• No results found

Voluntary disclosure of non-financial information and the cost of debt

N/A
N/A
Protected

Academic year: 2021

Share "Voluntary disclosure of non-financial information and the cost of debt"

Copied!
65
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

Amsterdam Business School

Voluntary Disclosure of Non-financial Information and the Cost of

Debt

Name: Mitchell Wansink Student number: 10846166

Thesis supervisor: Dr. Réka Felleg Date: 20 June 2016

Word count: 18173

MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam

(2)

Statement of Originality

This document is written by student Mitchell Wansink who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

Abstract

This study provides evidence on the relationship between the voluntary disclosure of non-financial information and the cost of public and private debt. Specifically, this study examines if the effect of voluntary disclosure of non-financial information on the cost of debt is larger for firms with private debt than for firms with public debt. Overall, my results suggest that sustainability reporting provide non-financial information that private debt distributers are likely to evaluate as important. I find significant evidence for the fact that the effect of voluntary disclosure of non-financial information on the cost of debt is larger for the cost of private debt than for the cost of public debt. These results have implications for financial management. Organizations can reduce their cost of private debt by improving their CSR performance.

Keywords Non-financial information, Corporate Social Responsibility Reporting, Voluntary Disclosure, Cost of Public Debt, Cost of Private Debt.

(4)

Table of Content

List of Tables ... 6

1 Introduction ... 7

2 Literature Review ... 11

2.1 The concept of non-financial information ... 11

2.2 Agency theory and information asymmetry ... 11

2.3 Legitimacy and stakeholder theory ... 13

2.4 Signaling theory and the credibility of non-financial information ... 14

2.5 Reporting frameworks ... 16

2.5.1 Reporting frameworks of non-financial information ... 16

2.6 Cost of capital ... 19

2.6.1 Cost of equity ... 19

2.6.2 Cost of debt ... 20

3 Hypothesis ... 22

3.1 Hypothesis development ... 22

4 Sample and Empirical Design ... 24

4.1 Sample ... 24 4.2 Empirical design ... 24 4.2.1 Main variables ... 27 4.2.2 Control variables ... 29 4.2.3 Hypothesis testing ... 30 5 Results... 32 5.1 Descriptive statistics ... 32

5.2 Results of hypothesis tests ... 41

(5)

5.2.2 Regression results cost of private debt ... 42

5.2.3 Seemingly unrelated estimation regression results ... 43

5.3 Additional analyses ... 46

5.3.1 Lagged time periods... 46

5.3.2 Multiple subsamples ... 46

6 Conclusion ... 51

References ... 54

Appendix A: CSR Areas of Strengths and Concerns ... 62

(6)

List of Tables

Equation 1: Regression Model Cost of Public Debt ... 25

Equation 2: Regression Model Cost of Private Debt ... 26

Equation 3: Total amount of Public Debt ... 28

Equation 4: Cost of Public Debt ... 28

Equation 5: Cost of Private Debt ... 29

Table 1: Number of observations per fiscal year ... 32

Table 2: Cost of public debt: Descriptive statistics for the variable DISC ... 33

Table 3: Cost of private debt: Descriptive statistics for the variable DISC ... 34

Table 4: Descriptive statistics for the samples sorted by Industry ... 35

Table 5: Cost of public debt: Descriptive statistics for regression variables ... 37

Table 6: Cost of private debt: Descriptive statistics for regression variables... 38

Table 7: Cost of public debt: Correlation matrix ... 39

Table 8: Cost of private debt: Correlation matrix ... 40

Table 9: Regression results of the cost of public and private debt ... 45

Table 10: Regression results of the cost of public and private debt with lagged variables ... 48

Table 11: Results of the cost of public and private debt across different sample periods ... 49

(7)

1 Introduction

This study examines the effect of voluntary disclosure of non-financial information on the cost of public and private debt of US companies. Specifically, I compare the cost of public debt and private debt in relation to the voluntary disclosure of non-financial information as public and private debtholders significantly differ in their ability to monitor and process information effectively and efficiently (Denis & Mihov, 2003). The research question is motivated by the economic importance of debt markets (Bharath, Sunder, & Sunder, 2008) and the increasing level of interest in voluntarily disclosed non-financial information, such as environmental, social and ethical performance (EY, 2009).

Debt financing is a predominant source of long-term external funding for US companies and differs from that of equity financing (Bharath, et al. 2008). The relation between the cost of equity and non-financial information has been studied extensively (e.g., Dhaliwal, et al. 2011; Diamond & Verrecchia, 1991; Baron, Harjoto, & Jo, 2009; Deegan, 2004). However, the effect of voluntary disclosure of non-financial information on the cost of public debt and the cost of private debt has not been examined. Prior literature highlights the significant differences between the private and public debt markets (e.g., Bharath, Sunder, & Sunder, 2008; Friedman, 1987; Kale & Meneghetti, 2011). According to Bharath, et al. (2008), lenders differ across the private and public debt market in the flexibility of resetting contract terms, in the ability to access and interpret information, to monitor the borrower, and the cost of renegotiating the contract. This results in the fact that private debt has a significant advantage over public debt in terms of monitoring efficiency (Denis & Mihov, 2003). However, Rajan (1992) states that borrowers can also be negatively affected by private lenders by extracting rents and distorting management incentives.

Voluntary disclosure of non-financial information gives companies the opportunity to be transparent in communicating these subjects affecting their performance. Dhaliwal, et al. (2011) argues that disclosures of sustainability activities contain information beyond that contained in performance ratings. Belcagem and Omri (2014) state that the quality of this voluntarily disclosed information in annual reports is essential to the proper functioning of equity markets and at the heart of modern financial problems. Nowadays, it is important that companies report about non-financial information within their industry and work environment in order to acquire legitimacy within the social environment in which they operate (Deegan & Blomquist, 2006). The US Securities and Exchange Commission (SEC) encourages organizations to consider

(8)

non-financial measures when determining whether disclosable material trends or uncertainties have developed related to securities and derivatives (Greene, et al., 2015).

Traditionally, research has focused on voluntary disclosure and the cost of equity and the payoff function of equity holders (e.g., Botosan, 1997; Dhaliwal, Zhen Li, Tsang, & George Yang, 2011; Diamond & Verrecchia, 1991; Botosan & Plumlee, 2002). Botosan (1997) and Dhaliwal, et al. (2011) find a positive relation between voluntary disclosure of non-financial information and the cost of equity. They argue that the disclosure of financial information and CSR activities are associated with a higher prior year cost of equity capital. Furthermore, organizations that disclose CSR information attract dedicated institutional investors and analyst coverage. However, my study is interesting because of the fact that the payoff function of debt holders is concave and therefore different from that of equity holders (Dhaliwal, et al. 2011). The pay-off function of debtholders is concave because of the fact that it is based upon a fixed claim and debt holders do not benefit from the upside potential of the increasing value of the organization. The differences between public and private debtholders matter in this research as this might give one an advantage over the other when acquiring and interpreting the voluntarily disclosed non-financial information. Goss & Roberts (2010) argue that providers of private debt are, contrary to the providers of public debt, among the best able to engage in detailed monitoring when assessing the impact of CSR activities. Therefore, I expect that the effect of voluntary disclosure of non-financial information on the cost of debt is larger for firms with private debt than for firms with public debt because of the fact that the providers of private debt are among the best at monitoring the impact of sustainability activities, processing information and renegotiating debt contracts.

