The impact of Integrated Reporting on the debt financing of firms
An empirical study examining the association between Integrated Reporting and the cost of public and private debt capital.
Name: Ka-Wai Tang Student number: 11121165
Thesis supervisor: Ms. dr. R. Felleg Date: 26-06-2017
Word count: 12,333
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam
Statement of Originality
This document is written by student Ka-Wai Tang 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.
Abstract
This study examines the relationship between Integrated Reporting (IR) and the cost of public and private debt capital. Specifically, this study examines whether the effect of IR on the public debt market is stronger than the private debt market. The motivation for this study is due to the growing interest of firms, investors and regulators in IR, the important role of debt markets, the limited research on and limited knowledge of IR, as well as calls from academics to further investigate IR. My results suggest that IR does have a significant effect and it decreases the cost of debt capital of firms. However, I did not find significant evidence that this effect is stronger in the public debt market. This study provides empirical evidence regarding the association between IR and the cost of public and private debt capital which could encourage leveraged firms to do IR. Managers in leveraged firms could use this information to further invest in IR and to manage their cash flows and debt. Also, my study shows the relevance of IR in the debt market which could assist leveraged firms in financing decisions such as obtaining an optimal capital structure.
Contents
1 Introduction ... 5
2 Literature review and hypotheses development... 8
2.1 The need for integrated reports ... 8
2.2 Integrated Reporting ... 9
2.3 Agency theory, information asymmetry, and signalling ... 10
2.4 Cost of capital ... 12
2.4.1 Cost of equity capital ... 12
2.4.2 Cost of public and private debt capital ... 12
3 Hypothesis development ... 14
4 Research methodology... 15
4.1 Sample description and data sources... 15
4.2 Research design ... 16
4.3 Variables ... 18
4.3.1 Independent variable: Integrated Reporting ... 18
4.3.2 Dependent variable: Cost of debt capital ... 18
4.3.3 Control variables ... 19
4.4 Analysis ... 20
5 Results... 21
5.1 Descriptive statistics ... 21
5.2 Results of hypothesis tests... 28
5.2.1 Regression results cost of public debt ... 28
5.2.2 Regression results cost of private debt ... 29
5.2.3 Seemingly Unrelated Estimation regression results ... 30
6 Conclusion ... 32
7 References ... 34
Appendix A: Variable definitions and data sources ... 39
1 Introduction
This study examines the effect of Integrated Reporting (IR) on the cost of capital in the public and private debt markets. This research question is motivated due to the growing interest of firms, investors and regulators in IR, the important role of debt markets, the limited research on and limited knowledge of IR, as well as requests from academics to further investigate IR (Sengupta, 1998; Bharath, Sunder, & Sunder, 2008; Cheng, Green, Conradie, Konishi, & Romi, 2014; De Villiers, Rinaldi, & Unerman, 2014; Serafeim, 2015; Barth, Cahan, Chen, & Venter, 2015).
The growing interest in integrated reports is caused by the inadequacy of Sustainability reports such as Corporate Social Responsibility (CSR) to communicate efficiently and effectively environmental, social, and governance information (ESG) to a diverse range of stakeholders who require this information for their investment decisions (Hutton, 2004; Vanstraelen, & Chua, 2009; IFAC, 2012; Cohen, Holder-Webb, Nath, & Wood, 2012; Cheng et al., 2014; EY, 2015). Because of this inadequacy, there were moves to combine financial and non-financial information into one integrated report to meet the information needs of a range of stakeholders by explaining how the organization creates value (Dey & Burns, 2010; IIRC, 2013). These moves have ultimately led to the creation of the <IR> framework. The framework is created by the International Integrated Reporting Council (IIRC) to standardize economic, environmental, social, and governance information into one report to enhance the credibility and comparability of financial and non-financial information (De Villiers et al., 2014). Investors agree that ESG information is more useful when it is standardized and combined with financial information into one report instead of stand-alone reports that disclose only ESG information (e.g. Sustainability reports and CSR reports) separate from the financial reports (EY, 2015).
Interest in IR has been increasing in the past few years because these reports are broader in scope than current CSR reports and other Sustainability reports (Eccles & Serafeim, 2014; EY, 2015; Barth et al., 2015). IR presents and communicates information in a concise and effective manner to stakeholders (Zhou, Simnett, & Green, 2017). Also, it integrates the resources of an organization with its strategy and business model to explain value creation over a short, medium, and long-term period (IIRC, 2013). By showing these linkages, investors understand how ESG performance relates to financial performance and how sustainability issues affect the creation of value in the organization (Eccles & Serafeim, 2014). This insight is value relevant information that potentially reduces the information asymmetry (Zhou et al., 2017). Furthermore, the incorporation of business models in IR enables a better risk assessment of the investment which improves capital allocation decisions of investors (Eccles & Serafeim, 2014; Barth et al., 2015; Nielsen & Roslender, 2015).
The debt market is the predominant source of new (long-term) external funding and it is economically important for firms globally (ECB, 2007; Bharath, Sunder, & Sunder, 2008; Lin, Ma, Malatesta, & Xuan, 2013). In addition, debt financing plays an important role for firms located in Europe due to their capital structure (ECB, 2007; Lin, Ma, Malatesta, & Xuan, 2013). Also, lenders in these markets have the power to influence the behaviour of firms by setting up contract terms (Scholtens, 2006). Furthermore, debt has additional advantages which makes it interesting for firms to finance their operations through debt (Fama & French, 2005).
