Master Thesis
Analysing the relationship between Corporate Social Responsibility,
discretionary accruals and real earnings management.
Student name: Sheila Gracia Calvo Student number: 10603336
Date final version: 21st June 2015 Word count: 12,315
MSc Accountancy & Control, variant Accountancy
Amsterdam Business School
Faculty of Economics and Business, University of Amsterdam
Supervisor: dr. Georgios Georgakopoulos
Statement of Originality
This document is written by student Sheila Gracia Calvo 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.
Table of Contents
1.Introduction………...4
1.1Background………...4
1.2 Research question………6
1.3 Motivation of this study………..6
2. Literature review and hypotheses………..7
2.1 Literature review………..7
2.1.1 Earnings management………...7
2.1.1.1.Motives for earnings management………8
2.1.1.2. Detection of Earning management: accruals-based vs real earnings Management………..9
2.1.1.3. Agency theory………..9
2.1.2 Corporate Social Responsibility……….10
2.1.2.1. Legitimacy theory………...11
2.1.2.2. Stakeholder theory………..11
2.2. Research hypothesis……….12
3. Research methodology………14
3.1. Data and sample selection………14
3.2.Measurements of variables………15
3.2.1. Accrual-based earnings management………15
3.2.2. Real earnings management………16
3.2.3. Corporate social responsibility………..18
3.2.4. Control variables………19
3.3.Empirical model………...……….21
4. Results………...……...….……….……….22
4.1. Descriptive statistics…………..……….………….……….22
4.2 Regression analysis………..……….……….30
5.Results and conclusion………...………..34
Appendix……….37
1. Introduction
1.1Background
Nowadays a growing number of companies are trying to integrate Corporate Social Responsibility (hereafter CSR) into different aspects of their business activities (Hartojo and Jo, 2007). In the latest years, CSR has emerged and has developed into a global and broader concept; CSR embraces corporate social responsibility, responsiveness and socially beneficial activities of businesses (Carroll, 1991). The impact of business activities on the society have lead a widespread interest in CSR in research, online resources, business schools, politics and in society in general, which appear to be recognizing a significant expansion and degree in society’s demand for CSR (Mahoney and Thorne 2005). Additionally, to handle with the elevated attention given to firms’ impact on society, more than half of the Fortune 100 companies release CSR information in US (Galema et al., 2008). CSR is becoming more significant to firms in order to attend pressures from a wide group of stakeholders rather than an unique group, shareholders (Donalson and Preston, 1995) . CSR defenders suggest that firms should engage in socially responsible activities that benefit various stakeholders. Grant Thornton (2008) argues that CSR is an issue that concerns of all types of business rather than large corporations. Jamali (2006) considers CSR as a new concept that attracts a global interest and garners new repercussion in the global market.
Recent corporate accounting scandals have lead important losses for investors, legislative and corporate governance reforms that are encouraged to identify the causes of the scandals. A significant number of commentaries have been attributed to these accounting scandals that are related to the degeneration of the business ethics and morality, and the subsequent collapse of the CSR (New York Times 2002). Given that accountability and financial transparency are two of the key points of CSR, a closer examination of issues concerning earnings management is required (Chih 2008). 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 (Healy and Wahlen 1999). The extensive practice of earnings management negatively impacts on the quality of financial reports and increases information asymmetries between owners and managers. When an extensive practice of manage earnings occurs, financial reports inaccurately 4
reflect the performance of the firm, and, consequently, this reduces the capacity of outsiders to the understanding on the financial situation of a firm (Leuz 2003). Prior research on earnings management, suggested that capital markets, contractual arrangements and regulatory motivations induce managers to manipulate earnings reports (Walker 2013). Moreover the managerial opportunism based on agency theory predicts that conflicts of interest between managers and shareholders facilitate the opportunistic behaviour by managers with respect to earnings manipulations, it is an agency cost because managers give preference to their own interest (Baiman 1990).
Heal (2008) argues that the disclosure of accurate and precise financial earnings is essential for CSR because it provides to outsider participants a true and assured view of the firm’s operations. Consequently, it seems especially important to study how CSR and earnings management are related. Although the relationship between firm financial performance and corporate governance or the relation between CSR and financial performance has received important attention in numerous studies, empirical studies that examined directly whether CSR and earnings management are related have been few, and the results of these studies have been mixed (Kim et al. 2012). Prior et. al (2008) found out that a positive impact of earnings management practices on CSR and demonstrated that the combination of CSR and earnings management has a negative effect on financial performance. Kim. Y. et al (2012) investigated whether firms that exhibit CSR behave in a responsible manner to constrain earnings management, thereby delivering financial information to investors more transparent and reliable as compared to firms that do not converge the same social criteria. Their findings support that socially responsible firms are less likely to manage earnings through discretionary accruals and real activities manipulation. Moreover they find evidence that executives of the CSR firms are less likely to be subjected to Securities and Exchange Commission (SEC) investigations of GAAP violations. Regarding to Chih, H.L. et al. (2008) who studied the relationship between CSR and three different types of earnings management: earnings smoothing, earnings aggressiveness, and earnings losses, found out that a CSR firm tends not to smooth earnings, and displays less interest in avoiding earning losses. Scholtens (2013) concluded that firms that engage in CSR activities are less likely to manage earnings. Thus, increasing efforts in both improving CSR and investor protection may reduce earnings management. This study was carried out in an Asian context. Trèbucq and Russ (2005) did not find a significant relationship between CSR and earnings management. And Grougiou (2014) examined the bi-directional 5
relationship between CSR and earnings management in U.S commercial banks. He concluded that the mentioned relationship is not bidirectional, when the engagement in earnings management is higher, the engagement in CSR increases. On the other hand, involvement in CSR has not relationship with earnings managements.
The purpose of this study is to complement the existing literature and investigate how the CSR engagement and earnings management are connected. This study provides a comprehensive examination of the relationship between CSR and earnings management behaviour in U.S firms.
1.2 Research question
This paper explores the relation between earnings management and CSR. I intend to more precisely to analyse whether CSR mitigates or increases earnings managements, or on the other hand whether earnings management reduces or increases the CSR engagement. Specifically, I examine if firms that demonstrate CSR also behave in a responsible manner to constrain earnings management. Otherwise, this study examines whether firms engage in CSR activities to cover up the impact of corporate misbehaviour, earnings management precisely. And subsequently CSR firms are more likely to manage earnings. To be more precise, I will use to measure earnings management discretionary accruals and real activities manipulation to provide a more complete study of the trends in earnings management activities. Moreover, the relationship presented will be study in a bidirectional approach to the better and in depth understanding of this association.
Based on the aforementioned relationship between CSR and earnings management, the research question is presented as a follows:
“How and to what extent is the direction between Corporate Social Responsibility and earnings management?. Does the earnings management affected by CRS or does the CSR affected by the earnings management?”.
1.3 Motivation of this study
The aim of this study is to examine the effect between CRS and earnings management. As much as CSR as earnings managements have been raise a lot of concerns and research separately. While CSR has pushed environmental and social issues into mainstream, the main objective of the firms is maximizing financial performance and
firm value. In consequence, social responsibility activities have an important impact on the society in general.
First, this study contributes to the literature examining the role of CSR and earnings management and also the research fills the vacuum that exist in the literature investigating the role of ethics in financial reporting. Existing research on CSR focuses on the empirical connection between financial performance and firm’s social responsibility (El Ghoul et al. 2011). Although some studies attempt to analyse the relationship between CSR and earnings management, they provide mixed findings that limit the comprehension of the true association between CSR and earnings management (Grougiou et al. 2014). Due to the divergence of findings and the relevance of this relationship for the market and academics, Kim, Park and Wier (2012) proposed that more research in this field is needed.
Second, Prior literature on the association between earnings management and CSR has studied this relationship in one direction. Although Grougiou et al. (2014) explores the bi-directional relationship in U.S commercial banks, he points out that banks that manage earnings are actively concerned with CSR, but the opposite relation is not significant. I try to broaden the scope of this investigation using U.S listed companies.