The few studies that examine the relation between CSR activities and cost of debt only focus on bank loans (Goss & Roberts, 2010; Hoepner, Oikonomou, Scholtens, & Schröder, 2014), the credit rating assigned to the organization by rating agencies (Attig, Ghoul, Guedhami, & Suh, 2013) or the quality of the disclosed information (Sengupta, 1998). However, these studies do not provide empirical evidence on the effect of voluntary disclosure of non-financial information on the cost of public and private debt. Therefore, in this study I examine the effect of voluntary disclosure of non-financial information on the cost of public and private debt.

In order to empirically examine the difference in effect of sustainability reporting on the cost of public and private debt, I use the average CSR score based on data from the MSCI ESG KLD Database as a proxy for sustainability performance. Furthermore, I use the model of Lin, et al. (2013) as a proxy for the cost of public debt and the model of Francis, et al. (2005) as a proxy for the cost of private debt. In addition, I use two samples to examine if the effect differs

(9)

between the cost of public and private debt. The first sample is related to the cost of public debt and contains 3,648 observations of 1,003 distinct firms from 2003 until 2013. The second sample is related to the cost of private debt and contains 20,403 observations of 3,774 distinct firms from 2003 until 2013. I run an Ordinary Least Square (OLS) regression on both samples and store the estimation results of the two regression models into two separate variables. I use a seemingly unrelated estimation regression combined with a Chi2 test in order to compare the

regression coefficient of my main variable of interest, the overall measure of CSR performance, computed from the two different models of the cost of public and private debt. Based on the results of the individual regressions, I find that firms that engage in sustainability reporting and produce reports with a high average disclosure score have a lower cost of private debt. However, I do not find any evidence for the fact that this effect is significant for the cost of public debt. Furthermore, when comparing the effect of the variable Average Disclosure Score on the cost of public and private debt, I find significant evidence for the fact that this effect is larger for the cost of private debt than for the cost of public debt. Hence, based on the results of the individual regression and the Chi2 test, I conclude that the effect of voluntary disclosure of

non-financial information is larger for firms with private debt than for firms with public debt.

These results have implications for financial management, specifically debt management. Given the fact that organizations frequently raise external financing, organizations can reduce their cost of private debt by improving their CSR performance (Attig, Ghoul, Guedhami, & Suh, 2013). My study on the effect of voluntary disclosure of non-financial information on the firms’ cost of public and private debt contributes to the literature by extending the scope of traditional research that is concentrated on the disclosure of (non-)financial information and the relation to equity capital. My study provides managers with information about the long-term performance related to debt management, performance sustainability and long-term strategy development of a firm. The results of my study document a lower cost of private debt when organizations engage in sustainability and produce sustainability reports with a high average disclosure score. This result might provide managers, who include sustainability reporting in their long-term strategy development, with more information in order to obtain an optimal finance structure because of the fact that private debt distributers are likely to evaluate this information as important. In addition, the findings presented in my study might give managers an incentive to voluntarily disclose non-financial information in order to reduce debt financing costs, provide a better information environment for external users and potentially reduce information asymmetry. Furthermore, my study contributes to the existing literature from a standard-setting perspective because of the fact that the results of my study provide further support for policies emphasizing

(10)

sustainability commitment within organizations as voluntarily disclosed sustainability information can be used in order to reduce financing costs.

The rest of this study proceeds as follows. Chapter 2 provides a literature review about the agency, legitimacy and signaling theory in relation to sustainability reporting. I discuss the motivations and hypothesis development in chapter 3. Chapter 4 covers the sample description and the empirical design of this study. Chapter 5 describes the descriptive statistics, results and additional analyses. Chapter 6 provides the conclusion.

(11)

2 Literature Review

2.1 The concept of non-financial information

The interest of stakeholders in social, corporate environmental and ethical performance has increased significantly over the past years (EY, 2009). There are multiple terms used for non-financial information such as sustainability reporting, corporate social responsibility reports (CSR), corporate responsibility (CR) and environmental, social and governance (ESG) reports (Pohl & Tolhurst, 2010). Moon (2002) states that CSR is the voluntary contribution of goods, services and finance to a community and excludes activities directly related to the production and commerce of an organization. Organizations produce reports with voluntarily disclosed non-financial information for various reasons, including informing stakeholders, in order to reduce the information asymmetry between the organization and its stakeholders (Simnett, Vanstraelen, & Chua, 2009).

Corporate social responsibility reporting is an important tool for organizations in order to communicate with stakeholders about their corporate and social activities. According to Golob and Bartlett (2006), CSR reporting performs an important role for public relations in creating and the communication of a mutual understanding, handle potential conflicts and in achieving legitimacy. It is important for organizations to report about their social impact (Deegan & Blomquist, 2006). Furthermore, disclosing true and relevant information about social responsibility can have benefits for stakeholders, organizations and society (Golob & Bartlett, 2006).

2.2 Agency theory and information asymmetry

Market allocations under uncertainty will not be optimal in an unconstrained manner in the presence of moral hazard (Grossman & Hart, 1983). Moral hazard occurs when two parties come to an agreement with each other and the actions of these individuals cannot be observed or contracted upon (Holmström, 1979). Moral hazard is directly related to the principal-agency problem (Grossman & Hart, 1983). According to Wallace (2004) the agency theory, also called the Stewardship (Monitoring) Hypothesis, refers to the conflict of interest between the principal and the agent. Jensen and Meckling (1976) define this agency relationship as a contract between

(12)

one or more persons in order to perform a service on their behalf and which involves delegating the authority for decision making to the agent. This means that the principal engages with another person, the agent. The conflict of interest between the principal and the agent exists due to the fact that both parties want to maximize their own interest. Furthermore, there is a good reason to believe that the agent does not (always) act in the best interest of the principal (1976).

In order for the principal to contain this conflict of interest, the principal needs to establish the appropriate incentives and introduce monitoring costs to limit the aberrant activities of the agent. Given the dependence between the agent and the principal it is important to notice that there is an optimal degree of risk sharing between the two parties (Grossman & Hart, 1983). According to Jensen and Meckling (1976) it is impossible for the principal to ensure that the agent makes the best decisions in the interest of the principal at zero costs. In most principal-agency relationships, the principal and the agent will induce positive monitoring and bonding costs, which can be non-pecuniary and pecuniary. According to Wallace (2004) the absence of monitoring will result in the potential extraction of resources. Furthermore, most of the time there will be aberration between the decisions that will maximize the welfare of the principal and the decisions which the agent will make. The loss of welfare of the principal due to the difference in decisions is referred to as the residual loss (1976). The costs related to this issue are called the agency costs and is defined by Jensen and Meckling (1976) as the sum of the monitoring expenditures made by the principal, the bonding expenditures made by the agent and the residual loss.

The conflicting interest between the agent and the principal becomes a problem when this conflict is combined with information that is asymmetrically distributed between the agents and the principals (Linder & Foss, 2015). Asymmetric information among the parties is the source of the moral hazard problem as a result of the fact that actions of these individuals cannot be observed or contracted upon (Holmström, 1979). This information problem or “lemons” problem can, according to Akerlof (1970), lead to a potential breakdown in the functioning of the capital market. The solution to this problem is an optimal contract between the principal and the agent where the agent will provide full (voluntary) disclosure of (private) information (Healy & Palepu, 2001).