Prior literature has extensively examined the relationship between CSR reporting and the cost of capital. Also, the differences between public debt and private debt have been examined extensively (Orlitzky, Schmidt, & Rynes, 2003; Sharfman & Fernando, 2008; Menz, 2010; Goss & Roberts, 2011). In contrast, the association between IR and the cost of debt capital has not been examined before. The study of Zhou et al. (2017) is, to my knowledge, the only study that examines the effect of IR on the capital market. Empirical evidence regarding the association between IR and the cost of debt capital remains to be seen. I extend the IR literature by examining the effect of IR in the debt market. Specifically, I examine the effect of IR on the cost of public and private debt capital. The debt market is interesting to examine because investors in these markets have a different perception of risk. Debtholders have limited downward and limited upward potential from their investments which leads to a more conservative approach in their decision-making process (Kanda, 1992). Within the debt market, it is interesting to examine the effect of IR in the public and private debt market because private debtholders are better in monitoring debtors than public debtholders. Compared to public debtholders, private debtholders do not necessarily rely on publicly disclosed information because they have better access to private information (due to their close relationship with the debtor). Also, they have superior information gathering and processing abilities. Thus, private debtholders have a lower information asymmetry and a different assessment of (business) risk than public debtholders. As a result, the effect of IR in the public and private debt market will be different (Bharath et al., 2008; Goss & Roberts, 2011). Therefore, in this study I examine the association between IR and the cost of public and private debt capital. I expect that, due to IR, the cost of public debt will experience a stronger decline than the cost of private debt.
To examine the relationship between IR and the cost of public and private debt capital, I follow the method of Serafeim (2015), and Baboukardos and Rimmel (2016) and use the ASSET4 database to determine IR. After extracting the data, I transform the data into dummy variables and include it as an independent variable in the cost of public and private debt capital regression models. For this study, I use two regression models, which are based on the models of Lorca,
Sánchez-Ballesta and García-Meca (2011) and Pittman and Fortin (2004), to measure the effect of IR on the cost of public and private debt capital. The two regression models have different samples and they relate to the period 2002 until 2015. The first sample is the public debt sample and it consists of 39,629 observations of 3,976 distinct firms. The second sample is the private debt sample which consists of 35,763 observations of 3,741 distinct firms. To examine the effect of IR on the cost of debt, I run a robust Ordinary Least Square (OLS) regression that is clustered by firm. After interpreting the data separately, I use these variables to run a Seemingly Unrelated Estimation (SUEST) regression and a Chi² test to compare the coefficient of the cost of public debt with the coefficient of the cost of private debt. I confirm that firms that issue IR experience a lower cost of debt capital and that this effect is stronger in the public debt market than in the private debt market. However, the Chi² indicates that this difference in effect is not significant. Based on these results, I conclude that the negative effect of IR on the cost of public debt capital is not stronger than the negative effect of IR on the cost of private debt capital.
To my knowledge there is only one academic study that examines IR and its association with the cost of capital. However, prior research concentrate mainly on the cost of equity capital. The relationship between IR and the cost of debt capital has not been examined before. My study examines IR and its relation to the capital market by providing empirical evidence on the effect of IR on the cost of public and private debt capital. Therefore, my study fills the gap in current literature and contributes to a better understanding of the claimed benefits of IR. My study has practical implications because the results are valuable for managers in leveraged firms. My study document that IR has a significant impact in lowering the cost of public and private debt capital which might encourage managers to do IR. These managers could do IR in order to manage their cash flows and/or debt by decreasing their cost of debt capital. Furthermore, my study shows the relevance of IR in the debt market. This is important because debtholders in the public and private debt market are fundamentally different from other stakeholders. Private debtholders are quasi-insiders with superior tools to evaluate debtors. Also, these debtholders function in a market that has less negative spillover effects (which results in a loss of competitive advantage because competitors could benefit from this information) than the public debt market. As a result, providing empirical evidence regarding IR in the debt market might assist leveraged firms in financing decisions (debt placement decisions) to obtain an optimal capital structure.
The remainder of this study proceeds as follows. Chapter 2 presents theoretical explanations and a literature review. Chapter 3 discusses the hypothesis development. Chapter 4 provides the sample description and the empirical design of this study. Chapter 5 describes the descriptive statistics and results. Lastly, chapter 6 provides the conclusion of this study.
2 Literature review and hypotheses development
2.1 The need for integrated reports
Starting early in the 1990s, investment analysts witnessed a growing interest of firms disclosing ESG information to meet the expectations of their stakeholders. These stakeholders use ESG information to understand how an organization’s key ESG factors will impact its overall performance on a long-term basis which enables the stakeholder to gain trust in their investments (IFAC, 2012; DeutscheAWM, 2015). This form of transaction is also called the information function of corporate reporting. It is decision-useful information that aims to provide the investor assurance to transact with a firm (Eccles & Serafeim, 2014). Deutsche Asset & Wealth Management and the University of Hamburg issued a global meta-study in which they combine the findings of 2,200 individual studies on ESG and its relationship with financial performance. Their study show that the majority of academic studies found a positive relationship between ESG information and the financial performance of the firm (DeutscheAWM, 2015). The potential benefits obtained by disclosing ESG have increased the interests of financial institutions and other investors. Ernst & Young (EY) published the 2015 EY Global Institutional Investor Survey that reveals what investors want to know regarding corporate information. The report shows that ESG information is regarded as highly important. It became clear in the survey findings that the percentage of investors who consider the importance of non-financial information1 for all sectors, increased from 33.7% in 2014 to 61.5% in 2015. Also, the survey shows that investors in private and public markets, who see integrated reports as highly important or vital to investment decisions, increased from 61% in 2014 to 70.9% in 2015. These investors assess risk and return of their investments by focusing on ESG information that are tied to the operational performance. This means that ESG information impacts the cost of capital (Boutin-Dufresne & Savaria, 2004; EY, 2015).