2. Literature review and hypotheses
2.1 Literature review
2.1.1 Earnings management
Watts and Zimmerman (1978) define earnings management as managers exercising their discretion over the accounting numbers. Extensively, M. Walker (2013) defines earnings management as the use of managerial discretion over (within GAAP) accounting choices, earnings reporting choices, and real and economic decisions to influence how underlying economic events are reflected in one or more measures of earnings. Leuz et al. (2003) proposes that managers and controlling owners have incentives to manage reported earnings in order to hide the true firm performance and conceal their private control benefits from outsiders. Another point worth to noting in the literature is that accounting judgment to make financial reports has cost and benefits (Jiraporn 2008). The cost is the possible misallocation of resources that result from earnings management. On the other hand, benefits cover potential enhancement in 7
management’s communication of private information to external stakeholders, improving the decision making (Watts and Zimmerman,1986: Healy and Palepu, 1993). And there are some studies that argue the positive effect of earnings management. Subramanyan (1996) provides that the discretion of managers help to improve the capacity of earnings to reflect the underlying value.
2.1.1.1. Motives for earnings management
Healy and Wahlen (1999) identify three principal incentives that influence earnings management activities; capital market expectations and valuation, contracts in terms of targets related to reported earnings and government regulations.
Firstly, managers decide to manage earnings in order to affect short-term stock price performance. For example, Burgstahler and Eames (1998) suggest that managers take actions to manage earnings upward to avoid reporting earnings lower than analysts’ expectations. Ronen and Sadan (1981) suggested that managers may manage earnings in order to send private information to markets over future forecast about the firm. Zahra et al. (2005) find that firms engage in opportunistic behaviour by pressure and opportunity.
Secondly, management compensation contracts and lending contracts are stipulated on accounting data that create incentives for executives to manage earnings. Healy (1985) establishes that US directors manage earnings get their bonus covenant. However, there are some work that studies the influence of equity-based incentives for EM. Cheng and Warfield (2005) found out that managers with an important level of equity incentives are less likely to report strong positive earnings surprises and these firms are more likely to meet or beat the analysis forecast.
Lastly, earnings management literature has explored the effects of industry-specific regulation and anti-trust regulation. Accounting standard setters have demonstrated an interest in earnings management to go around industry regulation. The fall of Enron and other accounting scandals have led the introduction of new regulation as the Sarbanes-Oxley Act (SOX) in 2002. Watts and Zimmerman (1978) proved that firms who operate in regulated sector where authorities establish a tight control over prices and market shares, tend to manage earnings to appear less profitable.
2.1.1.2 Detection of earnings management: accrual-based and real earnings management
There are two different earnings management strategies, real activities manipulation (R-EM) and accrual based earnings manipulations (A-(R-EM). Real activities manipulation may change economic decisions to achieve financial targets; there may be real free cash flow consequences. Prior studies have shown evidence of firms altering real activities to manage earnings (Roychowdhury 2006; Graham et al. 2005). Accrual based earnings management is achieved by changing the accounting method used when presenting a transaction in the financial statements (Graham 2005 and Roychowdhury 2006). For example, depreciation rates; deferred tax assumptions; stock valuation assumptions; bad debt provisions; and revenue recognition. These kind of earnings have not impact on future cash flows.
There are evidence on previous literature that firms make choices between the two earnings management strategies and this choices vary on the industrial context. For example Cohen et al. (2008) proved that companies used more accrual-based earnings management before SOX, and before SOX firms tend to increase more earnings management. Cohen and Zarowin (2010) demonstrated that firms engaged more in both types of earnings in the period of a seasoned equity offering. Badertscher (2011) shows that managers engage in accruals earnings management in the early phase of overvaluation, while managers shift to real manipulations to maintain during more time the overvaluation of equity.
2.1.1.3. Agency theory
The most common approach that research uses to study earnings management is agency theory. The agency theory provides a structure for organizational relationships through a nexus of contracts. The traditional view of the agency theory was stablished by Jensen and Meckling (1976), suggest that the principal hires the agent to delegate responsibility and arises the separation of ownership (the owner) and control (the manager), together with the existence of information asymmetry within the firms, gives rise the possibility of opportunistic actions by the agent (the manager), who may have different objectives from those of the principal (the owner), and thus act in his self-interest (the agency problem). Prior research suggests that when there is an agency problem, managers may take operating decisions that are not in the best interest of the firm (Rutledge and Karim 1999).
Exits two components underlying the agency theory, there are adverse selection and moral hazard. Adverse selection appears when managers have access to information that the principal does not have, and this information is relevant to take decisions. And moral hazard is when the principal is unable to control and monitor the actions made by the manager (Lambright 2009).
Earnings management is considered a type of agency cost because managers look after their own interest by disclosing financial reports that do not represent an accurate way the firms ‘financial performance. As consequence, shareholders could not make the optimal decisions. Chih et al. (2008) argues that earnings management could increase information asymmetries between shareholders and management. Earnings management has been connected with decreases in stock prices (Dechow et al. 1995), increases in firm risk (Chatterjee et al. 1999) or increases of cost of capital (Botosan, 1997).
2.1.2 Corporate Social Responsibility
CSR holds that firms voluntarily integrate economic, social and environmental concerns in their operations and interactions with stakeholders (Branco and Rodrigues 2006) and maximize the expectations (Elkington, 1994, Dyllick ; Collison et al., 2003). Elkington (1997) describes the triple bottom line as the concept that firms should provide corporate social information about their economic, social and environmental impact.
There has been a significant amount of academic research dedicated to defining Corporate Social Responsibility, and in consequence the reviews the CSR literature have remained highly fragmented (Aguinis 2012).
During decades there have been studies examining firms ‘social concerns (Berle, 1931; Bowen,1953; Davis, 1960; Frederick, 1960). Initially Bowen (1953) defines CSR as the obligations of the business men to develop policies and to take decisions which are proper for the values and objectives of the society. And the CSR discipline has evolved thorough different research questions. For example, Peloza (2009) studies how to measure the impact of CSR on financial performance. Peloza and Shang (2011) conducted a research of how CSR can create value for stakeholders. Moreover, others studies have focused on specific disciplines such as marketing, organizational behaviour, human resources, operations or information systems (Aguinis 2012). Hartojo and Jo (2011) argue that there are different definitions of CSR, its generally involve to serving people, communities, and the environment in ways that go further what is
legally required. Aside from academics, there are organizations that provide different understanding of CSR. For instance, the Commission of the European Communities establish that being socially responsible means not only fulfil legal expectations, but also going beyond compliance and investing more in human capital, the environment and the relation with stakeholders.
2.1.2.1. Legitimacy theory
One perspective that is useful to understand the relationship between CSR engagement and social expectations is the legitimacy theory. Dowling and Pfeffer (1975) define legitimacy as a condition or status which exists when an entity´s value system is congruent with the value system of the larger social of which the entity is part. And society give to the companies the right to use resources in the figure of social contract. Organizations are constrained to act in compliance with the terms of their social contract, the social contract represents the multitude of explicit and implicit expectations that the society has about how the organization should conduct its operations (Donaldson et al. 1995). Deegan and Unerman (2011) determine that the social contract is how a firm should operate to fulfil the external expectations. It is common practice to argue that CSR engagement positively influences a company’s reputation, protects its license to operate and contributes to its legitimacy. Some researchers use legitimacy theory to explain the incentives that firms have to engage in CSR. Porter and Kramer (2006) suggest that CSR engagement contribute to attain the license to operate in the society and guarantee the future existence. On the other hand, the legitimacy gap originates when a company performances against the social norms and breaks the social contract (Brown and Deegan 1998).
Furthermore, legitimacy theory has been used to interpret the CSR disclosures. Cho (2009) have studied how firms can steer legitimacy through their CSR disclosures. Moreover, Hopwood (2009) presented how firms engage in environmental reporting to expand their legitimacy. Ashforth and Gibbs (1990) reported that when an organization operates correctly, it reports CSR information to maintain its legitimacy.
2.1.2.2. Stakeholder theory
Stakeholder theory studies the firm in the context of a wider range of implicit and explicit stakeholders having legitimate aspirations, important claims and power regarding the firm (Jones et al., 1999). Ullmann (1985) proposes that organizations will
react to the demands of those stakeholder groups that control the resources necessary to the organization´s operations.