(13)

2.3 Legitimacy and stakeholder theory

An organization depends on the willingness of society in which it operates to allow them to continue to exist (O'Donovan, 2002). Stakeholders offer organizations multiple resources such as capital, customers, employees, material and legitimacy in order to continue their operational activities (Deegan, 2002). An organization or business aims for maximizing profit, while they have a moral obligation to act in a socially responsible manner (Shocker & Sethi, 1973). This so called ‘social contract’ between the organization and individual members of society is the foundation of an intangible social agreement (O'Donovan, 2002).

Deegan and Blomquist (2006) state that organizations will adopt a specific disclosure strategy as a result of their own perceptions of what society expects the organization to do. They argue that when an organization fails to act in accordance with the terms stated in the social contract, it has a negative impact on the operations and existence of the organization (2006). However, according to Elsbach and Sutton (1992) acquiring and maintaining this legitimacy is very difficult for most organizations. Legitimacy theory is derived from the theory of organizational legitimacy (Dowling & Pfeffer, 1975). An organization is legitimate when there is congruence between the social environment where they operate in and the social values that are associated with their activities (Deegan & Blomquist, 2006). Legitimacy theory discusses the general expectations of society, while stakeholder theory provides a more refined resolution by referring to particular groups within society (2006).

The stakeholder theory let the management of an organization understand which group of stakeholders are deserving or requiring management attention and which groups are not (Mitchell, Agle, & Wood, 1997). Freeman (1984, p. 46) defines a stakeholder as “any group or individual who can affect or is affected by the achievement of the organization’s objectives”. In this way, different stakeholders will have different views on how an organization should carry out the operational activities. As a result, there will be multiple contracts with different stakeholders instead of one contract with society in general (Deegan & Blomquist, 2006). Furthermore, organizations do not only need to do what is expected by society, they also need to inform general society about their activities, especially when the organization changes the scope of activities (Deegan & Blomquist, 2006).

These legitimizing and stakeholder strategies can also be used by a corporation to provide information to counter or offset information that is publicly available and that can possibly be harmful to the company (Deegan & Blomquist, 2006). Legitimacy strategies can also be implemented at an industry level in order for companies to deflect attention from activities

(14)

undertaken by other industry participants. In order to do this, the annual report has been considered an important and major document for organizations in order to influence the reaction of the society in which they operate (O'Donovan, 2002). Salancik and Meidl (1984) argue that voluntarily disclosed information in the annual report is used to influence society by sending specific signals to the general public. Furthermore, a lot of research has been done about the fact that organizations voluntarily disclose non-financial information in order to send signals and messages about their social and environmental activities to society and other stakeholders (Frost & Wilmshurst, 1998; Deegan, Rankin, & Voght, 2000; Deegan & Blomquist, 2006).

The social environment where the organization operates can hold the organization responsible for their operating activities. In order for an organization to be legitimate, it needs to inform general society about their activities. Voluntary disclosure of non-financial information can help an organization to provide information to the general public and might tribute to the legitimacy of the organization within a social environment. According to Goss and Roberts (2010) firms can deviate from their main goal of wealth maximization of shareholders in order to meet the demand to perform some societal good. Therefore, legitimacy and stakeholder theory are directly related to and provide an explanation for the reason why organizations voluntarily disclose information since the success of an organization depends on the extent to which it satisfies the expectations of key stakeholders (Ye & Zhang, 2011).

2.4 Signaling theory and the credibility of non-financial information

In contract theory, signaling explains how information asymmetry can be reduced by the party with more information signaling or credibly conveying it to other members (Morris, 1987). Morris (1987) argues that signaling theory was originally developed to explain problems related to information asymmetry in different markets. This is a method where the informed agent takes costly actions in order to reveal the agent’s information or type (Farrel & Rabin, 1996).

However, there is some concern among researchers about the disclosure of non-financial information due to comparability and (potential) credibility issues of the information (Hobson & Kachelmeier, 2005). There is an ongoing debate about the quality and value relevance of non-financial information (Cohen, Holder-Webb, & Zamora, 2015). Most of the non-financial information is assured due to the fact that the audit of the information is mandated by regulators and required by law (Darus, Sawani, Zain, & Janggu, 2014). In order to disclose non-financial

(15)

information to certain groups, for example specific regulators, there may be some requirements set to the information disclosed in particular (Simnett, Vanstraelen, & Chua, 2009). Furthermore, the reporting choices which are available to managers for reporting the financial statement are regulated by accounting standards (Healy & Palepu, 2001). There is, however, no regulation available regarding non-financial stand-alone reports which are developed for the general public (2009). This results in the fact that there is no regulation that requires assurance about the disclosed non-financial information and might lead to cheap talk, meaning that it does not directly affect payoffs. However, given the fact that people respond to it, talk does affect payoffs (Farrel & Rabin, 1996). Farrel and Rabin (1996) argue that cheap talk often does matter, but it does not generally lead to efficiency.

According to Reeve (2003) sustainability assurance is assurance that covers auditing and verification and evaluates the quality of public reports, the management systems and competencies that deliver the required information and which underpin an organization’s performance. Cohen, et al. (2015) argue that voluntary disclosed non-financial information tends to be more positive and more in line with impression management than the objective of providing complete and detailed information. Simnett, et al. (2009) state that assurance of the non-financial information is related to a desire to enhance and improve the credibility of the disclosed information. However, due to the variation in the content of the reports, it can be difficult to provide the desired assurance on the disclosed information (2009). According to Darus, et al. (2014) there are significant variations and ambiguities in the assurance statements of sustainability reports, even with the current standard related to the assurance of these reports such as the G4 (GRI) and AA1000APS. However, assurance of the information provided by organizations is needed in order to mitigate the information asymmetry with institutional creditors (Blackwell, Noland, & Winters, 1998).

KPMG examined during multiple years whether the largest companies in various countries produce a non-financial information report and whether there is assurance provided on these reports (KPMG, 2015; KPMG, 2013; KPMG, 2011). The results state that there is an increase in the frequency of types of sustainability reports among the largest companies and an increase in assurance provided about sustainability reports. In order for the assurance process to be effective, the assurance statement need to provide the user with an accurate representation of the findings and the intended level of assurance by the practitioner (Hasan, Roebuck, & Simnett, 2003). The difference between the level of expected performance as envisioned by the accountant and the user of the financial statements is called the expectation gap (Liggio, 1974). In addition, ineffective communication in an assurance report can increase information

(16)

asymmetry (Hodge, Subramaniam, & Stewart, 2009). However, the credibility of information is important for public debtholders because they rely upon the published information but, due to the lack of regulation related to the assurance of sustainability reports, they have no means of verifying the information in particular.

2.5 Reporting frameworks

2.5.1 Reporting frameworks of non-financial information

According to Hahn (2012) organizations are increasingly held accountable for the impact of their activities on the environment and society. It is, however, very difficult for stakeholders to assess the performance regarding the sustainability and corporate social responsibility of an organization (Hahn & Lülfs, 2014). In order to reduce the information asymmetry, companies are expected to communicate the responsibilities and norms to their stakeholders (Phillipe & Durand, 2011). CSR reports have become an important topic in accounting and management in order to communicate these aspects to the stakeholders (Hahn & Lülfs, 2014). 95 percent of the 250 largest global companies report on their Corporate Social Responsibility activities in order to comply with the increasing demand of pressure to explain their business activities (KPMG, 2011). Archambeault, et al. (2008) state that organizations that voluntary disclose information and as a result increase their transparency, can enhance trust with different stakeholders.

