Business ethics : what is the relationship between being corporate socially responsible and the level of earnings management?

(1)

UNIVERSITEIT VAN AMSTERDAM

Business Ethics

What is the relationship between being Corporate

Socially Responsible and the level of Earnings

Management?

Christopher Seraus, 10579257

Supervisor: Dr. Silvia Dominguez-Martinez MSC Business Economics

Managerial Economics and Strategy 15th July 2018

15 ECTS

This study examines the relationship between being Corporate Socially Responsible (CSR) and the level of Earnings Management (EM). Although prior literature finds a negative relationship between being CSR and the level of EM, this study shows that there is no negative relationship. This research uses organizations from the S&P 500 in the time frame of 1999-2007. The variables used for calculating EM are accrual based Earnings Management and real activities manipulation. Also the effect of the Sarbanes-Oxley Act is included in order to examine a change in the relationship between CSR and EM before and after the SOX. For the examination of this relationship, the Ordinary Least Squares (OLS) method and the Fixed Effects (FE) method is applied. One of the main findings is that there are no differences in the relationship between the pre-SOX period and the post-SOX period for accrual based EM. In contrast to a negative change in the relationship for real activities manipulation.

(2)

1 Statement of Originality

This document is written by Christopher Seraus, who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are 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)

2

1. Introduction

Corporate Social Responsibility (CSR) becomes more important to organizations because of the increased awareness of society in this subject. The society, customers, employees, and the community expect that organizations focus on CSR. One of the first definitions of CSR is from Bowen (1953), he defines CSR as the obligations of organizations to follow the objectives and values of the society at that moment. In other words, CSR focuses on making decisions on ethical topics, like environment, employee relations, and corporate governance. Financial transparency is something that could be considered as CSR practices as this strengthens the relationship with stakeholders, a group of individuals that is related to the organization. As organizations should report their financial situation in a way that reflects the actual situation of an organization. In the past years, there were several scandals that violated this transparency statement, such as the Enron Scandal, Worldcom, Tyco, Healthsouth, and Satyam. In those scandals, the management of certain earnings played a role. Earnings Management (EM) has a consequence for the stakeholders of an organization, because an organization can mislead or misreport the stakeholders by it. Organizations commit EM for example to cover their financial instability and attract new stakeholders (Healy & Wahlen, 1999; Roychowdhury, 2006). This paper examines whether organizations that focus on CSR commit less EM than organizations, which do not focus on CSR. Therefore the following research question is created:

“What is the relationship between being Corporate Socially Responsible and the level of Earnings Management?”

The already existing literature, like Kim, Park and Wier (2012), Chih, Shen and Kang (2008), Gelb and Strawser (2001) conclude that being Corporate Social Responsible has a negative relationship with the level of EM. This means that organizations that focus on CSR, commit less EM because of both ethical and moral motives.

This paper contributes to the already existing literature as the available academic studies regarding this research topic is limited. Also an exogenous shock will be considered, the Sarbanes-Oxley Act (SOX), which is included to examine a change in the relationship between CSR and EM. No previous literature has been discovered that specifically researched the differences in the relationship between CSR and EM before and after the SOX. As proven by prior literature, the SOX has a significant effect on the level of EM.

(6)

5 Therefore, it is expected that the SOX will have an effect on the relationship between CSR and EM. Extending the empirical model by not only using the Ordinary Least Squares method, but also the Fixed Effects method, also contributes to the already existing literature. This research uses two different proxies for EM in order to formulate an answer on the research question: accrual based EM and real activities manipulation. The results in this paper show that there is no negative significant relationship between being Corporate Socially Responsible and the level of EM, regardless the proxy of EM. This suggest that some organizations that focus on CSR do practice EM and other organizations that focus on CSR do not practice EM. One of the models used, shows a positive relationship between being Corporate Social Responsible and the level of real activities manipulation. This suggests that organizations that focus on CSR, allow themselves to practice EM. We also find that the effect of SOX on the level of EM depends on the model used and varies in the direction of the relationship. Lastly, we found that the effect of CSR on accrual based EM does not change because of the SOX. Furthermore, when using real activities manipulation as proxy for EM, the relationship of CSR on EM is higher in the pre-SOX period compared to the post-SOX period.

The remainder of this paper is as follows. First, we will examine prior literature and develop the hypothesis used in this paper. Second, the methodology will be examined. Third, the descriptive statistics of the data used to test the hypotheses will be provided. Fourth, the results based on the methodology will be discussed and lastly the conclusion will be discussed including the limitations and recommendations for future research.

(7)

6

2. Literature research and hypotheses development

In this section, relevant literature will be discussed and the hypotheses will be derived and explained. Firstly, literature about the agency theory will be provided. Secondly, EM will be discussed and examined. Thirdly, literature regarding CSR will be reviewed. After discussing CSR, the relationship between CSR and EM will be examined and relevant literature will be provided. Lastly, the influence of SOX on EM will be discussed.

2.1 Agency theory

The agency theory is based on the risk sharing of a principal and an agent. Also, as Jensen and Meckling (1976) stated in their paper, an agency relationship is a contract under which a principal delegates authority to the agent. Whereas according to Eisenhardt (1989) agency theory consists out of the following problems. The first one is that the goals and desires of the two parties are not aligned and that the costs of monitoring the agent by the principal is relatively high. For example, a shareholder’s interest is a high return per share but the interest of the CEO is an as high as possible compensation, i.e. salary with as low as possible effort. Another problem from the agency theory is the information asymmetry between the principal and the agent. The principal and the agent have different information about for instance the financial performance. This will result in different behaviors from the two parties. For example, the CEO (agent) has more internal information than the stakeholders (principal). This creates opportunities for the CEO to act in a way that benefits him, but hurts the shareholders. For example, a CEO has incentives to conceal and misstate the annual report in order to increase the reputation of an organization. In this case, a CEO could consider to practice EM.

Another theory applicable to this paper, is the stakeholder theory. Stakeholders are a group of individuals who are in a certain way related to the organization, for example shareholders, suppliers, employees and customers. The concept of the stakeholder theory is that there are so many different stakeholders related to the organization, that each stakeholder has different interests and preferences for the organization, which could conflict (Freeman, 2001). Clarkson (2005) states that the success of an organization is normally focused on the interests of one stakeholder, which is the shareholder. But according to him, an organization is obliged to focus also on the other stakeholders, like the society. And as he stated, “The moment that corporations and their managers define and accept responsibilities and obligations to primary

(8)

7 stakeholders, and recognize their claims and legitimacy, they have entered the domain of moral principles and ethical performance, whether they know it or not.”(Clarkson, 1995, p.112). In other words, the stakeholder theory explains why some organizations focus on being Corporate Social Responsible. Organizations want to take into account the preferences of all the stakeholders, including the society.