Managers continually encounter demands from multiple stakeholder groups to dedicate resources to CSR. Many managers have responded to heightened stakeholder interest in CSR in a very positive way, by dedicating additional resources to promote CSR. Corporations conduct CSR not only because their purpose is to generate profits and obey laws, but also because they are required to be ethical and socially supportive (Carroll 1991, 1999). According to the CSR literature, CSR reporting is consistent with the shareholders ‘demand for information and CSR activities, and disclosures are primarily a response to shareholder demand (Dhaliwal et al. 2012, Kim et al. 2012). Additionally, Jones (1995) suggests that firms that engage in social activities that affect stakeholder relations can reduce transactions and agency cost. Fombrun et al. (2000) proposes that organizations that engage in CSR activities and project a positive reputation, improve their capacity to negotiate with suppliers and government and in subsequent these organizations reduce their cost of capital.
In summary, stakeholder theory predicts that managers conduct CSR to fulfil their moral, ethical, and social duties for their stakeholders and strategically achieve corporate goals for their shareholders (Jo and Hartojo 2011). Disclosure about CSR engagement and corporate behaviour help to create a positive impression between stakeholders (Orlitzky, Schmidt and Rynes, 2003).
2.2. Research hypothesis.
Previous literature on the exploration of the negative relationship of CSR and earnings management study this relation through an ethical and reputation perspective. While Jones (1995) argues that CSR firms have an incentive to be honest, trustworthy, and ethical because such behaviour is beneficial to the firm. Therefore, if managers engage in CSR, then they are more likely to constrain earnings management and to make responsible decisions maintaining transparency in financial reporting. Linthicum et al. (2010) shows that managers may use CSR to enhance the firm´s reputation and constraint earnings management to reduce the damage to its reputation. Here with these approaches are expected that CSR can mitigate agency problems, particularly the agency conflicts between powerful shareholders and minority shareholders. As consequence, CSR may diminish the incentives that executives have to manage earnings. Regarding these ethical and reputation approaches a negative relationship
between CSR and earnings management is expected and the following hypothesis are proposed:
Ethical hypothesis
Hypothesis 1: The extent of Corporate Social Responsibility is negatively associated with the extent of earnings managements.
Hypothesis 1a: Corporate Social Responsible firms are less likely to engage in accruals based earnings management.
Hypothesis 1b: Corporate Social Responsible firms are less likely to engage in real activities manipulation.
On the other hand, prior literature provides a positive relationship between CSR and earnings management suggesting an opportunistic use of CSR. From agency theory Carroll (1979) suggests that CSR may be linked to the pursuit of managers ‘self-interest. Hemingway and Maclagan (2004) argue that firms engage in CSR as a form of reputation insurance, which gives them a “license to operate” and camouflage some corporate misconduct. This wave indicates that the choice to engage in CSR activities may be to give the impression to stakeholders that the firm is transparent, when, in fact managers are more likely to attempt to mislead stakeholders as to the value of the firm and financial performance. Under such conception, CSR is used as an entrenchment mechanism (Cespa and Cestone 2007; Prior et al. 2008) which demonstrate that firms that employ earnings management are more likely to engage in responsible social activities to reduce the dismissal risk and increase the remuneration. Martinez-Ferreo and García Sánchez (2015) find evidence that CSR could be used as managerial entrenchment to avoid the detection of managerial discretion and this aforementioned relationship is moderated by stakeholder protection.
In consideration of this stream of literature a positive relation between CSR and earnings management could be expected.
Entrenchment hypothesis:
Hypothesis 2: More earnings management is positively associated with more Corporate Social Responsibility engagement.
Hypothesis 2a: Firms that engage in accrual based earnings manipulation are more likely to engage more in CSR.
Hypothesis 2b: Firms that engage in real activities manipulation are more likely to engage more in CSR.
3. Research methodology
3.1. Data and sample selection
Based on the research question, to test the hypotheses, the data of CSR are gathered from the Kinder Lydenburg and Domini (KLD) database to construct the measure of CSR. It is issued by KLD Reach & Analytics, Inc. KLD has covered the S&P 500 since 1991 and expanded its coverage in 2002 to the largest 3000 U.S publicly traded companies.
KLD is the largest multidimensional Corporate Social Performance database available to the public and is used extensively in research on corporate social performance (Deckop et al., 2006). Waddock (2003) presents KLD as the de facto corporate social performance research standard at the moment. Mattingly and Berman (2006) state that KLD has emerged as the standard to measure the corporate social engagement. In addition, Chatterji et al. (2009) asserts that KLD ratings are the most extensively accepted by scholastic area.
Data from Compustat are collected to measure earnings management; Compustat database contains annual industrial and research files from 1987. It contains balance
sheets, income statements, cash flow statements, and stock data.
The sample data used is from listed U.S firms and it is available across five years, from 2005 to 2009. The analysis does not expand its coverage until more recent period because the KLD database does not cover all its ratings during the same period of time. However, the present research analyses as well the relationship between earnings management and CSR during the beginning of the financial crisis in 2008, and if the financial crisis has changed the trends on earnings manipulations or CSR engagement.
After matching KLD data with Compustat database, an initial sample of 14,749 firm-year observations is obtained. I exclude 42 firm-year observations of financial institutions (SIC codes 6000-6999) due to different characteristics of accrual in these firms. Of the remaining 14,707 firm-year observations, I have sufficient information to calculate proxies for discretionary accruals, real activities manipulation, CSR ratings and control variables.
3.2. Measurements of variables
3.2.1. Accrual-based earnings management
I use discretionary accruals to proxy for accrual-based earnings management. The early literature has established several ways to measure earnings management. The model used to measure accruals earnings management is the modified cross-sectional Jones model (Jones 1991) as described in Dechow et al. (1995) and used in Cohen et al. (2008). The modified Jones model estimated for two-SIC digits is described as :
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝑇𝑇𝐴𝐴𝑡𝑡 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1 = 𝛼𝛼0+ 𝛼𝛼1∗ � 1 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1� + 𝛼𝛼2∗ � ∆𝑅𝑅𝑅𝑅𝑅𝑅𝑡𝑡 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1� + 𝛼𝛼3∗ � 𝑃𝑃𝑃𝑃𝑅𝑅𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1� + 𝜀𝜀𝑇𝑇 (1) Where:
Total Accrualst= EBIXTt – CFOt, represent the total accruals and is defined as the earnings before extraordinary items and discontinued operations (annual Compustat data item 123) and the operating cash flows reported in the statement of cash flows (annual Compustat data item 308-annual Compustat data item 124).
Assetst-1: total assets (annual Compustat data item 6).
∆REVt: revenues in year t less revenues in year t-1. (Annual Compustat data item 12)
PPEt: gross value of property, plant, and equipment in year t. (Annual Compustat data item 7).
The coefficient estimates from Equation (1) is used to estimate the firm-specific normal accruals (NAt):
𝑁𝑁𝐴𝐴𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1 = 𝛼𝛼0+ 𝛼𝛼^1∗ 1 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1+ 𝛼𝛼^2∗ (∆𝑅𝑅𝑅𝑅𝑅𝑅𝑡𝑡−∆𝐴𝐴𝑅𝑅𝑡𝑡) 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1 + 𝛼𝛼^3 ∗ 𝑃𝑃𝑃𝑃𝑅𝑅𝑡𝑡 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1 (2) Where:
∆ARt is the variation in accounts receivable from the previous year (annual Compustat
data item 2). Following Cohen et al. (2008) I measure for discretionary accruals is the difference between total accruals and the normal accruals.
DAt= 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝑇𝑇𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴 𝑡𝑡
𝑡𝑡−1 −
𝑁𝑁𝐴𝐴𝑇𝑇
𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1
(3)
I compute the absolute value of discretionary accruals to proxy for earnings management and refer to it as ABS_DA, because earnings management can comprise either income-increasing or income-decreasing accruals, and accruals can modify over time and change the direction .