Corporate Social Responsibility reports are subjected to very limited regulatory guidance (Deegan, 2004). Therefore, the decision usefulness of CSR reports can be questioned regarding the aspect of accountability and reliability (Hahn & Lülfs, 2014). In order to overcome this gap, projects such as the Global Reporting Initiative (GRI) and the AccountAbility Standards 2008 (AA1000AS) were developed (2014). Both standards are used worldwide in order to produce reports related to sustainability and CSR activities and provide assurance about the published information in an attempt to increase the credibility of the information. The GRI guidelines on sustainability reporting are nowadays considered ‘the facto global standard’ (KPMG, 2011).

(17)

2.5.1.1 The Global Reporting Initiative (GRI)

The Global Reporting Initiative (GRI) is one of the most used frameworks related to standalone CSR Reports (EY, 2009). This reporting standard provides a set of principles to be considered when reporting on sustainability which impacts the environment, society and the economy (Global Reporting Initiative, 2015). These guidelines enable organizations to generate standardized, reliable and relevant information for not only within the business but for its stakeholders as well (2015).

Research shows that organizations tend to use and emphasize positive information in voluntary sustainability reports (Lougee & Wallace, 2008; Holder-Webb L. , Cohen, Nath, & Wood, 2009). According to Hahn and Lülfs (2014) these reports are prone to ambiguity and arbitrariness due to the lack of regulatory guidelines. The Global reporting Initiative tries to solve this problem by providing organizations a framework and guidelines for sustainability reporting. In May 2013, the fourth generation (G4) of the GRI Guidelines was launched. The aim of the new generation guidelines is to help reporters prepare sustainability reports that matter, to make robust and purposeful sustainability reporting standard practice (Global Reporting Initiative, 2015). The ‘auditability’ of the information is one of the principles considered essential by the GRI guidelines in order to produce a balanced and reasonable report and will positively affect the credibility of the information (O'Dwyer & Owen, 2005). The initiative challenges companies to report on the positive and negative contributions in order to come to a reasonable assessment of overall performance (Hahn & Lülfs, 2014). In order to categorize the disclosed non-financial information, the Global Reporting Initiative introduced six categories of non-financial indicators. These indicators are: economic, environment, labor, human rights, product responsibility and society. Furthermore, the G4 provide organizations a set of required disclosures such as Strategy and analyses, Organizational profile, Governance, and Stakeholder engagement (GRI, 2015).

2.5.1.2 AccountAbility Standards 2008 (AA1000APS)

The AccountAbility Standards 2008, also called the AA1000APS, is a well-known framework that provides guidance for evaluating the adherence of a reporting organization to different principles. Furthermore, this set of principles does also give some guidance about the reliability of associated performance information (EY, 2009). The AA1000 Assurance Standard of 2008 provides a comprehensive way of holding a company accountable for its management, performance and reporting on non-financial information issues.

(18)

This standard evaluates if a specific company adhered to the AccountAbility principles and the reliability of the performance information (AccountAbility, 2008a). The AccountAbility principles are intended to help organizations develop an strategic and accountable approach to sustainability which is internationally accepted and provide a structure to understand, govern, administer, implement, evaluate and communicate the organizations accountability (AccountAbility, 2008a). In order to realize the accountability, the framework is based upon three principles inclusivity, materiality and responsiveness.

The inclusivity principle is the foundation of the framework and is needed in order to achieve materiality and responsiveness. This principle states that it is necessary to include the participation of stakeholders in order to develop and achieve an accountable and strategic sustainable organization. Inclusivity is the commitment to be accountable to whom the organization has an impact on and requires a defined process of engagement and participation (AccountAbility, 2008b). According to the standard it is important for an organization to have an extensive understanding/knowledge about who the stakeholders of the organization are in particular and their needs and concerns. In order for an organization to adhere to the principle of inclusivity, it needs to adhere to a selection of criteria defined by the AccountAbility Standard 2008 (2008b).

The materiality principle is, according to the standard, the relevance and significance of an issue to an organization and the stakeholders (AccountAbility, 2008b). In order to assess the sustainability performance, the organization needs to make decisions that are useful for the organization and its stakeholders. The AccountAbility Standard 2008 provides organizations a materiality process which helps organizations to determine what is material. This process is designed in such a way that the right information (comprehensive and balanced) is used as an input and analyzed. The analysis takes sustainability drivers in consideration. Furthermore, it takes the needs, concerns and expectations of the organization and the stakeholders into account. In this way the organization has an comprehensive understanding of their context regarding sustainability and the related material issues (2008b).

The principle of responsiveness, “an organization shall respond to stakeholder issues that affect its performance” (2008b), is about the fact that an organization needs to demonstrate how it responds to the stakeholders and is related to in example establishing policies, objectives, targets, management processes and stakeholder engagements. An organization needs to communicate these responses in such a way that it meets the needs and expectations of the stakeholders (2008b).

(19)

In order to provide assurance to an organization which developed a strategic and accountable approach to report the sustainability, the AccountAbility developed an assurance standard. This standard seeks to assure its stakeholders the credibility and quality of sustainability performance reporting. There are two different types of assurance engagements regarding the AA1000 Assurance Standard for sustainability reporting, type 1 and type 2 (AccountAbility, 2009). With a type 1 assurance engagement, the sustainability information from the organization is taken into consideration by the assurance provider when an organization evaluated regarding the adherence to the principles. The verification and the reliability of the information, however, is not included in the type 1 assurance engagement. With a type 2 assurance engagement the assurance provider does not only give assurance regarding the principles but also verifies the reliability of the sustainability information by reviewing the underlying systems, processes and data (2009).

2.6 Cost of capital 2.6.1 Cost of equity

A lot of research has been done about voluntary disclosure and the cost of equity and the payoff function of equity holders (Botosan, 1997, Dhaliwal, Zhen Li, Tsang, & George Yang, 2011, Diamond & Verrecchia, 1991, Botosan & Plumlee, 2002). Dhaliwal, et al. (2011) finds that the likelihood of a firm initiating in reporting of sustainability activities is associated with a higher prior year cost of equity capital. Furthermore, organizations that disclose CSR information attract dedicated institutional investors and analyst coverage (2011). The underlying theoretical assumption is that organizations who provide more corporate information reduce the information asymmetry on the capital markets (Petrova, Georgakopoulos, Sotiropoulos, et al. 2012). This assumption has two aspects, one is related to the fact that organizations voluntarily provide more information in order to try to overcome the unwillingness of investors to hold shares in an illiquid stock. The second aspect is related to the fact that organizations try to reduce the estimation risk of stakeholders by (voluntarily) disclosing more information.

Investors can base their estimates of return on information that is publicly available. Investors can better estimate share returns when an organization voluntarily discloses more information about their operating activities. Given the convex payoff function of equity holders, the potential benefit of the increasing value of an organization is unlimited.