2.2 Earnings Management

This part of the literature research will focus on prior literature regarding EM. The definition of EM will be provided. Furthermore, the two different types of EM will be mentioned. And lastly, the motives of organizations for practicing EM will be discussed.

2.2.1 What is Earnings Management

In the often cited paper of Healy and Wahlen (1999), they defined the term EM as follows: "Earnings Management occurs when managers use judgment in financial reporting and in structuring transactions to alter financial reports to either mislead some stakeholders about the underlying economic performance of the company or to influence contractual outcomes that depend on reported accounting numbers." (p. 368).

Managers have the opportunity to practice EM in organizations because of the information asymmetry between stakeholders and the management of the organization. But according Bhaumi and Gregoriou (2010), EM exists because of the “loopholes” in the accounting rules and standards. Highlighting that there are specific rules regarding displaying and calculating the financial results. But that those rules do not and cannot cover all possible circumstances, and therefore the possibility to practice EM is created. EM is a problem because of the transparency of the financial results towards their stakeholders. Therefore an organization that practices EM is considered as less ethical than an organization that do not practice EM, as CEO’s practice EM in own interests and not in the interests of the stakeholders (Prior, Surroca and Tribó, 2008).

2.2.2 Types of Earnings Management

The literature distinguishes different approaches to practice EM. In this part wewill discuss the difference between the two most often cited and explored approaches in literature. Furthermore, the relevant literature for each approach will be discussed.

(9)

8

2.2.2.1 Accrual based Earnings Management

Accrual based Earnings Management is based on the manipulation of accruals. Accruals are the differences between net income and cash flow. The goal of the accruals is to provide a sign of the performance of an organization. To illustrate, an organization sells products to another organization on credit, this increases the accrual of revenue. The cash flow increases because the cash flow consists out of the cash inflow and outflow, but the net income stays equal because the organization sells on credit. The net income does not take the income and outcome of sales or purchases on credit in to account. Organizations can manipulate their earnings by creating new accruals and this is called accrual based EM. According to Zang (2012) accrual based EM is defined as:

"Accrual-based earnings management is achieved by changing the accounting methods or estimates used when presenting a given transaction in the financial statements. For example, changing the depreciation method for fixed assets and the estimate for provision for doubtful accounts can bias reported earnings in a particular direction without changing the underlying transactions." (p. 676)

So accrual based EM is based on bookkeeping. But the real net income of the organization is not affected by accrual based EM. According to Dechow, Sloan and Sweeney (1995), different components of accrual based EM exist. Firstly, the expense manipulation, i.e. delayed recognition of expenses. The income stays equal, but the costs shift towards another period. Secondly, revenue manipulation, i.e. premature recognition of revenue. The revenue shifts towards an earlier period then when it actually occurs. And thirdly, margin manipulation, which is a combination of revenue and expense manipulation. Two examples of accrual based EM are the estimates of bad debt reserves and inventory depreciation methods. If those estimates are manipulated in order to distort the view of the real performance, accrual based EM is practiced.

2.2.2.2 Real activities manipulation

Real activities manipulation is a way to practice EM based on adjustments of the execution of real transaction. According Zang (2012), real activities manipulation is defined as “a purposeful action to alter reported earnings in a particular direction, which is achieved by changing the timing or structuring of an operation, investment, or financing transaction, and

(10)

9 which has suboptimal business consequences.” Roychowdhury (2006) defined real activities manipulation in his paper as:

“Departures from normal operational practices, motivated by managers’ desire to mislead at least some stakeholders into believing certain financial reporting goals have been met in the normal course of operations. These departures do not necessarily contribute to firm value even though they enable managers to meet reporting goals.”(p. 337)

So real activities manipulation is done by adjusting the real operating activities in organizations, while accrual based EM focuses on the choice of accounting methods. For example, an organization reduces the real research and development (R&D) expenses or marketing expenses to reduce the total expenses in the financial report.

Thus, when an organization postpones the purchase of a certain fixed asset, literature defines this as real activities manipulation. When an organization purchases a certain fixed asset on credit, this is defined as accrual based EM.

2.2.3 The motives of practicing Earnings Management

There are different incentives for organizations to practice EM, which will be discussed in the following section

First of all, the market expectations and valuations is one of the incentives. Investors and financial analysts make use of obtainable financial results and this will contribute to the valuations of the stocks. When for example the outcome of implementing a new product or a new machine did not go as predicted, organizations can get the incentive to practice EM in order to increase their stock value (Healy and Wahlen, 1999). On this subject Burgstahler and Dichev (1997) mentioned the transaction cost theory, which entails that transactions done by stakeholders are affected by the information about earnings. Therefore higher earnings result in more transactions.

Gunny (2010) examined the relationship between using real activities manipulation and performance. She claimed that real activities manipulation is positively associated with organizations benchmarking on earnings, so organizations practice real activities manipulation to increase their performance. Therefore, she argued that real activities manipulation is consistent for a management that wants to increase the performance. She

(11)

10 stated that because of meeting the financial benchmarks by practicing real activities manipulation, organizations have better performance in the future. As Burgstahler and Dichev (1997) stated that the reputation of the firm will increase, which also results in better future performance. In addition to this, the contract with lenders, suppliers and employees could be affected if the financial results were not as expected. This will impact the trust of those parties on the organization, i.e. a declining reputation. For example, when the earnings are below a certain threshold, the organizations cannot pay their debt or pay suppliers. This signals instability which results in less trust of third parties in committing business with them (Healy & Wahlen, 1999; Roychowdhury, 2006). In other words, an organization wants to practice EM, in order to signal stability which increases their reputation and therefore the performance. This entails that EM is not always a negative practice towards the stakeholders in the long run. Roychowdhury (2006) stated in his paper that institutional ownership is also a motive to practice EM. Where disappointing earnings could incentivize shareholders to sell shares of the organizations.

Another perspective for the incentive to practice EM is that the compensation schemes of executives are based on the financial results of the organizations. This gives an incentive towards managers to practice EM to obtain a higher compensation (Iatridis & Kadorinis, 2009; Bhaumi & Gregoriou, 2010). For example, a CEO who is in his last year in an organization, might practice EM in order to get a high bonus when his bonus is linked to earnings.

