control variables. There is a significant negative relation between KLD_SCORE and accrual based earnings management, which indicates that ethical firms, which participate in CSR activities are less associated with accrual-based earnings management than non-ethical firms.
Panel B shows a remarkable outcome, the regression indicates that there is a positive relationship, although not significant, between a good corporate governance score and engagement in accrual-based earnings management. This is strange, since there is extensive literature on the relation between corporate governance and earnings management, which show that there is a negative relationship between these variables.
The third panel examines the effect of combining two independent variables, CSR score and CG score, on accrual-based earnings management. The results show again a significant negative relation between CSR activities and accrual-based earnings management. However, there is now also a significant positive relation between corporate governance and accrual-based earnings management.
Table 2
Regression of Accrual-Based Earnings Management
Panel A: Accrual-based earnings management on CSR Panel B: Accrual-based earnings management on CG score Panel C: Accrual-based earnings management on CSR and CG score DA Coefficient (t-stat) DA Coefficient (t-stat) DA Coefficient (t-stat) KLD_SCORE -.036 (-2.743)** -.042 (-3.111)** CG_SCORE .017 (1.340) .026 (1.987)** COMB_RAM -.027 -.031 -.025
35 (-2.072)* (-2.396)* (-1.917) MB -.011 (-0.865) -.011 (-0.881) -.010 (-0.822) SIZE -.024 (-1.757) -.040 (-2.962)** -.029 (-2.068)* ADJ_ROA .576 (43.613)** .577 (43.673)** .576 (43.649)** LEV .088 (6.946)** .090 (7.078)** .090 (7.042)** Adj. R2 .328 .327 .329 N 4316 4316 4316
Note: This table examines accrual-based earnings management and reports the results of the relevant OLS regressions. ** and * indicate significance at the 1% and 5% significance level respectively. Variable definitions can be found in the Appendix.
The results show that there is significant negative relationship between accrual-based earnings management and CSR score. This indicates that companies that are involved in social activities are less willing to manage earnings using accruals. These results confirm the findings of Kim et al. (2012), but they are inconsistent with my hypothesis. My results indicate that firms engaging in moral social activities, are intrinsically motivated to act ethical. I therefore have to reject hypothesis 2a.
Regarding corporate governance, the results show no relation with earnings management when CSR score is not included. However, there is a significant positive relation between corporate governance score and accrual-based earnings management when both variables of interest are included in the regression. This result indicates that firms with a good corporate governance score are more likely to be involved in accrual-based earnings management. I therefore reject hypothesis 1a. This result contradicts previous literature. One possible explanation for this result is that I did not use suspect-firm years. As has been discussed before, previous literature has shown that management is more likely to engage in earnings management when profit is just below zero-earnings benchmark (Burgstahler & Dichev, 1997; Degeorge, et al., 1999). Therefore, suspect-firm years are those years that management has an incentive to manage earnings. Instead of examining these specific years, I used the whole sample to perform my regressions.