(20)

2.6.2 Cost of debt

According to Barath, et al. (2008) debt financing is a predominant source of external funding for organizations in the US. Debt contracts have multiple contract terms like interest, maturity and collateral (Bharath, Sunder, & Sunder, 2008). Debt holders can influence the relationship of risk return by setting terms for the interest costs, maturity term and collateral required in the contract. Furthermore, debtholders prefer less risky return distributions because of the fact that they have a concave payoff function (Jarrow, Maksimovic, & Ziemba, 1995). Francis, et al. (2005) defines the cost of debt as a weighted average of the interest costs arising from outstanding debt that have been issued in different economic and accounting environments. According to Ye and Zang (2011), it is critical for the financial success of an organization to have access to low-cost debt capital. Furthermore, debt financing can have additional advantages for organizations for example, the tax deductibility of interest costs (Fama & French, 2005).

The economic nature of a debtholder and an equity holder differs according to Kanda (1992). Debtholders want to obtain a safe investment environment after the debt contract has been signed with a debtor with the use of specific types of terms and conditions. This is possible for debtholders in response to the risk sharing and alternation problems that arise with debt financing (Kanda, 1992). The payoff function of debtholders is based upon a fixed claim and debt holders do not benefit from the upside potential of the increasing value of the organization.

However, debtholders do face the risk of a decrease in their investment which can be unexpected (Kanda, 1992). This principle is confirmed by the work of Merton (1973) where he states that the payoff function of a corporate bond is asymmetric and is similar to the payoff function of a put option. The potential benefit of a debtholder does not exceed the interest payments and the potential losses can be as much as the outstanding debt. In contrast, the benefits of the equity/shareholder are potentially unlimited. It is, therefore, more important for debtholders compared to equity holders to identify competent and responsible managers in order to reduce agency and monitoring costs (Hoepner, Oikonomou, Scholtens, & Schröder, 2014).

Debt markets are broadly divided into a dispersed public debt bond market and a private debt bank loan market (Bharath, Sunder, & Sunder, 2008). The two providers of debt, banks and bondholders, significantly differ in their ability of processing information and renegotiating debt contracts (2008). According to Krishnaswami (1999), most private debtholders are institutions such as banks and life insurance companies. These private lenders specialize in complicated credit evaluations, which are performed before issuing a debt and in monitoring the performance

(21)

of organizations after a debt placement. Private debt is the most important funding source for small organizations because of the fact that their access to public markets is limited by the high flotation and other costs that are related to issuing public debt (Krishnaswami, Spindt, & Subramaniam, 1999). As a result, organizations that are subject to information problems tend to use private debt. Goss and Roberts (2010) find a statistically significant but economically modest reaction to the disclosure of CSR and sustainability activities on the cost of bank loans measured by the all-in-drawn spread over the London Interbank Offer Rate (LIBOR) (2010). However, Carey, at al. (1993) conclude that private debt is also an important source of funding for large organizations who held and publicly traded securities. This implies that private debt, in comparison to public debt, might provide benefits other than lower issuance costs.

(22)

3 Hypothesis

3.1 Hypothesis development

Voluntary disclosure of non-financial information demonstrates the fact that firms have confidence in their non-financial performance. This sends a positive message to the stakeholders (i.e. debt holders) when the performance of the firm in question is good, or, when the performance is not good, it gives the firm a possibility to provide an explanation (Dhaliwal, Zhen Li, Tsang, & George Yang, 2011). Therefore, the voluntary disclosure of non-financial information provides stakeholders information that is beyond of that contained in (financial) performance ratios (Dhaliwal, et al., 2011). Frankel, et al. (1995) find in their research that companies increase their level of voluntary disclosure in order to raise capital in the future against a lower cost. This implies that firms with a high cost of capital have a greater incentive to voluntary disclose information. In addition, organizations that engage in sustainability reporting are considered to be more transparent. This reduces the cost of equity and might have an impact on the cost of debt as well (Dhaliwal, Zhen Li, Tsang, & George Yang, 2011).

Public and private debtholders differ significantly in the way they process information (including information that is related to sustainability activities) and renegotiating debt contracts (Bharath, Sunder, & Sunder, 2008). Denis and Mihov (2003) argue that private lenders are more efficient and effective monitors than public investors. Hence, organizations with a high degree of information asymmetry will prefer private debt and organizations with lower information asymmetry will borrow publicly (Denis & Mihov, 2003). In contrast, Diamond (1991) and Rajan (1992) hypothesize that the cost of bank monitoring outweigh the benefits for low-quality firms. As a result, these organizations borrow publicly. Goss and Roberts (2010) find that from the perspective of a bank (as a quasi-insider), risks related to CSR concerns are taken into consideration and often banks respond with less attractive loan contract terms. They find a significant but economically modest reaction to CSR activities. This finding suggests that banks view sustainability activities as a second-order determinant of spreads and can result in a lower cost of private debt (2010). The study of Goss and Roberts (2010) focusses on bank loans, from a perspective of a bank, in order to determine the cost of debt.

Private debtholders have access to private (non-) financial information which can be used for designing the terms of a loan contract and monitoring them. Furthermore, private debt contracts are easier to renegotiate compared to public debt. As a result, banks can provide a

(23)

2008). This results in the fact that private debt has a significant advantage over public debt and this might lead to lower private debt costs (Denis & Mihov, 2003). However, Rajan (1992) states that borrowers can also be negatively affected by private lenders by extracting rents and distorting management incentives which can result in higher private debt costs.

Based on the fact that private lenders are more efficient and effective monitors than public debtholders combined with the importance of the information environment of a firm, I expect that the effect of voluntary disclosure of non-financial information on the cost of debt is stronger for firms with private debt than for firms with public debt. Accordingly, I state the following hypothesis:

H1: The association between the cost of debt and voluntary disclosure of non-financial information is stronger for firms with private debt than for firms with public debt.

(24)

4 Sample and Empirical Design

4.1 Sample

In order to compare the cost of public and private debt of a company in relation to the disclosure of non-financial information I use two samples of US firms who report on CSR and sustainability activities using data from the MSCI ESG database, the former KLD STATS (Statistical Tool for Analyzing Trends in Social and environmental performance) database. This is a database that provides detailed performance ratings for CSR reports given by the Risk Metrics Group (RMG) and is available through the Wharton Research Data Services (WRDS).

Furthermore, I use data from the Compustat1 database in order to determine the cost of

public and private debt. To develop the sample of organizations that voluntarily disclose non-financial information, I start with the MSCI ESG database and merge the data with the data from the Compustat database. The sample of the MSCI ESG database contains 34,383 observations for the fiscal years 2003 until 2013. I limit my sample to the years 2003 until 2013 in order to get the best database coverage. After merging the sample of the MSCI ESG database with the sample of the Compustat database, 30,437 observations are left. After removing the missing data the first sample, related to the cost of public debt, is made up of 3,648 observations of 1,003 distinct firms from 2003 until 2013, respectively. The second sample, related to the cost of private debt, is made up of 20,403 observations of 3,774 distinct firms from 2003 until 2013, respectively.

4.2 Empirical design

Hypothesis 1 examines the effect of voluntary disclosure of non-financial information on the cost of public and private debt. This study measures the level of difference in the cost of public and private debt over time related to the voluntary disclosure of non-financial information rather than measuring the initial factor that organizations report CSR activities for the first time and have not reported any CSR or sustainability activities before. In order to test my hypothesis I use an empirical model based upon the models of Dhaliwal, et al. (2011), Francis, et al. (2005), Lin,

1 Compustat is a database of financial, statistical and marketing information and provides Income statement, Balance

(25)

et al. (2013), Krishnaswami, et al. (1999) and Attig, et al. (2013). I use the model of Dhaliwal, et al. (2011) as a basis for the two individual regressions of the cost of public and private debt in order to compare the effect of voluntary disclosure of non-financial information on the cost of public and private debt.