In the paper of Graham et al. (2005) 400 CFOs were surveyed to determine the usage of real activities manipulation. The first result of this paper, was that CFOs conduct EM because their focus lays on earnings rather than cash flows. CFOs are convinced that this will increase their creditability and stock price. Another result is that if the CFOs conduct EM, most of the time this entails real activities manipulation. This surprises Graham et al, because accrual based EM is cheaper to practice than real activities manipulation. But they also specify that this is possibly due to the implementation of the SOX, because real activities manipulation is harder to detect.

But practicing EM has also costs for an organization. As Roychowdhury (2006) mentions that real activities manipulation might reduce the value of the organization as some actions to increase or decrease the earnings could decline the cash flows in the future period. And as Prior et al. (2008) mentioned that the biggest consequence of EM is that an organization

(12)

11 might lose the support of the stakeholders which might result in activism or alertness from the stakeholders. This will harm the reputation of the organization.

2.3 Corporate Social Responsibility

This part will focus on the definition of CSR and the motives of an organization to focus on CSR, supported by prior literature.

2.3.1 What is Corporate Social Responsibility

CSR has a long history and it is complex to find a well formulated definition of CSR. Bowen (1953) was one of the first authors who provided a definition. He defined CSR as “It refers to the obligations of businessmen to pursue those policies, to make those decisions, or to follow those lines of action which are desirable in terms of the objectives and values of our society” (p. 6). Moir (2000) addressed in his article the social responsibilities for businesses regarding CSR. According to Moir, examples of the issues that CSR covers are: human rights, environment, community relations, employee relations, and corporate ethics. The World Business Council for Sustainable Development defines CSR as “CSR is the continuing commitment by business to behave ethically and contribute to economic development while improving the quality of life of the workforce and their families as well as of the local community and society at large” (as cited in Moir, 2000, p. 17).

Dahlsrud (2006) concludes in his article that there are different definitions of CSR. The challenge for organizations is how they define their CSR practices and how they interpret the definitions of CSR. Because of the existence of different definitions of CSR, there is no universal explanation how organizations should implement the concept of CSR. This means that besides the definition also the implementation is subjective and should be guided to face the challenges and increase the outcome of focussing on CSR.

As discussed earlier, Clarkson (1995) mentioned in his article that the stakeholder theory is an important feature of the focus on CSR. Besides the organizations’ purpose to create wealth, also valuing all stakeholders equally is even important. The society is also one of the stakeholders and therefore focusing on CSR should be valued as high as other stakeholders.

(13)

12

2.3.2 Motives for focusing on Corporate Social Responsibility

There are several sources that discuss the benefits and motives of focusing on CSR. The motives of prior literature of focusing on CSR are:

 CSR has a positive effect on the public image of the organization and reputation. This will result in customer engagement and can help to engage new customers. Organizations that focus on CSR are perceived philanthropic, which is beneficial for the public image. All of this, will increase revenue from sales and market share (Weber, 2008; Grupta & Sharma, 2008; Epstein-Reeves, 2012; Double the Donation, n.d.). And according Gray and Balmer (1998) public image and reputation has a positive effect on the competitiveness of an organization.

 CSR has a positive effect on employee engagement, such as motivation, retention and recruitment. This results in a positive public image and reputation. Also, employees might be more motivated working for an organization that commits to CSR (Weber, 2008; Epstein-Reeves, 2012; Double the Donation, n.d.). On the other hand, it could be that certain employees search for organizations that focus on CSR. This could result in a selection bias, but might result in motivated employees as they might feel connected to the organizations strategy.

 CSR could result in competitive advantages as it could save costs for the organization. If taking the environmental aspect of CSR into account, cutting on packaging or using less energy could result in less costs (Weber, 2008; Epstein-Reeves, 2012). However, it could also be the case that costs increase, because environmental products can often be more expensive. So this argument can be a motive to focus on CSR, but can also be a motive for not focusing on CSR.

 CSR stimulates long-term thinking, the focus shifts from direct impact towards results in the future. As reputation and increasing environmental awareness, are not results that will be immediately visible, but they will be over time. CSR also stimulates innovative thinking and creative implementation of CSR practices (Epstein-Reeves, 2012; Double the Donation, n.d.)

(14)

13

2.4. Corporate Social Responsibility versus Earnings Management

The introduction explained the importance and relevance of examining the effect of EM on CSR. This part will provide relevant literature and provide prior research that examined this effect. The existing literature of the effect of EM on CSR is modest, but there are still several papers that examine this effect.

Kim, Park and Wier (2012) examined two hypothesis, the transparent financial reporting hypothesis and the opportunistic financial reporting hypothesis. The transparent financial reporting hypothesis entails that the relationship between CRS and EM is mainly focused on business ethics and doing the right thing. This suggest that an organization that is socially responsible, is ethical and therefore transparent towards the stakeholder. This indicates a negative relationship between CSR and EM. Another argument for the transparent financial reporting hypothesis is that CSR provides a signal regarding the reputation of the organization. In other words, focusing on CSR is based on building a reputation and practicing EM might decline the reputation of an organization. This means that organizations that care about their reputation, focus on CSR, resulting in a negative relationship between CSR and EM. The opportunistic financial reporting hypothesis entails that being ethical and socially responsible, is only used to cover or balance other unethical things, and that a manager focuses on CSR only for self-benefits. This indicates that organizations allow themselves to practice EM, because they are focusing on CSR. In other words, organizations compensate their EM level with their CSR practices. In the paper mentioned above, real manipulation and accrual based EM are used as a proxy of EM. They find that CSR organizations are less likely to practice EM through both proxies of EM, as support for the transparent financial hypotheses. Their explanation for this finding is thus that managers who focus on CSR, are incentivized to operate fair, honest and ethical. And this results in less EM, as this is perceived as more ethical.

Chih, Shen and Kang (2008) used different measures of EM: earnings smoothing, earnings aggressiveness, and loss avoidance. They found that the effect of CSR on EM depends on the type of EM. When earnings smoothing, which is an accounting technique to level fluctuations of earnings in different periods so that the earnings stay the same, and loss avoidance, which occurs when organizations manage their earnings to avoid losses in earnings when the earnings are lower, are used as proxy, CSR has a negative effect on EM. Their argument is the same as the transparent financial reporting hypothesis of Kim, Park and Wier (2012).