3.2.2. Real earnings management
I rely on prior studies ( Roychowdhury 2006; Cohen et al. 2008; Cohen and Zarowin 2010; Zang 2012) to develop the proxies for real activities manipulation. Specifically, I use the following three measures to detect real activities manipulation: abnormal levels of operating cash flows from operations (AB_CFO), abnormal production costs (AB_PROD) and abnormal discretionary expenses (AB_DISEXP). First I generate the normal level of cash flows, productions costs and discretionary expenses using the model developed by Dechow et al. (1998) and used in Roychowdhury (2006):
I express normal CFO as a linear function of sales and variation in sales:
𝐶𝐶𝐶𝐶𝐶𝐶𝑡𝑡 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1= 𝛼𝛼0+ 𝛼𝛼1� 1 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1� + 𝛽𝛽1� 𝑆𝑆𝑇𝑇𝑇𝑇𝐴𝐴𝐴𝐴𝑡𝑡 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1� + 𝛽𝛽2� ∆𝑆𝑆𝑇𝑇𝑇𝑇𝐴𝐴𝐴𝐴𝑡𝑡 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1� + 𝜀𝜀𝑇𝑇 (4) Where:
CFOt: cash flow from operations in year t (annual Compustat data item 308-annual
Compustat data item 124). Assetst: total assets
Salest: net sales
∆Salest= St-St-1. Change in sales.
The abnormal cash flow from operations (AB_CFO) is the residual from the model, actual cash flows less CFO from equation (4).
Prior studies ( Roychowdhury 2006; Cohen et al. 2008; Zang 2012) define production costs as the sum of cost of goods sold (COGS) (annual Compustat data item 41) and change in inventory (annual Compustat data item 3).
𝐶𝐶𝐶𝐶𝐶𝐶𝑆𝑆𝑡𝑡 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1= 𝛼𝛼0+ 𝛼𝛼1� 1 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1� + 𝛽𝛽1� 𝑆𝑆𝑇𝑇𝑇𝑇𝐴𝐴𝐴𝐴𝑡𝑡 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1� + 𝜀𝜀𝑇𝑇 (5) Where: 16
COGSt= the cost of goods sold in year t.
Similarly, the model for normal inventory growth is estimated following the equation:
∆𝐼𝐼𝑁𝑁𝑅𝑅𝑡𝑡 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1= 𝛼𝛼0+ 𝛼𝛼1� 1 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1� + 𝛽𝛽1� ∆𝑆𝑆𝑇𝑇𝑇𝑇𝐴𝐴𝐴𝐴𝑡𝑡 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1� + 𝛽𝛽2� ∆𝑆𝑆𝑇𝑇𝑇𝑇𝐴𝐴𝐴𝐴𝑡𝑡−1 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1� + 𝜀𝜀𝑇𝑇 (6) Where:
∆INVt is the change in inventory in year t.
Relying in Roychowdhury (2006), Cohen et al. (2008), and Zang (2012), production costs are defined as PRODt = COGSt + ∆INVt. Normal production cost are estimated from equation (5) and (6).
𝑃𝑃𝑅𝑅𝐶𝐶𝑃𝑃𝑡𝑡 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1= 𝛼𝛼0+ 𝛼𝛼1� 1 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1� + 𝛽𝛽1� 𝑆𝑆𝑇𝑇𝑇𝑇𝐴𝐴𝐴𝐴𝑡𝑡 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1� + 𝛽𝛽2� ∆𝑆𝑆𝑇𝑇𝑇𝑇𝐴𝐴𝐴𝐴𝑡𝑡 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1� + 𝛽𝛽3� ∆𝑆𝑆𝑇𝑇𝑇𝑇𝐴𝐴𝐴𝐴𝑡𝑡−1 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1� + 𝜀𝜀𝑇𝑇 (7)
Similarly as AB_CFO, abnormal production cost (AB-PROD) is the residual from the model.
The normal discretionary expenses is estimated using the following equation used by Roychowdhury (2006), Cohen et al. (2008) and Zang( 2012):
𝑃𝑃𝐼𝐼𝑆𝑆𝑅𝑅𝐷𝐷𝑃𝑃𝑡𝑡 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1= 𝛼𝛼0+ 𝛼𝛼1� 1 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1� + 𝛽𝛽1� 𝑆𝑆𝑇𝑇𝑇𝑇𝐴𝐴𝐴𝐴𝑡𝑡−1 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑇𝑇𝐴𝐴𝑡𝑡−1� + 𝜀𝜀𝑇𝑇 (8) Where:
DISEXPt: discretionary expenses in year t, determined as the sum of R&D expenses (annual Compustat data item 46), advertising expenses (annual Compustat data item 45), and selling, general and administrative expenses (annual Compustat data item 189). For every firm-year, abnormal discretionary expenditure (AB_EXP) is the residual from the model.
The abnormal CFO (AB_CFO), abnormal production cost (AB_PROD), and abnormal discretionary expenses (AB_DISEXP) are computed as the difference
between the actual values and the normal level predicted from equations (4), (7), and (8). In order to capture the effects of real earnings management through the three variables, I rely on Cohen et al. (2008). To obtain the effects of real management through these three variables in a comprehensive measure, I compute a separate variable, RM_PROXY, by combining the three real earnings management variable. RM_PROXY is the sum of the standardized variables. However, the three individual variables have different effects for earnings management and their values and direction are uncommon. I will study the results to the aggregate real earnings management (RM_PROXY) as well as the three individual variables (AB_CFO, AB_PROD, and AB_DISEXP).
3.2.3. Corporate social responsibility
The KLD database covers a company's strengths and concerns for approximately 80 variables in seven major social areas. The seven inclusive social ratings are: community relations, corporate governance, diversity, employee relations, environment, human rights and product characteristics. For each variable positive rating indicate strengths, and negative ratings indicate concerns. Additionally to those seven categories, KLD also rates six exclusionary issues that include alcohol, gambling, firearms, military, nuclear power and tobacco activities. Engagement in any of these six controversial items result in a negative rating. Exclusionary screen differ from the inclusive ratings in that only concern ratings, but no strength ratings are assigned. I only use the inclusive screen in constructing CSR scores1. Following Kim et al. (2012) corporate governance
is constructed differently from CSR, in order to resolve the influence of CSR and corporate governance, I control for corporate governance and construct CRS scores based on the five remaining categories .
Relying on prior research ( Deckop et al., 2006 Johnson and Greening, 1999; Griffin and Mahon, 1997; Waddock and Graves, 1997), the KLD index score is computed subtracting total concerns from total strengths and assign equal weight to each area. If the resulting CSR index is positive, the company is identified as socially responsible. If the index is negative or zero, the company is identifies as less socially responsible.
1
These dimensions do not refer to firms ‘discretionary activities.
18
CSR2 = (total strengths of community - total concerns of community) + (total strengths of diversity - total concerns of diversity) + (total strengths of employee relations- total concerns of employee relations) + (total strengths of environment - total concerns of environment)+ (total strengths of human rights - total concerns of human rights) + (total strengths of product- total concerns of product)
3.2.4. Control variables
The model that I will use includes various control variables that could affect financial reporting behaviour and CSR performance. And to avoid correlation problems between variables I insert the following control variables in the model :
• GROWTH OPPORTUNITIES: Using the market-to-book ratio, measured as market value equity divided by the book value of common equity. I consider the market’s perception with regard to future growth. Barth et al. (1999) and Chih et al. (2008) argue that the market reacts negatively to firms that change their sequence of earnings increase, and corporations with higher growth have motivations to manage their earnings upwards It is expected that firms with higher growth have more incentives to meet earnings targets. Roychowdhury (2006) suggests that firm specific growth and the size of the organization may potentially explain significant variation in earnings management.
• FIRM SIZE: measured as total assets. Prior studies (Waddock and Graves 1997; McWilliams and Siegel 2000; Prior et al. 2008) show that firm size is correlated with CSR performance. Besides the relationship between earnings management and firm size is not clear. Dechow and Dichev (2002) find that larger firms tend to have more stable operations and hence report smaller amounts of discretionary accruals. Moreover, larger firms have higher incentive to manage earnings (Richardson et. Al, 2002). On the other hand, firm size can be used as a proxy for information asymmetry. Larger firms are subject to closer examination by outsiders and are required to divulge their information, and in consequence there is more difficult for larger firms to manage earnings. Insiders of small firms are able to reserve their private information more easily that large firms (Lee and Choi, 2002).