Regression 1: Cost of public debt

In order to determine the effect of voluntary disclosure on the cost of public debt, I use the following regression formula:

∆%𝐶𝑂𝐷_𝑃𝑈𝐵𝐿𝐼𝐶(𝑖,𝑡) = 𝛽0+ 𝛽1𝐷𝐼𝑆𝐶(𝑖,𝑡)+ 𝛽2𝐹𝐼𝑅𝑀_𝑆𝐼𝑍𝐸(𝑖,𝑡)+ 𝛽3𝐿𝐼𝑇(𝑖,𝑡)+ 𝛽4𝑅𝑂𝐴(𝑖,𝑡) + 𝛽5𝐿𝐸𝑉(𝑖,𝑡)+ 𝛽6𝐵𝐼𝐺4(𝑖,𝑡)+ 𝛽7𝐿𝑂𝑆𝑆(𝑖,𝑡)+ 𝜀(𝑖,𝑡)

Equation 1: Regression Model Cost of Public Debt

Where:

COD_PUBLIC is the cost of public debt, calculated as the total cost of debt multiplied by the percentage of debt that was publicly issued.

DISC is the overall measure of social performance defined as the total of the MSCI ESG Score of Corporate Social Responsibility Strengths minus the concerns.

FIRM_SIZE is a control variable for firm size, calculated as the natural logarithm of the market value of equity at the beginning of each year.

LIT is a control variable related to the anticipation of potential lawsuits against the organization. This is an indicator variable that equals 1 if the firm operates in a high-litigation industry.

ROA the total return on assets measured as the ratio of income before extraordinary items over total assets at the end of each year

LEV is defined as the ratio of total debt divided by total assets and is a control variable to control for default risk.

BIG4 is a dummy variable, set to 1 if the organization hired a Big 4 auditor, 0 otherwise. LOSS is a dummy variable, set to 1 if the net income before extraordinary items is negative in the current year, and otherwise 0.

(26)

Regression 2: Cost of private debt

In order to determine the effect of voluntary disclosure on the cost of private debt, I use the following regression formula:

∆%𝐶𝑂𝐷_𝑃𝑅𝐼𝑉𝐴𝑇𝐸(𝑖,𝑡) = 𝛽0+ 𝛽1𝐷𝐼𝑆𝐶(𝑖,𝑡)+ 𝛽2𝐹𝐼𝑅𝑀_𝑆𝐼𝑍𝐸(𝑖,𝑡)+ 𝛽3𝐿𝐼𝑇(𝑖,𝑡)+ 𝛽4𝑅𝑂𝐴(𝑖,𝑡)

+ 𝛽5𝐿𝐸𝑉(𝑖,𝑡)+ 𝛽6𝐵𝐼𝐺4(𝑖,𝑡)+ 𝛽7𝐿𝑂𝑆𝑆(𝑖,𝑡)+ 𝜀(𝑖,𝑡)

Equation 2: Regression Model Cost of Private Debt

Where:

COD_PRIVATE is the cost of private debt, calculated as the cost of total debt minus the cost of public debt.

DISC is the overall measure of social performance defined as the total of the MSCI ESG Score of Corporate Social Responsibility Strengths minus the concerns.

FIRM_SIZE is a control variable for firm size, calculated as the natural logarithm of the market value of equity at the beginning of each year.

LIT is a control variable related to the anticipation of potential lawsuits against the organization. This is an indicator variable that equals 1 if the firm operates in a high-litigation industry.

ROA the total return on assets measured as the ratio of income before extraordinary items over total assets at the end of each year

LEV is defined as the ratio of total debt divided by total assets and is a control variable to control for default risk.

BIG4 is a dummy variable, set to 1 if the organization hired a Big 4 auditor, 0 otherwise. LOSS is a dummy variable, set to 1 if the net income before extraordinary items is negative in the current year, and otherwise 0.

(27)

4.2.1 Main variables

4.2.1.1 Independent variable: CSR performance

In order to measure the overall social performance of an organization I use data from the MSCI ESG database. The MSCI ESG database is often used in research related to the disclosure of non-financial information (Baron, Harjoto, & Jo, 2009). According to Waddock (2003), it is “The research standard at the moment”. The MSCI ESG database ranks the performance of a firm in thirteen categories, namely: (1) Products, (2) Environment, (3) Corporate Governance, (4) Employee Relations, (5) Human Rights, (6) Community, (7) Diversity, (8) Alcohol, (9) Gambling, (10) Firearms, (11) Military, (12) Tobacco and (13) Nuclear Power. On the first seven dimensions, strengths and concerns are measured by MSCI ESG with an performance indicator. Regarding the last six dimensions, companies can only register a concern. In this study, I will only use the first seven dimensions as the last six are considered exclusionary categories. In order to research the effect of voluntary disclosure of non-financial information on the cost of debt, I will use data from US firms as this is the main category of the MSCI ESG database. The database provides a binary summary of positive and negative Environmental, Social, and Governance (ESG) ratings. If the RMG assigned a rating to the category (this can be positive or negative), this is indicated with a 1 in the corresponding cell. If the organization did not have a strength or concern in that particular category, this is indicated with a 0 (RMG, 2010).

For each area MSCI ESG assigns a binary (0/1) rating to a set of concerns and strengths. Following the empirical model of Attig, et al. (2013) I calculate a score for each qualitative issue area equal to the number of strengths minus the number of concerns. I sum the scores related to the issues in the qualitative areas to obtain an overall CSR score (DISC) in order to estimate the effect of voluntary disclosure of non-financial information on the cost of public and private debt. The estimated variable DISC is the key independent variable in my research.

4.2.1.2 Dependent variable regression 1: Cost of public debt

My study focuses on the effect of voluntarily disclosed non-financial information on the cost of public and private debt. In order to estimate the cost of public debt of a firm, I need to estimate the amount of public debt first (equation 3). I define public debt as debt that is publicly traded with a maturity greater than one year (Krishnaswami, Spindt, & Subramaniam, 1999). Based

(28)

upon the model of Lin, et al. (2013) I measure public debt as the sum of senior bonds and notes (SN_BOND), subordinated bonds and notes (SB_BOND), and commercial papers (C_BOND).

𝑃𝑈𝐵𝐿𝐼𝐶_𝐷𝐸𝐵𝑇(𝑖,𝑡) = 𝑆𝑁_𝐵𝑂𝑁𝐷(𝑖,𝑡)+ 𝑆𝐵_𝐵𝑂𝑁𝐷(𝑖,𝑡)+ 𝐶_𝐵𝑂𝑁𝐷(𝑖,𝑡)

Equation 3: Total amount of Public Debt

Where:

PUBLIC_DEBT is the implied amount of public debt, as defined earlier.

SN_BOND represents the total amount of senior bonds and notes for firm i in year t. SB_BOND represents the total amount of subordinated bonds and notes for firm i in year t.

C_BOND represents the total amount of commercial papers for firm i in year t.