(15)

14 Gelb and Strawser (2001) used financial disclosure quality as proxy for EM. Financial disclosure quality is used to solve the adverse selection problem, where different parties have different information. In this paper the proxy for the financial disclosure quality, i.e. Earnings Management, is taken from an organization that provides reports of financial disclosure of different organizations. One of their findings is the positive relationship between CSR and the financial disclosure quality, i.e. the relationship between CSR and EM is negative. This supports their argument that financial disclosure is a form of ethical behavior that fits in the CSR practices.

In the literature there are several authors who focus on the effect of EM on being socially responsible. For example Choi, Lee and Park (2013) examined the effect of how a certain corporate governance affects a manager’s intention to promote CSR. In other words, the effect of EM on CSR and the reason behind this effect. They found that organizations which practice EM focuses less on being social responsible. Their argument for this is the opportunistic financial reporting hypothesis as mentioned in the article of Kim, Park and Wier (2012). Prior et al. (2008) explore the hypothesis that managers practice EM to benefit as an organization and use CSR as a compensation towards stakeholders. Indeed, they find a positive impact of EM on CSR, confirming their hypothesis. Leading to the same conclusion as the article of Choi, Lee and Park (2013).

As discussed, there are discrepancies about the expected direction of causality from EM to CSR. The direction is ambiguous and no clear argument for the expected directions is given in the literature. Therefore in this paper, the assumption is made that CSR has a relationship with the level of EM. So this paper will follow the papers of Kim et al. (2012), Chih et al. (2008) and Gelb and Strawser (2001) to answer the research question. As argued in Kim et al. (2012), Choi et al. (2013) and Gelb and Strawser (2001), the relationship between CSR and EM is mainly focused on business ethics and doing the right thing. This is labeled as the transparent financial reporting hypotheses, which suggests that an organization that is socially responsible, behaves ethical and therefore transparent towards the stakeholder, resulting in a negative relationship between CSR and EM. So the following hypothesis is created:

Hypothesis 1: The more socially responsible an organization is, the less likely that it commits Earnings Management.

(16)

15 But as discussed, prior literature also focuses on the opportunistic financial reporting hypothesis. Therefore if we find a contrary effect of CSR on EM, this would be in line with the opportunistic financial reporting hypothesis. This means that organizations allow themselves to practice EM, because they are already focusing on CSR.

2.5 Sarbanes-Oxley Act versus Earnings Management

According to the literature the SOX act has a significant effect on the level of EM. The SOX is funded in 2002 after several accounting scandals, like Enron and WorldCom. It is funded by Paul Sarbanes and Michael Oxley. The SOX applies towards listed organizations in the United States. The SOX consists of 69 articles that are introduced to prevent accounting scandals as Enron and Worldcom. The most important acts are sections 302 and 404. Section 302 obliges the management to state an opinion about the fairness, completeness and accuracy of the financial statements every quarter. This should increase the transparency between the management and stakeholders. Section 404 describes the responsibility of the management to report the financial results and enforce the procedures and regulations. If the management misstates the financial reports, this will result in sanctions in terms of fines or prison sentence.

Researchers have examined the effect of the SOX on the use of EM. As Cohen et al. (2008) examined in their paper. They found that after the implementation of the SOX, the accrual based Earnings Management decreased significantly. This in contrast to real earnings manipulation, which decreased before the implementation and increased significantly after the implementation of the SOX. They concluded that after the regulation of the SOX, organizations use real earnings manipulation as substitute of accrual based Earnings Management. Real activities manipulation is more costly to implement, because this form of EM implies changes in operational activities, whereas accrual based EM is solely based on the style of bookkeeping and not on cash flows. But at the same time, because real activities are based on operational activities, it is harder to detect for outsiders. Therefore, organizations consider to shift towards real activities manipulation instead of accrual based EM after the SOX. The trade-off between the two most common forms of EM, is thus based on cost and visibility.

Lobo and Zhou (2010) also conclude in their paper that accrual based EM increases prior to the SOX and declines after the implementation of the SOX. They used a sample of Canadian organizations that are also listed on the US exchange market. Thus, they are subjected to the

(17)

16 rules of the SOX. As control sample they used Canadian organizations that are not listed on the US exchange market. They find as expected that the level of accrual based EM does not differ before and after the SOX for the control sample. However, it does differ for the Canadian organizations that are listed on the US exchange market. A limitation of this paper is that it is likely that the control group differs from the treatment group and therefore the comparison cannot be made. However, besides the limitation, the paper is in line with Cohen et al. (2008).

Bartov and Cohen (2009) tested two hypothesis regarding the level of real earnings management in the post-SOX period. Firstly, the substitution effect of Cohen et al. (2008), managers compensate the decline of accrual based EM with real activities manipulation, because real activities manipulation it is less visible. Secondly, managers choose more often real-activities manipulation, even in the pre-SOX period. Therefore there would be no increase in the level of real activities manipulation in the post-SOX period. Bartov and Cohen applied an empirical model in order to test those hypothesis. They found that in the post-SOX period, an increase of real activities manipulation and a decrease of accrual based EM occurs. But in contrast, Ghosh et al. (2010) found no evidence that EM declined in the post-SOX period. They found a decline of the discretionary accruals after the SOX, but this effect appears not to be significant. It is unclear what could be the reason of this insignificant effect. The biggest difference between the paper of Ghosh and the other papers discussed, is the way of calculating the accrual based EM. Instead of using the modified Jones model, the paper used the performance-adjusted Jones model. The performance-adjusted Jones model uses sales in the calculation of the accrual based EM, where the modified Jones model uses the revenue.

In conclusion, most literature argues that EM is less in the post-SOX period than in the pre-SOX period. As the rules provide incentive for managers to commit less EM. But literature makes a distinction between the effects on accrual based EM and real activities manipulation. As this decline of the level of EM is only visible when looking at the accrual based EM (Cohen et al, 2008; Lobo & Zhou, 2010; Bartov & Cohen, 2009). When looking at the real activities, the level increases in the post-SOX period (Cohen et al, 2008; Bartov & Cohen, 2009). Thus the second hypothesis created in this paper is:

Hypothesis 2a: The level of accrual based Earnings Management in the post-SOX period is lower compared to the pre-SOX period for an organization.

(18)

17 Hypothesis 2b: The level of real activities manipulation in the post-SOX period is

higher compared to the pre-SOX period for an organization.