2
See appendix for KLD variables used.
19
• ROA: financial performance. I use return on assets (ROA), which is the ratio of earnings before interest and taxes to the total value of assets. ROA measures financial performance and its returns of assets. It is worthy to notice that accounting dimensions are more sensitive to manipulations than market measures. Ortitzky et al. (2003) show that measurement as ROA or ROE, are exposed to discretionary allocations of capital. Similarly, Dechow, Sloan, and Sweeney (1995) demonstrate that accruals-based earnings management are associated with performance.
• AUDITOR: Further, to the extent earnings management might differ for firms audited by large audit firms (Becker et al. 1998; Francis et al. 1999). DeAngelo (1981) presents that auditors that have more clients “have more to lose” by not reporting accurately, in consequence auditing carried out by larger firms enhance the audit quality. An indicator variable, for those firms using one of the Big 4 auditors (PricewaterhouseCoopers, Deloitte, Ernst &Young and KPMG) is included. In order to measure the auditor quality, a dummy variable is used, which takes a value equal to 1 when the sample firm is audited by the Big Four auditors, and 0 when the firm is not audited by the Big Four auditors.
• LEVERAGE: long term debt to equity scaled by total assets. Debt to equity ratio measures the risk of the firm and reflects the firm’s leverage. There is not clear relationship between leverage and earnings management on the literature. The first approach stablishes that firms which are subjected to more risk, and in consequence higher leverage will be predisposed to manage earnings more than less risky firms (Kim and Yi 2006). And high leveraged organizations tend to manage their earnings in an aggressive tendency Sweeney (1994). The second approach suggests that high leverage firms may imply less earnings management. Dechow and Skinner (2000) find out that firms with bigger leverage are less likely to report small increases in earnings. Finally, Chung and Kallapur (2003) do not obtain any significant relation between leverage and earnings management.
• RD INTENSITY: R&D (research and development) is the ratio of expenditures to total revenues. McWilliams and Siegel (2001) find that R&D intensity in industry is positively associated with CSR and earnings. Several studies argue that R&D investments perform in favour of earnings management to achieve
targets (Clinch, 1991; Dechow and Sloan, 1991). On the other hand, R&D investments make a firm’s technology more flexible, allowing the introduction of customer preferences into the design of products. This resource enhances customer satisfaction and, consequently the firm’s CSR.
• GOVERNANCE: corporate governance and CSR influence firms ‘financial reporting behaviours (Kim, Park and Wier 2012). I control for corporate governance in the models by introducing a net score of corporate governance dimension from the KLD.
3.3 Empirical models
To capture the relation between earnings management and CSR firms, I estimate two models. In both specifications, I consider the same set of control variables, although the independent variable vary depending the direction between CSR and earnings management.
In particular, in order to test hypothesis 1 and study the association of earnings management on CSR, I rely on the following regression:
Regression 1: Relation of accruals earnings management on CSR.
𝐴𝐴𝐴𝐴𝐴𝐴_𝐷𝐷𝐴𝐴𝑇𝑇 = 𝛼𝛼0+ 𝛼𝛼1∗ 𝐶𝐶𝐴𝐴𝐶𝐶𝑇𝑇+ 𝛼𝛼2∗ 𝐶𝐶𝑀𝑀_𝑃𝑃𝐶𝐶𝑃𝑃𝑃𝑃𝑃𝑃𝑇𝑇+ 𝛼𝛼3∗ 𝐺𝐺𝐶𝐶𝑃𝑃𝐺𝐺𝐺𝐺𝐺𝐺𝑇𝑇−1+ 𝛼𝛼4∗
𝐴𝐴𝑆𝑆𝑆𝑆𝑆𝑆𝑇𝑇−1+ 𝛼𝛼5∗ 𝐶𝐶𝑃𝑃𝐴𝐴𝑇𝑇−1+ 𝛼𝛼6∗ 𝐴𝐴𝐴𝐴𝐷𝐷𝑆𝑆𝐺𝐺𝑃𝑃𝐶𝐶𝑇𝑇+ 𝛼𝛼7∗ 𝐿𝐿𝑆𝑆𝐿𝐿𝑇𝑇−1+ 𝛼𝛼8∗ 𝐺𝐺𝑃𝑃𝐿𝐿𝑆𝑆𝐶𝐶𝐺𝐺𝐴𝐴𝐺𝐺𝐶𝐶𝑆𝑆𝑇𝑇+ 𝛼𝛼9∗ 𝐶𝐶𝐷𝐷𝑇𝑇+ 𝜀𝜀𝑇𝑇
The results are consistent with the hypothesis 1a if α1 is negative and significant. A negative association between the absolute value of accruals-based earnings management and CSR score is expected.
Regression 2: Relation of real earnings management on CSR.
𝐶𝐶𝑀𝑀_𝑃𝑃𝐶𝐶𝑃𝑃𝑃𝑃𝑃𝑃𝑇𝑇 = 𝛼𝛼0+ 𝛼𝛼1∗ 𝐶𝐶𝐴𝐴𝐶𝐶𝑇𝑇+ 𝛼𝛼2∗ 𝐴𝐴𝐴𝐴𝐴𝐴𝑃𝑃𝐴𝐴𝑇𝑇+ 𝛼𝛼3∗ 𝐺𝐺𝐶𝐶𝑃𝑃𝐺𝐺𝐺𝐺𝐺𝐺𝑇𝑇−1+ 𝛼𝛼4∗ 𝐴𝐴𝑆𝑆𝑆𝑆𝑆𝑆𝑇𝑇−1+ 𝛼𝛼5∗ 𝐶𝐶𝑃𝑃𝐴𝐴𝑇𝑇−1+ 𝛼𝛼6∗ 𝐴𝐴𝐴𝐴𝐷𝐷𝑆𝑆𝐺𝐺𝑃𝑃𝐶𝐶𝑇𝑇+ 𝛼𝛼7∗ 𝐿𝐿𝑆𝑆𝐿𝐿𝑇𝑇−1+ 𝛼𝛼8∗ 𝐺𝐺𝑃𝑃𝐿𝐿𝑆𝑆𝐶𝐶𝐺𝐺𝐴𝐴𝐺𝐺𝐶𝐶𝑆𝑆𝑇𝑇+ 𝛼𝛼9∗ 𝐶𝐶𝐷𝐷𝑇𝑇+ 𝜀𝜀𝑇𝑇
The results are consistent with the hypothesis 1b if α1 is negative and significant. A negative association between the proxy for real earnings management and CSR score is expected.
Prior literature point out that firms can choose between the two different mechanisms to manage earnings (Cohen 2008). Zang (2012) supports that the use between two earnings management method depends on their relative costs. To control for the substitutive nature of these two earnings management methods, as in Cohen (2008), I include ABS_DA, a proxy for accrual-based earnings management, as a control variable in the real earnings management (RM-PROXY) regression and a proxy for real activities manipulation as a control variable in the accrual-based manipulations (ABS_DA) regression.