Despite the growing interests and the acknowledgement of the usefulness of ESG information in CSR reports, investors agree that this information is more useful when it is standardized and combined into one report instead of disclosing ESG information in stand-alone reports (e.g. Sustainability reports and CSR reports) separate from the financial reports. Often these stand-alone sustainability reports are lengthy, complex and not standardized which create understandability and comparability issues. As a result, (external) stakeholders have difficulties to determine the materiality of information, to assess the financial performance, and to assess the risk
of the disclosing firm (KPMG & FERF, 2011; IFAC, 2012; Cheng et al., 2014; EY, 2015). Furthermore, the information disclosed in CSR reports are often not verified (Hutton, 2004; Simnett, Vanstraelen, & Chua, 2009; Cohen et al., 2012; EY, 2015). Even if they are verified, these reports usually get negative assurance instead of investor-useful (positive) assurance. This is due to the lack of rigorous measurement and reporting standards (Eccles & Serafeim, 2014). Also, these reports lack credibility, timeliness, and relevance because they are mostly published several months after the annual financial report (Eccles & Serafeim, 2014). As a response to the problems of CSR, there were moves to combine ESG information with financial information into a single report to meet the information needs of a range of stakeholders by explaining how the organization creates value (Dey & Burns, 2010; IIRC, 2013). To show their value creation, an organization needs to communicate their strategy, governance, performance and prospects in the context of its external environment over the short, medium and long term (IIRC, 2013). This integration of ESG information together with financial information, and its explanation regarding value creation, have ultimately led to the creation of the International Integrated Reporting Framework <IR> (Cheng et al., 2014).
2.2 Integrated Reporting
In 2010, the GRI and Prince’s Accounting for Sustainability Project jointly formed the International Integrated Reporting Council (IIRC) to develop and promote IR on a global level (De Villiers et al., 2014). According to The International Integrated Reporting Council (IIRC), an integrated report is “A concise communication about how an organization’s strategy, governance, performance and prospects, in the context of its external environment, lead to the creation of value over the short, medium and long term” (IIRC, 2013, p. 7). The aim of such an integrated report is to standardize, social, environmental, financial and governance information into one report to enhance the credibility and comparability which results in the improvement of the quality of information to providers of financial capital (IIRC, 2013; De Villiers et al., 2014).
The IIRC published the International <IR> Framework in 2013 to provide formal guidance to firms that want to do IR. The <IR> Framework clearly states that it promotes a cohesive and efficient approach to corporate reporting which enables efficient and productive allocation of capital by improving the information quality to providers of financial capital. Furthermore, it states that the framework benefits all stakeholders who are interested in an organization’s ability to create value (IIRC, 2013; De Villiers et al., 2014; Eccles & Serafeim, 2014).
The whole framework is based on integrated thinking in which the firm considers the relationship between operating and functional units and the capitals that the organization uses or affects (Eccles & Serafeim, 2014). Integrated thinking can be achieved when the decision-making of operating and functional units can create value over short, medium and long term. To do so, <IR> provides seven guiding principles and eight content elements. The most important are the capitals, which are Financial, Manufactured, Intellectual, Human, Social, and Natural (IIRC, 2013). These six capitals need to integrate with the strategy and business model to explain and communicate effectively the value creation of an organization (Cheng et al., 2014). Integrated reports are broader in scope than traditional corporate reports because it includes all capitals into one report and connects these capitals to value creation activities. It creates more content in the report that captures the strategy, the business model and future oriented information. Linking the capitals to business model and strategy of an organization enables stakeholders to assess where the value comes from, how the firm performs, and what the prospect of the organization is. As a result, it reduces the uncertainty in assessing the firm’s long-term performance by investors (Nielsen & Roslender, 2015; Zhou et al., 2017).
Overall, IR integrates financial and non-financial information into one report and addresses all resources and relationships that materially impact the value-creation activities (Barth et al., 2015; Cohen & Simnett, 2015).
2.3 Agency theory, information asymmetry, and signalling
Prior literature states that corporate disclosures are critical for the functioning of an efficient capital market. In the capital market, managers and investors match their savings to investment opportunities. The manager wants to receive capital to finance their operations and the investors want to allocate their resources efficiently. However, the information asymmetry and the agency problem impede this efficient allocation of resources (Healy & Palepu, 2001).
The information asymmetry is also called the lemons problem which could potentially lead to a breakdown of the functioning of the capital market (Akerlof, 1970). This problem arises from information differences between the manager and the investor. The managers are better able to value their business investment opportunities than the savers because they have more and better quality of information, which can be used to value their business investment opportunities. Information asymmetry creates incentives for the entrepreneur to overstate the value of their business investment opportunities. In addition, good and bad business investment opportunities
are difficult to distinguish because a rational investor would undervalue good investments and overvalue bad investments which impedes the efficient allocation of capital in the market (Healy & Palepu, 2001).
The agency problem arises after the capital allocation decision of the investor. These investors usually don’t play an active role within the firm. After the investment, the self-interested manager could expropriate the received funds by acquiring perquisites, pay excessive compensation, and make suboptimal investment decisions. These activities could harm the firm because the interest of these managers misalign with the interest of outside investors (Healy & Palepu, 2001).
Firms have several incentives to reduce the information asymmetry between the manager and the investor. One method is to disclose (private) information, which allows external parties to value and monitor the disclosing firm (Healy & Palepu, 2001). This method communicates the qualifications of the issuing firm and sends a credible signal to external parties. This is also known as the signalling theory (Morris, 1987). Prior research used this theory to explain job markets (Spence, 1973). According to Watson, Shrives and Marston (2002), the signalling theory is transferable to studies regarding voluntary disclosure. Signalling is used to decrease the information asymmetry between firms and investors. Especially, it is used to counter the lemons problem. High quality firms use signalling to distinguish themselves from low quality firms through voluntary disclosures (Watson et al., 2002). Another study shows that voluntary disclosure of CSR is used to signal the future prospects of a firm which investors can use to assess the risk of the firm (Lys, Naughton, & Wang, 2015).