In order to combine hypothesis 1 and hypothesis 2, the effect of CSR on EM should be examined in the pre-SOX period and the post-SOX period. The assumption when looking at those effects is that the hypothesis discussed earlier are confirmed. We still expect a negative relationship between CSR and EM. But accrual based EM is in the post-SOX period lower compared to the pre-SOX period. Therefore, we expect that the effect of CSR on accrual based EM is lower in the post-SOX than in the pre-SOX period. In contrast to real activities manipulation, where we expect that the effect between CSR and EM is higher in the post-SOX period. Therefore the following hypothesis is created:

Hypothesis 3a: The relationship between CSR and accrual based Earnings Management is lower for the post-SOX period than the pre-SOX period.

Hypothesis 3b: The relationship between CSR and real activities manipulation is higher for the post-SOX period than the pre-SOX period.

(19)

18

3. Methodology

This part describes the research methodology applied in this research. First, the sample design will be discussed. Second, the measurements of the dependent variables will be reviewed. Thirdly, the independent variable, CSR, will be reviewed. Fourthly, the control variables used in the empirical model will be given. Finally, the empirical model will be provided.

3.1 Sample design

This research uses the organizations of the S&P 500, a stock market index, which consists out of the 500 biggest firms in the United States of America. The United States of America is used because of the availability of the data and the possibility of comparison with prior research. The data used in this paper is extracted from the database Wharton Research Data Services (WRDS). For the organization’s financial information the database Compustat is used. In order to obtain data for calculating CSR, we used the database MSCI, formerly called KLD or GMI. This database gives us the strengths and concerns for all categories of CSR. The years 1999 till 2007 are used in this research, in order to eliminate any effects of the financial crisis. This dataset contains four years before the SOX and four years after the SOX.

After collecting the data of Compustat and MSCI, the two databases were combined. When removing all missing matches and observations we end up with a database of 2,155 observations. In the end, the data sample consists out of 318 different companies of the S&P 500. These observations will be used in order to test the hypotheses and formulate an answer on the research question.

3.2 Dependent variables

As explained in the literature review, there are two different approaches to engage in EM practices, accrual based Earnings Management and real activities manipulation. This part will elaborate on the measurements of the two different variables.

3.2.1 Accrual based Earnings Management

Dechow et al. (1995) showed that the Modified Jones Model is the best way to measure the accrual based Earnings Management. Previous literature that examined accrual based Earnings Management also made use of the modified Jones, model such as Kim, Park and Wier (2012), Klein (2002) and Bergstresser and Philippon (2006). The formula calculates the

(20)

19 discretionary accruals from an organization, which is the difference between normal accruals and the real total accruals. Jones uses the discretionary accruals as a proxy for EM. In this paper the use of the absolute value of the discretionary accruals will be used, as EM can imply accrual-increasing or accrual-decreasing. So according to Dechow et al. (1995) the modified Jones can be calculated by the following three steps:

Step 1: Calculate the total accruals for an organization by the following formula: _t C _t C _t Cash_t Dep_t

Where

= Total accruals in year t

= Change in current assets in year t = Change in current liabilities in year t

= Change in cash and cash equivalents in year t = Depreciation and amortization expenses in year t

= Change in short term debt included in current liabilities in year t

Step 2: Estimate the modified Jones model with the following calculation:

Where

= Change in revenues in year t

= Change in accounts receivables in year t

= Property, plant and equipment in year t

= Total assets in year t – 1

Step 3: Calculate the discretionary accruals by predicting , i.e. the residuals. In this data sample the absolute value of the discretionary accruals is used as proxy for accrual based EM.

3.2.2 Real activities manipulation

To measure the real activities manipulation of an organization, the paper of Roychowdhury (2006) is studied. In his paper he stated that he examined three different methods to create a quantitative variable for real activities manipulation: abnormal levels of discretionary

(21)

20 spending, abnormal levels of production expenses and abnormal cash flow from operating. Kim, Park and Wier (2012) also applied these methods to estimate real activities manipulation. In case of the three methods used by Roychowdhury (2006) all estimations are used for abnormal discrepancies. Therefore the interest lays with the residuals of the estimations below.

CFO

The first method to estimate the real activities manipulation is the CFO method. The CFO method is based on sales manipulation, as a manager can increase sales by offering price discounts or give credits. This will result in an abnormal amount of CFO. This will be examined with the following regression:

Where

= Cash flow from operations in year t = Total sales in year t

= Total sales in year t-1

Prod

This method is based on the sum of the cost of goods sold and inventory. This manipulation method is based on overproduction or to report lower cost of goods sold. In this case organizations manage earnings upwards, organizations can for example produce more goods to reach the demand of stakeholders. This will result in lower fixed costs per unit, which increases the reputation of the organization (Roychowdhury, 2006). The following estimation is used: Where

= Sum of the cost of goods sold in year t and the change in inventory in year t

DIX

This method of detecting real activities manipulation is based on the discretionary expenses, which is defined as the sum of advertising expenses, R&D expenses and selling, general and

(22)

21 administrative expenses. Organizations can reduce the reported expenses in the financial results by for example postponing expenses and this results in an increase in the earnings (Roychowdhury, 2006). The following estimation is used as proxy:

Where = Discretionary expenses: sum of advertising expenses, R&D expenses and selling, general and administrative expenses in year t.

To create a comprehensive variable as proxy for real activities manipulation, we follow the research of Kim, Park and Wier (2012) where they calculated the combined RAM. Because of the expected direction of Kim, Park and Wier they conducted the combined real activities manipulation as . Table 1 provides a summary of which measurements and methods were used to estimate accrual based Earnings Management or real activities method.

Measurement for EM: Model: Estimation of :

Accrual based Earnings Management

Modified Jones model (Dechow et al., 1995)

Discretionary accruals

Real activities manipulation

CFO Method (Roychowdhury, 2006)

Abnormal cash flow from operations

Prod Method (Roychowdhury, 2006)

Abnormal production costs

DIX Method (Roychowdhury, 2006)

Discretionary expenses

CFO-PROD+DISEXP (Kim, Park and Wier, 2012)

Combined real activities manipulation

(23)

22

3.3 Corporate Social Responsible

In order to create a variable that measures the CSR, the database MSCI (former KLD and GMI) is used. This database provides ratings for several categories. For each category they give a rating for the concerns and for the strengths. The categories used in this paper are: Community, Corporate Governance, Diversity, Environment, Employee Relations and Human Rights. Appendix A provides an extensive list with all the concerns and strengths per category. From this database, a proxy for a rating of CSR is created. The total sum of the ratings of the strengths of the organization minus the total sum of the ratings of the concerns implies the score on CSR. This indicates that there could also be a negative CSR score, which implies that more concerns than strengths exist.