I derive the cost of public debt from the total cost of debt (equation 4). Francis, et al. (2005) defines the total cost of debt as a weighted average of the interest costs arising from outstanding debt that have been issued in different economic and accounting environments. Based upon the model of Francis, et al. (2005) and Krishnaswami, et al. (1999) I define the dependent proxy cost of public debt (COD_PUBLIC) as the total cost of debt multiplied by the percentage of debt that was publicly issued. The total cost of debt is defined as the ratio of the firms’ interest expense in year t + 1 (INT) to average interest-bearing debt (AVG_IDEBT) outstanding during the years t and t + 1. The cost of publicly issued debt is defined as described before: 𝐶𝑂𝐷_𝑃𝑈𝐵𝐿𝐼𝐶(𝑖,𝑡)= ( 𝐼𝑁𝑇(𝑡+1) 𝐴𝑉𝐺_𝐼𝐷𝐸𝐵𝑇(𝑡,𝑡+1) ) × ( 𝑃𝑈𝐵𝐿𝐼𝐶_𝐷𝐸𝐵𝑇(𝑖,𝑡) 𝑇𝑂𝑇𝐴𝐿_𝐷𝐸𝐵𝑇(𝑖,𝑡) )

Equation 4: Cost of Public Debt

(29)

INT is the interest expense in year t + 1.

AVG_IDEBT is the average interest-bearing debt that is outstanding during the years t and t + 1.

PUBLIC_DEBT is the implied total of public debt, as defined earlier. TOTAL_DEBT is the total debt that is outstanding during the year t.

4.2.1.3 Dependent variable regression 2: Cost of private debt

Based upon the model of Francis, et al. (2005) and Krishnaswami, et al. (1999) I define the dependent proxy cost of private debt (COD_PRIVATE) as the total cost of debt minus the cost of debt that was publicly issued. The total cost of debt and the cost of public debt are defined before:

𝐶𝑂𝐷_𝑃𝑅𝐼𝑉𝐴𝑇𝐸(𝑖,𝑡) = ( 𝐼𝑁𝑇(𝑡+1) 𝐴𝑉𝐺_𝐼𝐷𝐸𝐵𝑇(𝑡,𝑡+1)

) − (𝐶𝑂𝐷_𝑃𝑈𝐵𝐿𝐼𝐶(𝑖,𝑡))

Equation 5: Cost of Private Debt

Where:

COD_PRIVATE is the implied cost of private debt, as defined earlier. INT is the interest expense in year t + 1.

AVG_IDEBT is the average interest-bearing debt that is outstanding during the years t and t + 1.

COD_PUBLIC is the cost of public debt, as defined earlier.

4.2.2 Control variables

To determine the effect of voluntary disclosed non-financial information on the cost of public and private debt, I include a number of control variables in order to isolate the effect of sustainability and CSR activities. Following the model of Dhaliwal, et al. (2011) and Attig, et al. (2013) I include a control variable for firm size (FIRM_SIZE). Lang and Lundholm (1993) argue

(30)

that firm size include various factors that motivate firms to disclose CSR reports, for example financial resources and public pressure. I measure firm size as the natural logarithm of the market value of equity at the beginning of each year (Dhaliwal, Zhen Li, Tsang, & George Yang, 2011). According to Skinner (1997) firms are more likely to engage in voluntary disclosure of information in order to anticipate on potential lawsuits. Therefore, I control for litigation (LITIGATION) with the use of an indicator variable that equals 1 if an organization operates in an industry with high-litigation and 0 otherwise. Organizations that operate in a high-litigation industry is determined by the SIC codes of 2833-2836 (Medicinal Chemicals and Botanical Products), 3570-3577 (Computer Technology), 3600-3674 (Electronics), 5200-5961 (Retailing) and 7370 (Computer-programming services) based upon the research of Francis, Philbrick, and Schipper (1994) and Matsumoto (2002). Organizations that have a better financial performance are likely to have more resources at their disposal to engage in sustainability and CSR activities and produce reports (Dhaliwal, Zhen Li, Tsang, & George Yang, 2011). I include the Return on Assets (ROA) in order to control for this effect and is calculated as income before extraordinary items divided by total assets at the end of each year. I control for default risk (LEVERAGE) with the use of the ratio of long term debt to total assets (Attig, Ghoul, Guedhami, & Suh, 2013). According to Sengupta (1998) the lower the risk of default, the lower the cost of borrowing for an organization. Furthermore, this study examines the cost of public and private debt across different industries. Following the model of Lin, et al. (2013) I control for year effects and I include the Fama and French’s 48 industry indicators in my regression model in order to control for possible industry differences (Fama & French, 1997).

I include two dummy variables. First, the dummy indicator variable BIG4. If the organization is audited by a Big 4 auditor this indicator variable is set to 1, and otherwise 0. Second, I include the dummy variable LOSS. This indicator variable is set to 1 if the net income before extraordinary items is negative in the current year, and otherwise 0 (Attig, Ghoul, Guedhami, & Suh, 2013). This dummy variable is included because of the fact that organizations report more information voluntarily as a result of a loss which occurred during the year in order to provide an explanation to convince the stakeholders..

4.2.3 Hypothesis testing

In order to determine the effect of voluntary disclosure of non-financial information on the cost of public and private debt, I regress the cost of debt on non-financial information with multiple

(31)

debt and the cost of private debt, taking linearity of the dependent and independent variables into account. In order to ensure that the regression is valid, I check if the residuals of both regressions are normally distributed, linear and independent. I use a pairwise correlation matrix in order to check for multicollinearity among the variables after winsorizing the variables at the 1st and 99th percentile in order to extract the outliers from the sample and taking normality and

linearity of the sample distribution into account. Furthermore, I check the homoscedasticity of the residuals with the use of the White’s test and the Breush-Pagan test. I combine these tests with a residual versus fitted plot because of the fact that the tests are very sensitive to model assumptions (Field, 2013). Furthermore, in order to prevent the fact that one or more variables are omitted or irrelevant (model specification errors) I perform a link test on both regressions.

After running the robust and clustered regressions individually, I run the regression again without the two options. I store the estimation results of the two regression models into two separate variables. With the use of a seemingly unrelated estimation (SUEST) regression I combine the two estimation results into one parameter vector in order to compare the regression coefficient of the Average Disclosure Score (DISC) variable computed from the two different models with the use of a Chi2 test. I use a seemingly unrelated estimation regression because of

the fact that it is possible to run this estimation regression on two different samples. The estimation results are estimated without using a robust or clustered option as this option is prohibited in the base regression models when using a seemingly unrelated estimation regression (SUEST automatically returns the robust variance estimators). Thus the standard errors reported in the original regression are potentially flawed (Hausman & Ruud, 1987). In order to correct for this problem I specify the cluster and robust option when using the seemingly unrelated estimation regression. Specifically, in order to confirm my hypothesis, I expect a larger effect of my main variable of interest (β1), the coefficient of the overall measure of CSR performance

(32)

5 Results

5.1 Descriptive statistics

Table 1 provides the distribution of the observations of the two samples used in my study sorted by year from the period of 2003 until 2013. Descriptive statistics show that the two samples are equally distributed among the years, respectively.