Specifically, to test hypothesis 2 and analyse the association of CSR on earnings management, I rely on the following regression:
Regression 3: Relation of CSR on earnings management
𝐶𝐶𝐴𝐴𝐶𝐶𝑇𝑇 = 𝛾𝛾0+ 𝛾𝛾1∗ 𝐴𝐴𝐴𝐴𝐴𝐴_𝐷𝐷𝐴𝐴𝑇𝑇+ 𝛾𝛾2∗ 𝐶𝐶𝑀𝑀_𝑃𝑃𝐶𝐶𝑃𝑃𝑃𝑃𝑃𝑃𝑇𝑇+ 𝛾𝛾3∗ 𝐺𝐺𝐶𝐶𝑃𝑃𝐺𝐺𝐺𝐺𝐺𝐺𝑇𝑇−1+ 𝛾𝛾4∗
𝐴𝐴𝑆𝑆𝑆𝑆𝑆𝑆𝑇𝑇−1+ 𝛾𝛾5∗ 𝐶𝐶𝑃𝑃𝐴𝐴𝑇𝑇−1+ 𝛾𝛾6∗ 𝐴𝐴𝐴𝐴𝐷𝐷𝑆𝑆𝐺𝐺𝑃𝑃𝐶𝐶𝑇𝑇+ 𝛾𝛾7∗ 𝐿𝐿𝑆𝑆𝐿𝐿𝑇𝑇−1+ 𝛾𝛾8∗ 𝐺𝐺𝑃𝑃𝐿𝐿𝑆𝑆𝐶𝐶𝐺𝐺𝐴𝐴𝐺𝐺𝐶𝐶𝑆𝑆𝑇𝑇+ 𝛾𝛾9∗ 𝐶𝐶𝐷𝐷𝑇𝑇+ 𝜀𝜀𝑇𝑇
Hypothesis 2a is supported if γ1 is positive and significant. A positive relation between the CSR net score and the absolute value of discretionary accruals is predicted. Hypothesis 2b is supported if γ1 is positive and significant. A positive relationship between the CSR score and the real earnings management is expected.
4. Results
4.1 Descriptive statistics
I start with the sample distribution by the two-digit SIC industry code presented in Table 1. Business services is the most represented industry (47.24 % of the sample, SIC code 73) and followed by electronic and other electronic equipment (10.99% of the sample, SIC code 28) and instruments and related products (8.49% of the sample, SIC code 38) . On the contrary, oil and gas extraction (0.20% of the sample, SIC code 13)
and petroleum and coal products (0.33% of the sample, SIC code 29) are the least represented industry within the sample.
TABLE 1: Industry frequency table
SIC 2
digits N. % sample % cumulative
Oil & Gas Extraction 13 30 0.20% 0.20%
Food & Kindred Products 20 109 0.74% 0.95%
Furniture & Fixtures 25 70 0.48% 1.42%
Paper & Allied Products 26 110 0.75% 2.17%
Chemical & Allied Products 28 1,046 7.11% 9.28%
Petroleum & Coal Products 29 47 0.32% 9.60%
Rubber & Miscellaneous Plastics Products 30 118 0.80% 10.40%
Stone, Clay, & Glass Products 32 67 0.46% 10.86%
Primary Metal Industries 33 109 0.74% 11.60%
Fabricated Metal Products 34 146 0.99% 12.59%
Industrial Machinery & Equipment 35 993 6.75% 19.34%
Electronic & Other Electric Equipment 36 1,617 10.99% 30.34%
Transportation Equipment 37 334 2.27% 32.61%
Instruments & Related Products 38 1,249 8.49% 41.10%
Miscellaneous Manufacturing Industries 39 105 0.71% 41.825
Communications 48 117 0.80% 42.61%
Wholesale Trade - Durable Goods 50 202 1.37% 43.99%
Wholesale Trade - Nondurable Goods 51 108 0.73% 44.72%
General Merchandise Stores 53 103 0.70% 45.42%
Food Stores 54 23 0.16% 45.58%
Automotive Dealers & Service Stations 55 59 0.40% 45.98%
Apparel & Accessory Stores 56 182 1.24% 47.22%
Furniture & Home furnishings Stores 57 49 0.33% 47.55%
Eating & Drinking Places 58 223 1.52% 49.07%
Miscellaneous Retail 59 200 1.36% 50.42%
Business Services 73 6,948 47.24% 97.67%
Amusement & Recreation Services 79 94 0.645 98.31%
Health Services 80 161 1.09% 99.40%
Engineering & Management Services 87 88 0.60% 100%
Total 14,707 100%
Table 2A provides summary of the descriptive statistics of the full sample. The descriptive analysis shows that total accruals is negative at -0.078 with standard 23
deviation of 0.148. In addition, discretionary accruals (DA) is positive at 0.006. I estimate the absolute value of discretionary accruals (ABS_DA) for the study because I cannot predict the direction of the earnings management, they could present as increases or decreases of the accruals. The mean value for ABS_DA is 0.063, it is slightly smaller than the results obtained by Cohen et al. (2008) or Kim et al. (2012). The mean variables for the real earnings management are 0.028, -0.025, and -0.023 respectively for CFO, PROD and DISEXP, while the mean value for the RM_PROXY is -0.02. One of the main variable of interest is net score of CSR (CSR_SCORE), the mean has a negative value (-0.176) and standard deviation of 1.662 this indicates that companies studied on the sample do not perform well in their social activities, and in consequence obtain a negative net scone as a mean.
TABLE 2: Descriptive statistics of selected variables
Panel A: Descriptive statistics of full sample
Standard 25th 75th
n Mean Median Deviation Percentile percentile
ABS_DA 14,707 0.063 0.039 0.094 0.012 0.075 DA 14,707 0.006 0.003 0.113 -0.029 0.043 TOTAL ACCRUALS 14,707 -0.078 -0.056 0.148 -0.094 -0.025 CFO 14,707 0.028 0.058 0.155 -0.041 0.122 PROD 14,707 -0.025 -0.022 0.174 -0.138 0.039 DISEXP 14,707 -0.023 -0.012 0.259 -0.132 0.068 CSR_SCORE 14,707 -0.176 0.000 1.662 -1.000 1.000 RM_PROXY 14,707 -0.020 -0.034 0.184 -0.115 0.087 GROWTH 14,707 3.078 1.693 38.080 1.180 3.752 SIZE 14,707 5.729 5.724 2.061 4.966 6.915 LEVERAGE 14,707 0.343 0.086 10.306 0.007 0.007 AUDITOR 14,707 0.549 1.000 0.498 0.000 1.000 ROA 14,707 -0.131 0.057 4.475 -0.056 0.110 GOVERNANCE 14,707 -0.185 0.000 0.693 -1.000 0.000 RD 14,707 0.316 0.071 13.767 0.028 0.089
The results of the descriptive statistics show that the mean value of ROA is -0.131, a market book ratio of 3.078 and on average companies have 34.43% of liabilities over its assets. Additionally, on average the 54.9 percent of the sample
companies are audited by the Big4 audited companies and the mean average of R&D expenditure is 31.6 percent.
Additionally, I conclude that CSR firms are bigger (SIZE), with higher growth expectations and ROA , while non CSR firms allocate more R&D expenses and their leverage ratio is bigger than CSR firms. Regardless, auditor I find that 55.5 percent of CRS firms are audited by one of the big 4 companies while 54.7 percent of non CSR.
Panel B: Descriptive statistics between CSR and non-CSR firms
CSR NON CSR n mean standard deviation n mean standard deviation ABS_DA 3,841 0.062 0.093 10,866 0.064 0.095 DA 3,841 0.0048 0.1113 10,866 0.007 0.114 TOTAL_ACCRUALS 3,841 -0.080 0.155 10,866 -0.077 0.146 CFO 3,841 0.025 0.152 10,866 0.029 0.155 PROD 3,841 -0.020 0.179 10,866 -0.027 0.173 DISEXP 3,841 -0.022 0.256 10,866 -0.023 0.260 RM_PROXY 3,841 0.018 0 .185 10,866 -0.021 0.183 CSR_SCORE 3,841 1.719 1.243 10,866 -0.846 1.215 GROWTH 3,841 3.120 21.944 10,866 3.063 42.338 SIZE 3,841 5.752 2.064 10,866 5.721 2.060 LEVERAGE 3,841 0.319 5.677 10,866 0.351 11.505 AUDITOR 3,841 0.555 0.497 10,866 0.547 0.498 ROA 3,841 -0.089 1.192 10,866 -0.146 5.158 GOVERNANCE 3,841 -0.228 0.716 10,866 -0.169 0.684 RD 3,841 0.256 8.050 10,866 0.338 15.285
Table 2B contrasts the mean and standard deviation between CSR and non CSR companies. CSR firms are defined as those with a positive CSR net score. And non CSR firms are those with a null or negative CSR net score. I have 3,841 observations for CSR firms while non CSR firms are composed by 10,866 observations .The average value of the net CSR score for the CSR firms is 1.719, while the mean for the non CSR is -0.846. Furthermore, I observe that the ABS_DA is greater in the non CSR than in CSR firms ( 0.064 for non CSR and 0.062 in CSR). In addition, DA and total accruals are as well greater in those firms than are considered non CSR, this higher value of accruals could indicate that companies that engage more in CSR, engage less in discretionary accruals.