3.4 Control variables

In order to perform a statistical test as precise as possible, the inclusion of several variables is crucial. The correlation between the two main variables, CSR and EM, will be controlled by those included variables. Therefore, the correlation in the empirical model will be more precise than without control variables. If the inclusion of the control variables are neglected, the problem of omitted variable bias occurs. This will affect the effect between the variable of interest. According to Stock and Watson (2015) the control variables are included to control for omitted factors that have a causal relationship with the dependent variable and are correlated with the independent variables. The control variables used in this research, are based on prior literature and will be elaborated in this paragraph.

Roychowdhury (2006) argues that the size of an organization is correlated with EM. Therefore, the first control variable is the size measured in the total assets. This variable is also included by Choi et al. (2013), Kim et al. (2012) and Chih et al. (2008). Kim et al. (2012) also argue that the level of EM in an organization, depends on the organizations that controls the audit of that organization. Therefore the control variable BIG4 is included, which is an indicator if the organization’s audit is done by a BIG 4 auditor. Choi et al. (2013) and Kim et al. (2012) use the variable leverage, as long term debt, to control for the level of financial distress. And in order to control for the incentive of focusing on CSR and practicing EM, we include the total amount of shares outstanding. An organization that has more shares outstanding could be more incentivized to focus on EM, in order to keep the shareholders satisfied regarding the financial results. Barnea and Rubin (2010) found that the amount of shares outstanding and CSR score are not correlated with each other. The control variable

(24)

23 often used in prior literature is the Return On Assets (ROA), a measure for the performance of an organization. The ROA is calculated as net income divided by total assets. Choi et al. (2013) and Kim et al. (2012) also include this variable as a control variable. In our empirical model, the ROA will not be included. Researchers claim that the level of EM has an effect on the ROA and not vice versa. Therefore including the ROA in this empirical model, will not be valid.

3.5 Empirical model

In the prior parts, the different variables used in this research were described and also the proxy and the estimations. In this paper a panel data regression will be used in order to test the hypotheses. A panel data regression is important to test the hypotheses. Only examining the correlations between variables, do not give a valid answer on the hypotheses. Therefore we have to perform a regression, with controlling for several variables. Because of the inclusion of the control variables, the omitted variable bias will be reduced. In this paper the following model will be estimated:

Model 1:

Where:

= Proxy for EM as elaborated in part 3.2

= Proxy for CSR as elaborated in part 3.3 = Binary variable for post-SOX period = Natural logarithm of total assets

= Dummy variable, whether audit is done by a BIG 4 auditor. = Natural logarithm of total long term debt

= Natural logarithm of total shares outstanding

= Error term

(25)

24

4. Descriptive statistics

This chapter will provide a summary of the data of the variables are given. In addition, the correlation between the variables of interest are given and discussed.

4.1 Summary

In order to prevent the effect of outliers in the statistical part, all variables are winsorized at 1th and 99th percentile. In table 2 the descriptive statistics of the EM variables are given. Table 2 shows that all organizations in the sample used, use a form of EM. This can be seen in the column minimum, where the level of EM is not zero regardless of the proxy of EM. If we compare the statistics in this paper with those of Kim, Park and Wier (2012), it can be concluded that the mean of the absolute value of discretionary accruals in this paper are significantly lower. A possible explanation is because of the sample and industry differences. In contrast to Kim, Park and Wier, the mean of the combined RAM is in our paper negative, where in their paper this is positive. A difference between that paper and this paper, is the time period they consider. In their paper they use the observations from 1991-2009. This time period has a part of the financial crisis in their sample, what could explain the difference in mean. As organizations could practice real activities manipulation in the financial crisis, to manipulate their earnings.

Variable Mean Standard Deviation Minimum Maximum

Abs_DACC 0.05612 0.06470 0.00098 0.33268

RAM -0.00857 0.18427 -0.43680 0.40302

DCFO -0.00433 0.07110 -0.16644 0.17735

DPROD -0.00518 0.19232 -0.44617 0.45527

DISEXP -0.00331 0.08265 -0.21915 0.21326

Where Abs_DACC is the proxy for accrual based earnings management, RAM the combined proxy for real activities manipulation, DCFO the proxy for the cash flow part of real activities manipulation, DPROD the proxy for the production part of real activities manipulation, DISEXP the proxy for the expense part of real activities manipulation.

Table 2 Descriptive statistics of Earnings Management variables

In table 3, the descriptive statistics of CSR variables are given. The table shows that the mean of each CSR variable is around zero. This means that there are approximately as much

(26)

25 organizations with a negative score and a positive score. If we compare the total CSR score with the score of Kim, Park and Wier (2012), it can be concluded that the scores are similar. The table also shows that organizations have different CSR scores, as for example the total score where the organization with the lowest score has -7 and the organization with the highest score have 8. This means that they differ fifteen points in CSR score.

CSR-Environment -0.29513 1.11007 -4 2 CSR-Community 0.19675 0.79067 -1 3 CSR-Human Rights -0.16381 0.37018 -1 0 CSR-Employee Relations 0.03527 0.98910 -2 2 CSR-Diversity 0.93086 1.45765 -1 5 CSR-Product -0.41903 0.86095 -3 1 CSR-Corporate Governance -0.54942 0.70332 -2 1 CSR-Total -0.27192 2.9621 -7 8

Table 3 Descriptive statistics of CSR variables

Table 4 discusses the control variables used in this paper. As shown in the table the control variable Big4 is skewed towards 1, because the mean is 0.93271. This means that almost every organization has an organization of the Big 4 doing their audit. This is as expected, as we use the S&P 500 as sample and this contains big organizations of the United States. This increase the probability that they also work with an organization of the Big 4. Compared to the research of Kim, Park and Wier (2012) the mean does not differ significantly. The other control variables are not comparable, as in this paper the variables are transformed to log variables in order to avoid large outcomes in the regression analysis.