Table 1: Number of observations per fiscal year

Fiscal year Cost of Public Debt Cost of Private Debt

N % N % 2003 345 9.46 1,864 9.14 2004 381 10.44 1,918 9.40 2005 343 9.40 1,729 8.47 2006 333 9.13 1,703 8.35 2007 335 9.18 1,786 8.75 2008 340 9.32 1,831 8.97 2009 320 8.77 1,781 8.73 2010 333 9.13 1,921 9.42 2011 314 8.61 1,867 9.15 2012 218 7.70 1,868 9.16 2013 323 8.85 2,135 10.46 Total 3,648 100 20,403 100

Table 2 provides the descriptive statistics related to the Average Disclosure Score over the sample period of the cost of public debt. The variable Average Disclosure Score shows a non-negligible variation over time with a minimum of -10 to a maximum of +15. In addition, the overall median is -1, suggesting that the distribution of the Average Disclosure Score is slightly negative. This implicates that the firms included in the sample period for the cost of public debt report slightly more CSR concerns than CSR strengths on average. The results related to the

This table provides the distribution of the number of observations per fiscal year. The first sample is related to the cost of public debt and consists of 3,648 observations of 1,003 unique firms. My second sample is related to the cost of private debt and consists of 20,403 observations of 3,774 unique firms. Both samples are from the period of 2003 until 2013.

(33)

average disclosure score for the cost of public debt sample are in line with the results of the research of Attig, et al. (2013).

Table 2: Cost of public debt: Descriptive statistics for the variable DISC

Years Mean Min Q1 Median Q3 Max SD

2003 -.484058 -9 -1 0 0 9 2.000118 2004 -.67979 -9 -2 -1 0 9 2.152682 2005 -.7959184 -8 -2 -1 0 9 2.24762 2006 -1.081081 -8 -2 -1 0 11 2.340587 2007 -.9940299 -8 -2 -1 0 13 2.333041 2008 -1.055882 -10 -2 -1 0 13 2.301685 2009 -.9875 -8 -2 -1 0 13 2.358171 2010 -.7297297 -8 -2 -1 0 14 2.750507 2011 -.8980892 -7 -3 -2 0 15 3.251315 2012 .6405694 -5 -1 0 1 8 2.214157 2013 1.095975 -6 -1 0 2 12 3.04318 Total -.5641447 -10 -2 -1 0 15 2.560215

Table 3 provides the descriptive statistics related to the Average Disclosure Score over the sample period of the cost of public debt. The variable Average Disclosure Score shows a non-negligible variation over time with a minimum of -11 to a maximum of +19. In addition, the overall median is -1, suggesting that the distribution of the Average Disclosure Score is slightly negative. This implicates that the firms included in the sample period for the cost of private debt report slightly more CSR concerns than CSR strengths on average. The results related to the average disclosure score for the cost of private debt sample are in line with the results of the research of Attig, et al. (2013).

This table provides the mean, minimum, first quartile, median, third quartile, maximum and the standard deviation of the variable Average Disclosure Score (DISC) score by fiscal year. The sample is related to the cost of public debt and consists of 3,648 observations of 1,003 distinct firms from 2003 until 2013. ‘Appendix B’ provides the variable definitions presented in this table.

(34)

Table 3: Cost of private debt: Descriptive statistics for the variable DISC

Years Mean Min Q1 Median Q3 Max SD

2003 -.2280043 -11 -1 0 1 9 1.776584 2004 -.4827946 -10 -1 0 0 10 1.895955 2005 -.5361481 -9 -2 -1 0 10 2.051139 2006 -.6300646 -10 -2 -1 0 12 2.197764 2007 -.6556551 -9 -2 -1 0 15 2.206081 2008 -.6302567 -10 -2 -1 0 14 2.239814 2009 -.6709714 -10 -2 -1 0 14 2.249992 2010 -.7636648 -8 -2 -2 0 17 2.727054 2011 -1.017675 -8 -3 -2 0 19 3.290704 2012 .8501071 -6 0 0 2 16 2.335313 2013 1.229977 -7 -1 0 3 17 3.109685 Total -.2952507 -11 -2 -1 1 19 2.527308

Table 4 presents the sample distribution sorted by industry based on the Fama and French’s (1997) industry classification. Descriptive statistics show that the two samples are equally distributed among industries, respectively, with the exceptions of the Insurance industry (11.81%) related to the cost of private debt sample and the Banking (12.02%) and the Trading industry (12.54%) related to the cost of public debt sample.

This table provides the mean, minimum, first quartile, median, third quartile, maximum and the standard deviation of the variable Average Disclosure Score (DISC) score by fiscal year. The sample is related to the cost of private debt and consists of 20,403 observations of 3,774 distinct firms from 2003 until 2013. ‘Appendix B’ provides the variable definitions presented in this table.

(35)

Table 4: Descriptive statistics for the samples sorted by Industry

Industry Cost of Public Debt Cost of Private Debt

N % N %

Agriculture 7 0.19 73 0.36

Food Products 41 1.12 291 1.43

Candy & Soda 0 0.00 47 0.23

Beer & Liquor 12 0.33 54 0.26

Tobacco Products 4 0.11 34 0.17

Recreation 16 0.44 69 0.34

Entertainment 133 3.65 265 1.30

Printing and Publishing 6 0.16 139 0.68

Consumer Goods 30 0.82 244 1.20 Apparel 26 0.71 191 0.94 Healthcare 94 2.58 305 1.49 Medical Equipment 73 2.00 370 1.81 Pharmaceutical Products 173 4.74 733 3.59 Chemicals 73 2.00 462 2.26

Rubber and Plastic Products 22 0.60 102 0.50

Textiles 0 0.00 35 0.17

Construction Materials 44 1.21 324 1.59

Construction 87 2.38 272 1.33

Steel Works Etc. 74 2.03 253 1.24

Fabricated Products 5 0.14 19 0.09

Machinery 86 2.36 673 3.30

Electrical Equipment 29 0.79 213 1.04

Automobiles and Trucks 36 0.99 237 1.16

Aircraft 28 0.77 132 0.65

Shipbuilding, Railroad Equipment 5 0.14 44 0.22

Defense 10 0.27 34 0.22

Precious Metals 14 0.38 43 0.21

Non-Metallic and Industrial Metal Mining 17 0.47 93 0.46

Coal 20 0.55 59 0.29

Petroleum and Natural Gas 261 7.15 878 4.30

Utilities 6 0.16 807 3.96

Communication 124 3.40 493 2.42

Referenties

GERELATEERDE DOCUMENTEN

Due to the fact that previous studies mostly focused on voluntary reporting, previous research on the Directive was only focused on the potential contribution of the Directive

If the evidence gathered for question 2 shows that the decision-usefulness theory can be applied in setting standard on NFI in general, but the users, uses or criteria for choices

Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright

Pre-S&OP and S&OP meeting: consideration and comparison of different risk- treatment options based on financial implications; decisions depending on the cost of measures –

Nou dat die volk uit die mond van die Britse min i ster self vcrneem hct wat presics die Britse konneksies, kan daar miskicn met grotcr nut gckon- sentreer word op

As mentioned before, a negative relationship between auditor independence and cost of debt is expected since prior literature found for example a negative and significant

Approach/Design/Methodology/Purpose – Building on signalling and legitimacy theory and drawing from literature on corporate transparency and information disclosure, this study uses

Furthermore, we reveal that the weight of bank credit in the financing of the private sector affects the debt maturity of multinationals less negatively compared