Reducing information asymmetry contributes to the risk mitigation view of investors. The risk mitigation view argues that corporate reporting on environmental, social and governance may contribute to a firm by reducing the risk of the investor and the risk of the firm itself. Investors argue that screening socially responsible firms also screens the overall excellence of the management by looking at whether the firm invests optimally in social responsibility (Boutin-Dufresne & Savaria, 2004). This comprehensive information can be used by investors to assess the (idiosyncratic) risk2 of a firm (Nielsen & Roslender, 2015). Idiosyncratic risk results in an increase in repair expenses and lawsuit or fines that have a negative effect on the financial health of a firm (Lee & Faff, 2009).
2 Idiosyncratic risk is a risk on a particular asset. An example of an idiosyncratic risk is that a pipeline company may have a damaged pipeline which led to oil leakage (Lee & Faff, 2009).
Therefore, disclosing more (private) information decreases the information asymmetry and controls the agency problem between managers and investors. As a result, investors can better evaluate the value of their investments.
2.4 Cost of capital
2.4.1 Cost of equity capital
Financing through equity is one of the two ways for a firm to raise capital to fund their operations or other investments. Cost of equity is the expected reward an investor receives in exchange for their capital (Investopedia, 2017). Zhou et al. (2017) find that firms that improve their alignment with the <IR> framework experience a decline in the cost of equity capital. The underlying theoretical assumption is that IR provide new value relevant information in a concise and useful manner which potentially reduces the information asymmetry (Zhou et al., 2017). Some investors are better informed than others when the information asymmetry is high. The investors with less information have a higher risk and are hesitant to trade which leads to an increase in illiquidity (Verrecchia, 2001). This ultimately leads to an increase in the cost of equity capital (Amihud & Mendelson, 1986).
According to Diamond (1984, 1991), the equity market consists of diffused investors who experience higher agency costs than the investors within the debt market (Diamond, 1984, 1991). These investors base their returns and investment decisions on publicly available information and incorporate this information in their decision-making process (Fama, 1985).
2.4.2 Cost of public and private debt capital
Debt plays an important role in financing operational activities for firms globally due to their capital structure (ECB, 2007; Bharath et al., 2008). This provides lenders incentives to monitor firms by setting up interest, maturity, and security in a loan contract (Scholtens, 2006). Also, financing through debt has additional advantages because interest costs are deductible from taxable income (Fama & French, 2005).
Debt can be split up into public and private debt due to their different characteristics. First, (institutional) investors in the private debt market are quasi-insiders because of their low information asymmetry. These investors have a close relationship with the borrowing firm which results in a better access to information than investors in the public debt market and therefore,
these investors have lower agency costs than investors in the public debt market (Boyd & Prescott, 1986; Diamond 1984, 1991; Fama, 1985; Goss & Roberts, 2011; Vashishtha, 2014). Yosha (1995) and Bhattacharya and Chiesa (1995) argue that the information that borrowers disclose in the private debt market are difficult to observe by competitors (i.e. the private debt market has less negative spillover effects which could result in a loss of competitive advantage) (Yosha, 1995; Bhattacharya & Chiesa, 1995). Second, the public debt market consists of dispersed arm’s-length lenders while the private debt market consists of a small group of institutional investors. These small group of institutional investors have better tools to process acquired information and use it to evaluate the borrower and design loan contracts (Bharath et al., 2008). The processed information is useful to increase the monitoring efficiency of borrowers (Denis & Mihov, 2003). Furthermore, investors in the private debt market are specialized in credit evaluations which is useful in assessing and monitoring the performance of borrowers (Krishnaswami & Subramaniam, 1999). Therefore, determining the cost of debt capital is different for the investors in the private debt market than the investors in the public debt market.
3 Hypothesis development
Firms that are committed to sustainability reporting reduce their risk and the risk of their investors. Prior studies have shown that investors assess the/a firm’s risk and the capabilities of its management through incorporating sustainability information into their decision-making process (McGuire et al., 1988; Boutin-Dufresne & Savaria, 2004).
Investors in the equity and debt market agree that information needs to be integrated into one report (EY, 2015). The characteristics of integrated reports differ from those of sustainability reports because IR is more comprehensive and it integrates financial and non-financial information. Furthermore, IR also lowers information asymmetry by providing new relevant information which benefits all stakeholders (De Villiers et al., 2014; Barth et al., 2015; Zhou et al., 2017).
The debt market can be distinguished into public and private debt markets because the composition of investors in these markets are different from each other. Compared to the public debt market, investors in the private debt market have a close relationship with borrowers which lead to lower information asymmetry, lower agency costs, and better monitoring. Furthermore, these investors are specialized in evaluating their investments (Diamond, 1984, 1991; Boyd & Prescott, 1986; Fama 1985, Goss & Roberts, 2011; Vashishtha, 2014). As a result, investors in the public debt market, who have no access to private information, are dependent on corporate reporting (to acquire their information and incorporate this information) for their investment valuation. In contrast, investors in the private debt market, with access to private information, are less dependent on corporate reporting. Therefore, these investors should be less affected by corporate reporting, including IR, and as a result the negative effect of IR on cost of debt in private debt markets should be lower.
Overall, due to the different composition of investors in the public and private debt market and the characteristics of these investors, I expect that IR have a stronger negative effect on the cost of public debt than the cost of private debt. Therefore, my hypothesis is as follows:
H1: The effect of IR on the cost of debt is more negative in the public debt market than in the private debt market.
4 Research methodology
4.1 Sample description and data sources
Following the study of Serafeim (2015), the data regarding IR is collected through Datastream by using the Thomson Reuters ASSET4 database. In addition, the database is also used to collect data regarding Big 4 auditor and CSR reporting. ASSET4 is a database that provides objective, relevant and systematic environmental, social and governance information that is based on key performance indicators and individual data points along with their original data sources (Serafeim, 2015).