Log (Size) 8.87584 1.27304 6.17286 11.60078

Big4 0.93271 0.25057 0 1

Log (Long term debt) 7.06609 1.74645 1.43509 9.87493

Log (Shares outstanding) 5.59172 1.16340 3.48278 8.43059

(27)

26

4.2 Correlation

Table 5 gives the Pearson correlation test of all variables, the bold coefficients are significant at a level of at least ten percent. The table shows that the variable discretionary accruals and combined real activities manipulation are significant positively correlated with the total CSR. This indicates that there is a positive relationship between CSR and EM. This is not in line with hypothesis 1 of this paper. We also see that the three components of the real activities manipulation variable are positively significantly correlated with each other, but the direction of this correlation varies. Regarding hypothesis 2, we do not see a significant correlation between SOX and any EM variables. This is not in line with what we expected. Size of an organization, in terms of total assets, are negatively correlated with EM. An argument for this correlation could be that large organizations, do not want to sacrifice their reputation in order to commit EM. The table also shows that the CSR score is positively correlated with the shares outstanding, which could be because of the incentive on an organization to focus on CSR in order to keep the shareholders satisfied. Leverage has a negative correlation with all EM and total CSR score. Which indicates that when leverage is high, an organization will manage their earnings less and focus less on CSR. In terms of Earnings Management, the explanation could be that an organization that commits EM choose to decrease their debt. Or that EM practices deter stakeholders from lending money to an organization. In terms of CSR, it could be the case that if the leverage is high that for example the costs of CSR practices outweigh the benefits. Therefore organizations will focus less on CSR. An important result in the table is the correlation of assets and shares outstanding and the correlation of assets and leverage. In other to avoid the multicollinearity problem, we eliminate the assets in the empirical model.

(28)

1 2 3 4 5 6 7 8 9 10 11 1.abs_DACC 1.000 2.RAM 0.105*** 1.000 3.DCFO 0.087*** 0.637*** 1.000 4.DPROD -0.103*** -0.962*** -0.590*** 1.000 5.DISEXP -0.050** -0.591*** -0.800*** 0.699*** 1.000 6.TotalCSR 0.090*** 0.344*** 0.231*** -0.329*** -0.206*** 1.000 7.Log_assets -0.239*** -0.117*** -0.017 0.117*** 0.006 -0.030 1.000 8.Log_sho -0.015 0.251*** 0.269*** -0.241*** -0.248*** 0.156*** 0.783*** 1.000 9.Log_lev -0.262*** -0.241*** -0.179*** 0.226*** 0.120*** -0.137*** 0.769*** 0.482*** 1.000 10.SOX 0.020 -0.036 -0.032 0.028 0.028 -0.082*** -0.011 -0.043* -0.062** 1.000 11.BIG4 0.049** 0.045* 0.028 -0.043* -0.017 0.013 0.021 0.020 -0.011 0.339*** 1.000

*significance at 10%, **significance at 5%, ***significance at 1% Table 5 Correlation

(29)

28

5. Results

This part discusses the results of the hypotheses drawn in this paper. Firstly, the empirical model used in this part will be validated. Secondly, the results on the three hypothesis created in chapter 2 will be discussed.

5 .1 Validation of empirical model

In order to analysis the empirical model, the model has to be validated. . There will be three different regression methods examined in order to validate the empirical model. The regression methods used are: Ordinary Least Squares (OLS), Fixed Effects (FE), and Random Effects (RE).

OLS is the most basic method for performing a regression. This method ignores the firm fixed effects and year effects. The assumption of this model is that all observations are independent and identically distributed. The FE model is a regression method for panel data, wherein they do not assume that the observations are independent and identically distributed. This means that this model expects that there is a correlation between time-invariant characteristics of an organization. RE method is almost the same as FE method, only in the RE method the assumption is made that the time-invariant characteristics are random.

As shown in appendix B and C, we validate for the use of the FE method and but not the use of the RE method. So when testing the hypotheses, we will examine the OLS method and the FE method. As the OLS is the basic method and FE is an extension of the OLS . When estimating the model with accrual based EM as proxy for EM, we do not include year fixed effects. For the model with real activities manipulation as proxy, we do include year fixed effects.

(30)

29

5.2 Testing hypothesis 1

In this part, we will present the results of hypothesis 1 according the empirical method discussed earlier. Table 6 shows the output in order to formulate an answer on hypothesis 1. As discussed earlier hypothesis 1 is: The more socially responsible an organization is, the less likely that it commits Earnings Management. This indicates that the coefficient of the independent variable, CSR score, is negative. The table shows three different empirical models per EM proxy. Number (1) and (4) are simple OLS regressions, where regression (2), (3), (5), and (6) are panel regression with fixed effects. The coefficients in the table that are bold, are significantly different from zero. The table shows different observations per model, this is due to use of lagged values and changes in variables for creating the accrual based EM and real activities manipulation. Also missing observations in the control variables, results in different observations per model.

Dependent Variable Independent variables ABS_DACC RAM (1) (2) (3) (4) (5) (6) Total CSR 0.0004 0.0001 -0.0002 0.0129*** -0.0003 -0.0017 (0.0004) (0.0009) (0.0009) (0.0012) (0.0012) (0.0012) Log (Leverage) -0.0114*** -0.0011 -0.0462*** -0.0263*** (0.0012) (0.0027) (0.0034) (0.0039) Log ( Shares outstanding) 0.0056*** (0.0017) -0.0067 (0.0056) 0.0668*** (0.0040) 0.0206** (0.0105) BIG4 0.0071 0.0054 0.0416** 0.0229** (0.0050) (0.0051) (0.0175) (0.0109)

Firm Fixed Effects No Yes Yes No Yes Yes

Year Fixed Effects No No No No Yes Yes

R2 0.0862 0.4979 0.5104 0.2836 0.8899 0.9038

N 1859 1949 1859 1614 1688 1614

*significance at 10%, **significance at 5%, ***significance at 1%

Between parentheses are the robust standard errors. A constant is included in the regression model. The data sample consists out of 318 organizations.

(31)

30 If we consider accrual based EM as the proxy of EM, than table 6 shows indeed a negative coefficient of the total CSR score on the accrual based EM in model (2) and (3). This means that the direction are in line with hypothesis 1, but they are not significantly different from zero. Model (1) has a positive coefficient of the total score, which is not in line with hypothesis. However, the coefficient is also not significantly different from zero. Therefore, we cannot confirm hypothesis 1 based on accrual based EM. If we look at significant effects based on accrual based EM, the table shows that the shares outstanding have a positive effect on the level of accrual based EM. This implies that organizations with more shares outstanding, i.e. more external investors, perform more accrual based EM in order to keep them satisfied. Furthermore, leverage has a significant negative effect on the level of EM, which indicates that the more leverage ( i.e. long term debt) an organization has, the less EM it practices.