For the real earnings management I find that while CFO is higher in non CSR firms, PROD, DISX and RM_PROXY have a mean value bigger in CSR companies. On the whole, the results indicate that while non CSR firms are more predisposed to manage accruals based earnings management, CSR companies are more likely to engage in real activities manipulation.
Table 2C shows the values of the studied variables during the period between 2007 and 2009 to study how these variables perform during the financial crisis in 2008, and if there is any change of trend for the variables most essential in the study.
I observe that the ABS_DA increases in 2008 with the beginning of the financial crisis and in consequence in 2009 decreases from 0.072 to 0.051 to the lower value in these three years. The value of the discretionary accruals increases during 2008 and decreases drastically in 2009, DA reaches its maximum value in 2008. Regarding the real earnings manipulation (RM_PROXY) I perceive a drastically reduction during the study period of time from 0.032 in 2007 to -0.030 in 2008, and -0.090 in 2009. ABS_DA, total accruals, CFO, DISX, RM and CSR_SCORE decreases while DA and PROD increases.
In addition, I observe that during the crisis variables as LEVERAGE, GOVERNANCE and expenditure in R&D increased, while GROWTH, SIZE, AUDITOR, and ROA decreased their values.
The examination of the correlation matrix (table 3) shows a significant and negative relationship between discretionary accruals (DA) and real earnings management (RM_PROXY), which indicates the shift between earnings.
I observe that CRS is negatively correlated with DA and CFO. And positively correlated with PROD, DISEXP and RM_PROXY. But these relationships are not statically significant.
Panel C: Descriptive statistics, 2007-2009
2007 2008 2009
n mean median
Standard
deviation n mean median
Standard
deviation n mean median
Standard deviation ABS_DA 2,926 0.060 0.029 0.090 2,923 0.072 0.043 0.089 2,902 0.051 0.012 0.095 DA 2,926 -0.011 -0.029 0.108 2,923 0.018 0.043 0.113 2,902 -0.003 -0.008 0.108 TOTAL_ACCRUALS 2,926 -0.086 -0.093 0.142 2,923 -0.087 -0.056 0.148 2,902 -0.107 -0.094 0.142 CFO 2,926 0.059 0.099 0.155 2,923 0.045 0.108 0.133 2,902 -0.066 -0.095 0.136 PROD 2,926 -0.052 -0.127 0.182 2,923 -0.061 -0.147 0.179 2,902 -0.033 -0.080 0.149 DISX 2,926 0.025 0.068 0.233 2,923 -0.014 -0.034 0.242 2,902 0.009 0.023 0.225 RM_PROXY 2,926 0.032 0.084 0.160 2,923 -0.030 -0.073 0.164 2,902 -0.090 -0.117 0.176 CSR_SCORE 2,926 -0.199 0.000 1.680 2,923 -0.197 0.000 1.713 2,902 -0.190 0.000 1.707 GROWTH 2,926 4.888 4.099 44.692 2,923 3.004 1.340 64.256 2,902 2.107 0.973 26.142 SIZE 2,926 6.252 6.915 2.092 2,923 5.477 5.724 2.047 2,902 5.507 5.192 2.035 LEVERAGE 2,926 0.215 0.117 1.308 2,923 0.377 0.048 9.993 2,902 0.692 0.086 20.539 AUDITOR 2,926 0.400 0.000 0.490 2,923 0.378 0.000 0.485 2,902 0.389 0.000 0.488 ROA 2,926 -0.021 0.096 0.842 2,923 -0.100 0.100 2.400 2,902 -0.429 -0.200 9.571 GOVERNANCE 2,926 -0.207 0.000 0.705 2,923 -0.202 0.000 0.693 2,902 -0.187 0.000 0.691 RD 2,926 0.115 0.047 0.482 2,923 0.575 0.071 23.511 2,902 0.490 0.089 17.836 27
TABLE 3: Correlations matrix
*,**,*** Indicate statistical significance at 0.1, 0.05, and 0.01 levels respectively.
1 2 3 4 5 6 7 1.ABS_DA 1 2.DA -0.1742*** 1 3.TOTAL_ACCRUALS -0.4018*** 0.7676*** 1 4.CFO -0.2015*** -0.0748*** 0.1016*** 1 5.PROD 0.0674*** 0.0109 -0.0212** -0.2464*** 1 6.DISEXP 0.2252*** -0.1102*** -0.2174*** -0.448*** -0.4261*** 1 7.RM_PROXY 0.212*** -0.208*** -0.2411 -0.0243*** 0.1423*** 0.6279*** 1 8.CSR_SCORE -0.012 -0.0064 0.0021 -0.0044 0.0101 0.0043 0.0119 9.GROWTH 0.0425*** 0.0244*** -0.0179** -0.0314*** 0.0142* 0.0566*** 0.0668*** 10.SIZE -0.2875*** -0.0333*** 0.2106*** 0.2518*** -0.0666*** -0.0362*** 0.0975*** 11.LEVERAGE 0.0511*** -0.0466*** -0.1516*** -0.0308*** 0.027*** 0.0429*** 0.0602*** 12.AUDITOR -0.0763*** 0.0119 0.1579*** 0.1557*** 0.1905*** -0.158*** 0.0892*** 13.ROA -0.0983*** 0.0999*** 0.2232*** 0.0751*** -0.0454*** -0.0726*** -0.0823*** 14.GOVERNANCE -0.0149* 0.0049 0.0086 0.0073 -0.0199** 0.004 -0.0071 15.RD 0.0451*** -0.0174** -0.0784*** -0.0675*** 0.0099 0.1149*** 0.1146*** 28
TABLE3: CONTINUED 8 9 10 11 12 13 14 15 1.ABS_DA 2.DA 3.TOTAL_ACCRUALS 4.CFO 5.PROD 6.DISEXP 7.RM_PROXY 8.CSR_SCORE 1 9.GROWTH -0.0126 1 10.SIZE 0.0003 0.0078 1 11.LEVERAGE -0.0152* -0.0019 -0.0371*** 1 12.AUDITOR 0.0212** -0.0026 0.4107*** -0.0193** 1 13.ROA 0.0251*** 0.0028 0.0664*** -0.8604*** 0.0361*** 1 14.GOVERNANCE 0.0527*** -0.0294*** -0.0075 -0.0073 -0.0095 0.0043 1 15.RD -0.0126 0.0006 -0.0226*** 0.0661*** -0.0165** -0.3272*** -0.0091 1
*,**,*** Indicate statistical significance at 0.1, 0.05, and 0.01 levels respectively.
According to the correlations, it is interesting to notice that ABS_DA is positively and significant correlated with GROWTH, LEVERAGE, and RD, this implies that the bigger growth expectations, the more leverage company and the more expenditure in R&D, more is the engagement in accrual-based earnings management. On the other hand, smaller companies, lower firms’ financial performance, weaken corporate governance and less audited by the Big4 are more prone to manage discretionary accruals.
Regarding the real earnings manipulations, a positive and significant correlation suggests that the more leveraged companies, the more growth carried out and the more investment in R&D, the more real earnings manipulations engagement. On the other side, a negative and significant correlation between AUDITOR and RM_PROXY is observed.
4.2 Regression analysis
Regression 1: Relationship between CSR and accruals-based earnings management. TABLE 4 ABS_DA Coefficient t-stat P>t CRS_SCORE -0.0007 (-1.66)* 0.097 RM_PROXY 0.121103 (10.36)*** 0 GROWTH 7.22E-05 (1.44) 0.15 SIZE -0.01466 (-26.37)*** 0 LEVERAGE -0.00119 (-2.08)** 0.038 AUDITOR 0.007152 (5.16)*** 0 ROA -0.00388 (-2.53)** 0.011 GOVERNANCE -0.00195 (-1.98)** 0.048 RD -0.00028 (-1.6) 0.111 R2 0.1502 Adjusted R2 0.1496 F-value (Pr>F) 101.45 n 14,707
*,**,*** Indicate statistical significance at 0.1, 0.05, and 0.01 levels respectively. All tests statistics and significance are calculated based on Huber-White standard errors. Table 4 tests ethical hypothesis 1.a, the results of accrual based practices on CSR. I use the absolute value of discretionary accruals (ABS_DA) to measure the reaction of
accruals-based earnings management on CSR. Consistent with hypothesis 1, there is a negative association between CSR_SCORE and the accruals based earnings management, the estimated coefficient on CSR is negative and significant (p<0.1). According with the ethical and reputation hypothesis, CSR firms tend to engage less in earnings management through accruals.