The sample from ASSET4 is a global sample and it contains 5,154 firms. To determine the cost of public and private debt and the control variables leverage, firm size, return on assets, operating cash flow, and loss, I use Bureau van Dijk’s Osiris database. The Osiris database can be accessed through Wharton Research Data Services (WRDS) and it contains detailed information on listed, and major unlisted/delisted firms around the world (WRDS, 2017b). The data collected from ASSET4 is merged with the data from Osiris database. My sample is not restricted to a continent or a country, rather, it contains information of all firms that are covered by the databases to achieve a large enough sample to test my hypothesis. My sample covers the period 2002 – 2015. The first practices of IR started in 2002. I have chosen 2015 as the ending period of my sample due to limited data availability in 2016.
After merging the data from the ASSET4 database with the data from the Osiris database, the full sample consists of 72,156 observations of 5,154 distinct firms over the years 2002 until 2015. After removing missing observations, the cost of public debt sample consists of 39,629 observations of 3,976 distinct firms, the cost of private debt sample consists of 35,763 observations of 3,741 distinct firms.
4.2 Research design
This study examines the effect of IR on the cost of public and private debt. To measure the effect, my regression models are based upon the models of Lorca et al. (2011) and Pittman and Fortin (2004). The study of Lorca et al. (2011) examines the effect of the board of directors’ attributes on the cost of debt financing. From their model, I have taken the dependent variable cost of debt capital and the control variables: firm size (measured as the natural logarithm of total assets), leverage, return on assets, Big 4 auditor, and industry dummy to control for industry effects. The variables interest coverage, collateral, and market-to-book value, board independence, director ownership, expertise, separation, and frequency are excluded because they are not relevant to my study or because the data is limited available. The model of Pittman and Fortin (2004) has the following variables: cost of debt capital, firm age, prime rate, default spread, leverage, cash flow, firm size, asset structure, Big 6 auditor, and negative book equity. From their model, I use leverage, cash flow, and firm size and include them into my model. Other variables, such as firm age, prime rate, default premium, asset structure, Big 6 auditor, and negative book equity are excluded in my model due to irrelevance or limited data availability. Due to my global sample and the chosen time-period, I have included year dummies and country dummies as additional variables to control for year effects and country effects.
Both studies focus on the total cost of debt. My study distinguishes the cost of debt into cost of public debt and cost of private debt. I follow the method of Lin et al. (2013) to distinguish the cost of public debt from the cost of private debt. My regression models are as follows:
△CODC_PUBLIC(i,t) = 𝛽0 + 𝛽1IR(i,t) + 𝛽2LEV(i,t) + 𝛽3SIZE(i,t) + 𝛽4ROA(i,t) + 𝛽5CF(i,t)
+ 𝛽6BIG4(i,t) + 𝛽7LOSS(i,t) + 𝛽8CSR(i,t) + ∑ 𝛽jYEAR_DUM(i,t) + ∑ 𝛽jIND_DUM(i,t) +
∑ 𝛽jCOUNTRY_DUM(i,t) + 𝜀 Equation 1: Change in cost of public debt capital.
△CODC_PRIVATE(i,t) = γ0 + γ1IR(i,t) + γ2LEV(i,t) + γ3SIZE(i,t) + γ4ROA(i,t) + γ5CF(i,t)
+ γ6BIG4(i,t) + γ7LOSS(i,t) + γ8CSR(i,t) + ∑ γjYEAR_DUM(i,t) + ∑ γjIND_DUM(i,t) + ∑
γjCOUNTRY_DUM(i,t) + 𝜀
Both models look at firms that do and don’t do IR during the period 2002 – 2015. The variables in the models are explained as follow:
CODC_PUBLIC: Cost of public debt capital measured as the total cost of debt capital minus the cost of private debt capital.
CODC_PRIVATE: Cost of private debt capital measured as the total cost of debt times the ratio of private debt to total debt.
IR: A dummy variable set to 1 if a firm issue an integrated report and 0 if the firm does not issue an integrated report.
LEV: A control variable for leverage. Calculated as the total debt divided by total assets. SIZE: A control variable for firm size, calculated as the natural logarithm of total assets.
ROA: Total return on assets, a control variable calculated as net income divided by total assets at the end of each year.
CF: Cash flow from operations. Defined as the net cash from operational activities.
BIG4: A dummy variable for Big 4 auditor. The value is set to 1 if a firm hires a Big 4 auditor and 0 if it does not hire a Big 4 auditor.
LOSS: A dummy variable for loss. The value is set to 1 if the net income is negative and 0 if it is positive.
CSR: A dummy variable set to 1 if a firm issue a CSR report and 0 if the firm does not issue a CSR report.
YEAR_DUM: Dummy variable to control for year effects. IND_DUM: Dummy variable to control for industry effects.
4.3 Variables
4.3.1 Independent variable: Integrated Reporting
I follow prior studies (Serafeim, 2015; Baboukardos & Rimmel, 2016) by using the ASSET4 database to determine which firms issue IR and which don’t. Specifically, I use the overall equal-weighted rating (A4IR), which measures the firm’s overall performance on economic, environmental, social and governance performance on a scale of 0 – 100. The score reflects whether a firm is able to show and communicate effectively and convincingly how it integrates the economic, environmental, social, and governance dimensions into its day-to-day decision making process. According to ASSET4, the higher the score, the better the performance of a firm in the pillars: economic, environmental, social and governance (Thomson Reuters, 2013; Serafeim, 2015). The scores are transformed into dummy variables to determine whether a firm issue IR or not. All scores above 0 will be transformed into a dummy variable with a value of 1. Missing scores or scores equal to 0 will be transformed into a dummy variable with a value of 0.