If we consider real activities manipulation as proxy for EM, we see different results. For the variable of interest, the total CSR score, we see that the effect differs between model (4) and model (5)-(6). When performing only a simple OLS, without fixed effects, the total CSR score has a positive effect on EM. This is not in line with the hypothesis, but when looking at the fixed effects model, we see a negative relationship between CSR and EM. Yet, the negative relationship in model (5) and (6) are not significantly different from zero compared to the positive relationship in model (4) which is significantly different from zero.

So based on table 6 and taking accrual based EM as proxy, we cannot confirm the hypothesis. The coefficient of interest in the models (1)-(3) are not significantly from zero. Therefore, we cannot conclude that the more socially responsible an organization is, the less likely that it commits Earnings Management. When taking real activities manipulation as proxy for EM, we still cannot confirm the hypothesis as model (4) gives an opposite relationship and model (5) and (6) are not significantly different from zero. A difference between model (4) and model (5) and (6), can be explained by the different regression models. When including firm fixed effects, the hypothesis is tested within a firm. As OLS will test the hypothesis between firms. It could be that there is not enough variation within a firm in the CSR score and level of EM and this will result in different outcomes compared to OLS. In table 11 in appendix D, the regressions of table 6 are performed without winsorzing the variables. The table shows that there are no differences in the coefficients of interests regarding the significance.

(32)

31 The direction of the significant control variables of model (4) and (6) are the same. Such as the model of accrual based EM, leverage and the shares outstanding have an effect on EM. The dummy variable BIG4 has a positive effect on the level of real activities manipulation, which means that if the audit is done by an audit firm of the Big 4 the level of real activities manipulation is higher.

As an addition to table 6, model (1) and (4) are estimated with the seven different categories of CSR. In this way, we can test the hypothesis based on separate categories. The results of the model with accrual based EM as proxy are shown in table 7. The table shows different relationships regarding the effect of CSR on accrual based EM. The categories: Community, Diversity, Product, and Corporate Governance have a negative relationship with accrual based EM, as hypothesized, but only corporate governance is significantly different from zero. Environment, Human Rights, and Employee Relations have a positive relationship with accrual based EM, which contradicts the hypothesis.

Dependent variable Independent variable Abs_DACC (1) (2) (3) (4) (5) (6) (7) Environment Score 0.0015* (0.0009) Community Score -0.0012 (0.0014) Human Rights Score 0.0085*** (0.0028) Employee Relations Score 0.0025** (0.0012) Diversity Score -0.0001 (0.0010) Product Score -0.0001 (0.0014) Corporate Governance Score -0.0039** (0.0018)

Control Variables Yes Yes Yes Yes Yes Yes Yes

Firm Fixed Effects No No No No No No No

Year Fixed Effects No No No No No No No

R2 0.0866 0.0860 0.0885 0.0875 0.0858 0.0858 0.0878

N 1859 1859 1859 1859 1859 1859 1859

(33)

32 Table 8 shows the same output as table 7 but with real activities manipulation as proxy. We see that only Human Rights has a negative relationship with real activities manipulation, but not significantly different from zero. The other categories have a positive relationship with real activities manipulation, where only employee relations is not significantly different from zero. In other words, the coefficients that are significantly different from zero in this table, contradict the hypothesis.

Dependent variable Independent variable RAM (1) (2) (3) (4) (5) (6) (7) Environment Score 0.0260*** (0.0029) Community Score 0.0431*** (0.0049) Human Rights Score -0.0019 (0.0100) Employee relations Score 0.0026 (0.0040) Diversity Score 0.0201*** (0.0030) Product Score 0.0107** (0.0048) Corporate Governance Score 0.0219 *** (0.0056)

Control Variables Yes Yes Yes Yes Yes Yes Yes

Firm Fixed Effects No No No No No No No

Year Fixed Effects Yes Yes Yes Yes Yes Yes Yes

R2 0.2657 0.2743 0.2403 0.2405 0.2594 0.2427 0.2471

N 1614 1614 1614 1614 1614 1614 1614

(34)

33

5.3 Testing hypothesis 2

In this part, the results regarding hypothesis 2 will be discussed. As mentioned, hypothesis 2a entails: The level of accrual based Earnings Management in the post-SOX period is lower compared to the pre-SOX period for an organization. And hypothesis 2b: The level of real activities manipulation in the post-SOX period is higher compared to the pre-SOX period for an organization. Table 9 shows the results from the empirical model. The variable SOX is created in order to test these hypotheses. SOX is a binary variable, what equals 1 if the SOX is implemented and 0 if it is not implemented (i.e. the variable is zero if it is before 2003 and one if it is after 2003). Because of the implementation of the SOX variable, we do not include year fixed effects to prevent multicollinearity. The table shows that the SOX has a negative effect on the level of accrual based EM. The SOX has in model (3), a negative effect on the level of accrual based EM which is significantly different from zero. Therefore, we can confirm hypothesis 2a based on model (3), but we cannot confirm the hypothesis on models (1) and (2). When looking at the real activities manipulation as proxy for EM, the table shows that the SOX has a negative effect on the level of real activities manipulation in model (4) which contradicts hypothesis 2b. The coefficient of SOX in model (5) and (6) are not significantly different from zero, therefore we cannot confirm hypothesis 2b based on model (5) and (6). Table 12 in appendix D, shows that instead of the SOX variable in model (3), the variable in model (2) is significant. For real activities manipulation there are no differences.

Dependent variable

Independent variables Abs_DACC RAM

(1) (2) (3) (4) (5) (6)

SOX -0.0008 -0.0036 -0.0046* -0.0254*** 0.0047 -0.0037

(0.0028) (0.0025) (0.0027) (0.0091) (0.0045) (0.0047)

Control Variables Yes No Yes Yes No Yes

Firm Fixed Effects No Yes Yes No Yes Yes

Year Fixed Effects No No No No No No

R2 0.0858 0.4985 0.5112 0.2380 0.8875 0.9015

N 1859 1949 1859 1614 1688 1614

Business ethics : what is the relationship between being corporate socially responsible and the level of earnings management?

Business Ethics

What is the relationship between being Corporate

Socially Responsible and the level of Earnings

Management?

Table of Contents

1. Introduction

2. Literature research and hypotheses development

2.1 Agency theory

2.2 Earnings Management

2.3 Corporate Social Responsibility

2.4. Corporate Social Responsibility versus Earnings Management

2.5 Sarbanes-Oxley Act versus Earnings Management

3. Methodology

3.1 Sample design

4. Descriptive statistics

4.1 Summary

4.2 Correlation

5. Results

5

.1 Validation of empirical model

5.2 Testing hypothesis 1

5.3 Testing hypothesis 2