Moreover, the real earnings management (RM_PROXY) is positively and significantly (p<0.01) associated with accruals based earnings management (ABS_DA). Consistent with Kim et al. (2012).
It is interesting to notice that small companies, with a leverage ratio and ROA smaller and firms with a worse governance are more likely to engage in discretionary accruals. I found SIZE, LEVERAGE, ROA, GOVERNANCE, and LEVERAGE negatively and significantly associated with ABS_DA, indicating that bigger companies, with higher leverage ratio, with better financial performance and with better governance structures are less likely to engage in discretionary accruals. On the contrary, AUDITOR is positively associated with ABS_DA (p<0.01), companies that engage more in accrual based earnings management are audited more by one of the Big 4 companies.
Taken into consideration, the regression analyses support the hypothesis 1a that CSR firms tend to engage less in discretionary accruals.
Table 5 presents the results of multiple analyses for real activities manipulation on CSR. For the regressions of CFO, PROD, and DISEXP, the estimated coefficient of CSR_SCORE is not significant. But I observe that CSR_SCORE is positive and significant (p<0.05) related with RM_PROXY. In this point if I take the overall RM_PROXY I can suggest that hypothesis 2b is not supported, because the regression indicates that CSR firms tend to engage more in real manipulation, while the hypothesis states that CSR firms engage les in real activities manipulation. I use ABS_DA as control variable, I found out ABS_DA positive and significant for PROD and RM_PROXY (P<0.01), consistent with Kim et al. (2012). While ABS_DA is negative and significant for CFO.
Regression 2: Relationship between CSR and real-based earnings management. TABLE 5
CFO PROD DISEXP
PM PROXY
Coefficient Coefficient Coefficient Coefficient
(t-stats) (t-stats) (t-stats) (t-stats)
CSR SCORE -0.0009 0.00079 0.00196 0.00181 (-1.33) -0.96 -1.63 (2.03)** ABS_DA -0.2221 0.06425 0.6467 0.48882 (-7.17)*** (1.91)* (11.3)*** (9.95)*** GROWTH -0.0001 6.6E-05 0.00031 0.00027 (-1.47) -0.84 (3.71)*** (4)*** SIZE 0.0135 -0.0136 0.0137 0.01365 (17.49)*** -(15.18)*** (9.85)*** (14.35)*** LEVERAGE 0.00109 -0.001 0.00113 0.00125 -1.12 (-1.45) -0.95 -1.48 AUDITOR 0.02137 0.09139 -0.0952 0.01753 (8.25)*** (33.89)*** -(22.72)*** (6)*** ROA 0.00349 -0.0038 0.00138 0.0011 -1.41 (-2.08)** -0.56 -0.63 GOVERNANCE 0.00154 -0.0046 0.00318 0.00012 -0.86 (-2.31)** -1.08 -0.06 RD -0.0003 -0.0002 0.00205 0.00149 (-0.94) (-1.06) (5.63)*** (6.53)*** R2 0.0916 0.066 0.0935 0.0913 Adjusted R2 0.0911 0.0654 0.093 0.0907 F-value (Pr>F) 71.37 145.24 73.18 73.41 n 14,707 14,707 14,707 14,707
*,**,*** Indicate statistical significance at 0.1, 0.05, and 0.01 levels.
All tests statistics and significance are calculated based on Huber-White standard errors. Returning to the aggregate variable for real earnings manipulation , I observe that GROWTH, SIZE, AUDITOR and RD are positively and significantly associated with RM_PROXY. Indicating that bigger firms, with a higher growth and larger levels of R&D expenditures are more likely to engage in real activities manipulation. Moreover the firms that are audited by Big4 tend to engage more in real earnings management.
In summary, the regression does not support the ethical hypothesis 1b, the evidence shows that CSR organizations tend to engage more in real earnings
manipulations. The results show a weak signification between CSR_SCORE and RM_PROXY, moreover CSR_SCORE is not significant with CFO, PROD or DISEXP. the evidence shows that CSR organizations tend to engage more in real earnings manipulations.
Regression 3: Relationship between earnings management and CSR. TABLE 6 CSR SCORE Coefficient (t-stats) P>t ABS_DA -0.25186 (-1.66)* 0.097 RM_PROXY 0.162113 (2.05)** 0.041 GROWTH -0.0005 (-0.92) 0.355 SIZE -0.01414 (-1.89)* 0.059 LEVERAGE 0.003962 (1.82)* 0.069 AUDITOR 0.083365 (2.77)*** 0.006 ROA 0.017226 (2.88)*** 0.004 GOVERNANCE 0.125554 (5.68)*** 0 RD -7.78E-07 0 1 R2 0.0046 Adjusted R 0.004 F-value (Pr>F) 9.59 n 14,707
*,**,*** Indicate statistical significance at 0.1, 0.05, and 0.01 levels.
All tests statistics and significance are calculated based on Huber-White standard errors.
Table 6 tests the results of CSR on earnings management, entrenchment hypothesis. Results indicate that the effect of accrual based earnings management (ABS_DA) on CSR is negative and significant (p<0.1), while the effect of real earnings manipulations (RM_PROXY) on CSR is positive and significant (p<0.05). The analyses demonstrate that more discretionary accruals is negatively associated with more CSR and in consequence results are not consistent with hypothesis 2a, on the other hand companies that engage in real activities manipulation are busily concerned with CSR activities. Hypothesis 2b is accepted and suggest than companies carry out CSR activities as a result of their engagement on real earnings management.
In addition, the regression reports that bigger firms tend to reduce their engagement in CSR, because SIZE is negatively related to CSR. Moreover LEVERAGE, AUDITOR, ROA and GOVERNANCE are positively and significant
related to CSR. This indicates that those companies with better financial performance, leverage ratio, with better governance structures and more audited by Big 4 firms engage more on CSR. GROWTH opportunities and RD expenditure have not statistical relationship with CSR.
Taken together, the results show that exists a statically significance of CSR on earnings management, although its effect and direction is not the same in both types of accounting manipulations. I observe that firms that engage more in DA are less likely to engage in CSR, while firms that engage more in RM are more likely to engage more in CSR. The regression, support in some extent the entrenchment hypothesis. The data rejects hypothesis 2a, because firms that engage in discretionary accruals tend to engage less in CSR. Nevertheless, firms that engage in real earnings manipulations tend to engage more in CSR.
5. Results and conclusion
In this paper, I examine the relationship between CSR and earnings management practices. I explore the aforementioned relation in a bi-directional approach to understand if CSR motivates managers to decrease accounting manipulations due to an ethical vision or earnings management are the incentive to engage in CSR and cover up the impact of their dishonest behaviour. While there are previous literature exploring the relationship between CSR and earnings management in one direction (Kim et al. 2012; Prior at a. 2008; Heltzer. 2011; Chil et al. 2008). Few investigations have been carried out exploring the bi-directional association, Grougiou et al. (2014) find out inconclusive results analysing the bi-directional relationship in U.S banks.
TABLE7: summary of the results.
Ethical hypothesis 1.a +CSR - AEM
Ethical hypothesis 1.b +CSR + REM
Entrenchment hypothesis 2.a + AEM - CSR
Entrenchment hypothesis 2.b + REM +CSR
Table 7 shows a summary of the results. The analysis suggest different relationship between CSR and earnings managements. On an ethical perspective, the study proposes that CSR firms tend to manipulate less discretionary accruals (AEM), regarding real earnings management (REM) I find that managers use CSR as an entrenchment mechanism, because firms that engage more in real activities