4.3.2 Dependent variable: Cost of debt capital
Based on the studies of Pittman and Fortin (2004), Lorca et al. (2011), and Lin et al. (2011), I calculate the cost of public and private debt capital as follows.
𝐶𝑜𝑠𝑡 𝑜𝑓 𝑝𝑢𝑏𝑙𝑖𝑐 𝑑𝑒𝑏𝑡 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 =𝑖𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑒𝑥𝑝𝑒𝑛𝑠𝑒
𝑡𝑜𝑡𝑎𝑙 𝑑𝑒𝑏𝑡 − 𝑐𝑜𝑠𝑡 𝑜𝑓 𝑝𝑟𝑖𝑣𝑎𝑡𝑒 𝑑𝑒𝑏𝑡 𝑐𝑎𝑝𝑖𝑡𝑎𝑙
Equation 3: Total cost of public debt.
𝐶𝑜𝑠𝑡 𝑜𝑓 𝑝𝑟𝑖𝑣𝑎𝑡𝑒 𝑑𝑒𝑏𝑡 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 = (𝑖𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑒𝑥𝑝𝑒𝑛𝑠𝑒 𝑡𝑜𝑡𝑎𝑙 𝑑𝑒𝑏𝑡 ) ∗ (
𝑝𝑟𝑖𝑣𝑎𝑡𝑒 𝑑𝑒𝑏𝑡 𝑡𝑜𝑡𝑎𝑙 𝑑𝑒𝑏𝑡 )
Equation 4: Total cost of private debt.
The total cost of debt is calculated as the interest expense divided by the interest-bearing debt of a firm (Pittman & Fortin, 2004; Francis et al., 2005; Lorca et al., 2011). Prior study first determines the cost of public debt and extract it from the total cost of debt to calculate the cost of private debt (Francis et al., 2005). However, due to limited data availability, I am restricted to determine the cost of private debt first and then extract it from the total cost of debt to calculate
the cost of public debt. The cost of private debt is measured as the total bank loans divided by the total interest bearing debt of a firm.
4.3.3 Control variables
Several control variables are included in my model to isolate the effect of IR on the cost of debt capital. Leverage (LEV) is used to control for differences in capital structure. This variable is measured as the total overall measure of debt divided by total assets of a firm. Several extant researches use this ratio to control for default risk because firms with greater debt present higher risk to debt providers (Pittman & Fortin, 2004; Francis et al., 2005; Lorca et al., 2011; Attig et al., 2013). Firm size (SIZE) is the natural logarithm of total assets and it is included because prior studies argue that creditors perceive larger firms as less risky due to their ability to better withstand negative economic shocks and are therefore less likely to default. Prior studies also argue that an increase in firm size have reputational effects which means that large firms are seen, by outsiders, as more trustworthy (Pittman & Fortin, 2004; Lorca et al., 2011; Goss & Roberts, 2011). Return on assets (ROA), calculated as the net income divided by the total assets at the end of each year, is also included as a control variable because it controls for default risk (Ashbaugh-Skaife et al., 2006; Lorca et al., 2011). Prior study argues that default risk has a positive relationship with the cost of debt. The lower a firm´s default risk, the lower the cost of debt capital (Sengupta, 1998). The control variable cash flow from operating activities (CF) is defined as the net cash from operational activities and it is scaled by dividing it with the total assets. According to Petersen and Rajan (1994), firms that generate more cash internally are in a better position to repay their debts and thus, are less risky which lowers the cost of capital (Petersen & Rajan, 1994). The control variable Big 4 auditor (BIG4) is a dummy variable that is set to 1 if the firm hires a Big 4 auditor and set to 0 if they don’t hire a Big 4 auditor. Prior studies argue that firms hire Big 4 auditors to counter information problems because they have superior monitoring which enhance the credibility of the financial statements. As a result, this enables firms to borrow at a lower cost (Mansi et al., 2004; Pittman and Fortin, 2004; Simunic et al., 2007). Other control variables are loss (LOSS) and CSR reporting (CSR). LOSS is a dummy variable set to 1 if a firm report a negative net income in a given year and 0 if otherwise. Prior study has found an increase in the cost of debt capital if a firm report a loss in a given year (Beatty, Weber, & Yu, 2008). Another study has found that the credit rating of a firm decreases if they report a loss. As a result, the credit rating affects the borrowing costs (Jiang, 2008). CSR is a dummy variable set to 1 if a firm issue a CSR report and 0 if otherwise. Issuing CSR could help a firm to further improve the information environment
of a firm (Zhou et al., 2017). Lastly, due to my global sample and the chosen period, I include dummy variables YEAR_DUM, IND_DUM, and COUNTRY_DUM to control for year, industry, and country effects.
4.4 Analysis
A quantitative method is used to determine the effect of IR on the cost of public and private debt capital. Multiple control variables are included in my model to isolate the effect of IR. For this study, I run an Ordinary Least Squares (OLS) regression. Before running the regression, I test the validity of my two regression models. First, I winsorize my variables at the 1st and 99th percentile to cut outliers which could influence the results of my regression. Second, I predict the residuals and see if both regression models have normally distributed residuals. Third, I use a correlation matrix of coefficients of my regression model to test if there are variables that lead to unreliable and unstable estimates of my regression. Lastly, I test for omitted variables in both regression models using the Ramsey Regression Equation Specification Error Test (RESET).
To confirm my hypothesis, I compare the equations and expect that the coefficient of IR (𝛽1)
is more negative for the cost of public debt than private debt. Before running the regression, I run a T-test to see if the cost of public debt sample is significantly different from the cost of private debt sample. After the T-test, I run the regressions individually with robust and clustering (on firm). Then, I run a Seemingly Unrelated Estimation (SUEST) to compare the regression coefficient of IR between the two regression models. In order to do that, I need to run the regression again without the options robust and cluster to store the estimation results of the two regressions into separate variables. The estimation results should be computed without the robust and cluster options because SUEST has a cluster option and it automatically computes robustness. After SUEST, the Chi² test is used to see whether there is a significant difference in effect between the two coefficients.
5 Results
5.1 Descriptive statistics
Table 1 shows the distribution of observations from 2002 until 2015 and the yearly difference in percentage. In total, the cost of public debt capital sample consists of 39,629 observations of 3,976 distinct firms. The cost of private debt capital sample consists of 35,763 observations of 3,741 distinct firms. As seen below, the sample of cost of public debt capital is slightly larger than the cost of private debt sample. In addition, both samples are on average evenly distributed among the years.
Table 1: Number of observations per fiscal year
Fiscal year Cost of public debt capital N Percentage Cost of private debt capital N Percentage 2002 2,296 5.79% 2,102 5.88% 2003 2,368 5.98% 2,155 6.03% 2004 2,473 6.24% 2,291 6.41% 2005 2,594 6.55% 2,370 6.63% 2006 2,685 6.78% 2,418 6.76% 2007 2,769 6.99% 2,430 6.79% 2008 2,833 7.15% 2,510 7.02% 2009 2,916 7.36% 2,572 7.19% 2010 3,029 7.64% 2,656 7.43% 2011 3,065 7.73% 2,718 7.60% 2012 3,135 7.91% 2,802 7.83% 2013 3,126 7.89% 2,861 8.00% 2014 3,178 8.02% 2,920 8.16% 2015 3,162 7.98% 2,958 8.27% Total 39,629 100% 35,763 100%
This table provides the yearly observations of the cost of public and private debt capital from the period 2002 until 2015.
Table 2 provides the descriptive statistics of the variables in the public debt and private debt sample. The dependent variables CODC_PUBLIC and CODC_PRIVATE were extremely positively skewed. Therefore, these variables were manipulated by dropping bad data (such as firms with a cost of borrowing above 1) and winsorizing at the 1st and 99th percentile. After dropping bad data and winsorizing, the data is transformed using natural logarithm. The CODC_PUBLIC and CODC_PRIVATE are normally distributed with a mean of -4.433 and -3.582, respectively. In addition, their standard deviations indicate that the sample is medium distributed around the mean (standard deviation of 1.969 for the public debt sample and 1.321 for the private debt sample). However, since the coefficient of variation of both dependent variables are low (value below 1), this would not cause any problems in my regression model. The independent variable IR has a mean of 0.624 for the public debt sample and a mean of 0.633 for the private debt sample. Both samples have more firms issuing IR than firms who don’t. In order to test the validity of my sample and the hypothesis, I compare my key variables between the two samples by using a two-sample t-test with equal variances in STATA. The t-test of the CODC_PUBLIC and CODC_PRIVATE has a p-value of 0.00, which indicate that there is a significant difference between the two means. I run another two-sample t-test with equal variances for IR. The results show a p-value of 0.994 which indicates that the means of IR between the two samples are not significantly different.
The distribution of LEV was manipulated due to extreme positive skewness. First, bad data was dropped. Then, the variable is winsorized at the 1st and 99th percentile. After dropping bad data and winsorizing, LEV was transformed using the square root to counter positive skewness. The mean of LEV is 0.466 for the public debt sample and 0.489 for the private debt sample. Furthermore, the control variable is in both samples normally distributed with a narrow distribution around the mean (standard deviation of 0.185 in the public debt sample and 0.177 in the private debt sample). The control variable SIZE is manipulated by dropping impossible values (below 0). Then, the variable is transformed using natural logarithm. The variable SIZE is normally distributed in both samples and it has a mean of 15.659 for the public debt sample and 15.598 for the private debt sample which indicate that the firms in both samples are similar in size. The standard deviation of SIZE for the public and private debt sample is 2.771 and 2.743, respectively. The control variable ROA is normally distributed and displays an average return of 4.50% in the public debt sample and 4.10% in the private debt sample. The standard deviation of this variable for the public and private debt sample is 0.096 and 0.090, respectively. This indicates that the distribution of the observations is narrow around the mean in both samples. Overall, the control variables LEV, SIZE, and ROA are in line with the study of Lorca et al. (2011). The control variable CF is manipulated by dropping bad data and winsorizing data at the 1st and 99th percentile.
Then, I follow Pittman and Fortin (2004) and scale the variable by dividing the cash flow from operating activities with the total assets. The mean of CF is 0.043 for the public debt sample and 0.034 for the private debt sample. In addition, the standard deviation for the public and private debt sample is 0.079 and 0.062, respectively. These results are in line with the results of Pittman and Fortin (2004).
Furthermore, 91.60% of the firms in the public debt sample and 93.40% of the firms in the private debt sample have hired a Big 4 auditor. Both samples are equally controlled for credibility of the financial statements. Moreover, 5.10% of the firms in the public debt sample and 3.80% of the firms in the private debt sample have a negative net income in a given year. In addition, 26.30% of the firms in the public debt sample and 26.10% of the firms in the private debt sample issue CSR additional to IR reports.
Table 2: Descriptive statistics of the public and private debt sample. Panel A: Public debt sample
Variable N Mean Min Q1 Median Q3 Max Std. dev.
CODC_PUBLIC3 39,629 -4.433 -17.013 -5.437 -4.226 -3.002 -0.798 1.969 IR 39,629 0.624 0.000 0.000 1.000 1.000 1.000 0.484 LEV4 39,629 0.466 0.035 0.350 0.480 0.590 0.923 0.185 SIZE5 39,629 15.659 2.833 14.016 15.434 17.235 26.226 2.771 ROA 39,629 0.045 -0.993 0.018 0.045 0.080 0.941 0.096 CF6 39,629 0.043 -2.042 0.000 0.000 0.078 3.041 0.079 BIG4 39,629 0.916 0.000 1.000 1.000 1.000 1.000 0.277 LOSS 39,629 0.051 0.000 0.000 0.000 0.000 1.000 0.220 CSR 39,629 0.263 0.000 0.000 0.000 1.000 1.000 0.440
Panel B: Private debt sample
Variable N Mean Min Q1 Median Q3 Max Std. dev.
CODC_PRIVATE7 35,763 -3.582 -19.524 -4.106 -3.246 -2.804 -1.265 1.321 IR 35,763 0.633 0.000 0.000 1.000 1.000 1.000 0.482 LEV 35,763 0.489 0.035 0.378 0.497 0.607 0.923 0.177 SIZE 35,763 15.598 5.193 13.991 15.300 17.100 26.226 2.743 ROA 35,763 0.041 -0.993 0.016 0.042 0.076 0.941 0.090 CF 35,763 0.034 -0.515 0.000 0.000 0.061 0.685 0.062 BIG4 35,763 0.934 0.000 1.000 1.000 1.000 1.000 0.248 LOSS 35,763 0.038 0.000 0.000 0.000 0.000 1.000 0.191 CSR 35,763 0.261 0.000 0.000 0.000 1.000 1.000 0.429 This table provides the number of observations, mean, minimum, first quartile, median, third quartile, maximum, and the standard deviation of the regression variables in the cost of public and private debt capital sample.
3 This variable is transformed due to extreme positive skewness. First, bad data is dropped. Second, the data is winsorized at the 1st and 99th percentile. Lastly, the data is transformed using natural logarithm.
4 This variable is transformed due to extreme positive skewness. First, bad data is dropped. Second, the data is winsorized at the 1st and 99th percentile. Lastly, the data is transformed using the square root.
5 Following prior studies, this variable is the total assets transformed using natural logarithm.
6 Following Pittman and Fortin (2004), the cash flow from operating activities is scaled by dividing it with total assets. 7 This variable is similar to CODC_PUBLIC. First, bad data is dropped. Second, the data is winsorized at the 1st and 99th percentile. Lastly, the data is transformed using natural logarithm.
The pairwise correlation matrix regarding the cost of public and private debt capital and its relation to the variables are presented in table 3 and 4. The correlation matrixes display a significantly (at 1% level) negatively association with the cost of public and private debt capital. Table 3 shows the pairwise correlation coefficient matrix of the public debt sample. The correlation coefficient of LEV and CODC_PUBLIC shows a negative coefficient of -0.3108 which indicates that these variables affect each other on a medium level. Furthermore, the table shows several slightly higher correlation coefficients such as CF and CODC_PUBLIC, BIG4 and CODC_PUBLIC, SIZE and IR, CF and ROA, and LOSS and ROA. Table 4 shows the pairwise correlation coefficient matrix of the private debt sample. The correlation coefficient of SIZE and CODC_PRIVATE shows a negative value of -0.3001 which indicates a medium association. Also, the table shows several slightly higher correlations between the variables SIZE and IR, CSR and IR, CF and ROA, and LOSS and ROA. Overall, there are no high correlations between the variables in either samples. These results are in line with Pittman and Fortin (2004) and Lorca et al. (2011). Therefore, there is no indication that multicollinearity exists in my regression model. Additional test for multicollinearity is done using VIF statistics (see Appendix B). The VIF statistics displayed no values higher than 4 which indicate that multicollinearity does not exist in either regression models.
Table 3: Correlation matrix of cost of public debt capital
Variable CODC_
PUBLIC
IR LEV SIZE ROA CF BIG4 LOSS CSR
CODC_PUBLIC8 1.0000 IR -0.0694* 1.0000 LEV9 -0.3108* 0.0055 1.0000 SIZE10 0.0389* 0.2364* -0.0041 1.0000 ROA 0.0447* 0.0307* -0.1742* 0.0549* 1.0000 CF11 0.2477* 0.0649* -0.1754* 0.1749* 0.2686* 1.0000 BIG4 -0.1036* 0.0925* 0.0555* -0.0412* -0.0099** -0.0714* 1.0000 LOSS 0.0978* 0.0322* 0.0219* -0.0088*** -0.2975* -0.0220* -0.0339* 1.0000 CSR -0.0380* 0.2422* 0.0138* 0.0772* -0.0089*** 0.0321** 0.0354* 0.0268* 1.0000 This table provides the pairwise correlation coefficients between the regression variables. See appendix A for the definition of the variables.
*Significant at 1% level **Significant at 5% level ***Significant at 10% level
8This variable is transformed due to extreme positive skewness. First, bad data is dropped. Second, the data is winsorized at the 1st and 99th percentile. Lastly, the data is transformed using natural logarithm.
9This variable is transformed due to extreme positive skewness. First, bad data is dropped. Second, the data is winsorized at the 1st and 99th percentile. Lastly, the data is transformed using the square root.
10Following prior studies, this variable is the natural logarithm of total assets.
Table 4: Correlation matrix of cost of private debt capital
Variable CODC_
PRIVATE
IR LEV SIZE ROA CF BIG4 LOSS CSR
CODC_PRIVATE12 1.0000 IR -0.1118* 1.0000 LEV13 0.0199* -0.0055 1.0000 SIZE14 -0.3001* 0.2364* -0.0041 1.0000 ROA -0.0152* 0.0307* -0.1742* 0.0549* 1.0000 CF15 -0.1274* 0.0649* -0.1754* 0.1749* 0.2686* 1.0000 BIG4 -0.0052 0.0925* 0.0555* -0.0412* -0.0099** -0.0714* 1.0000 LOSS -0.0197* 0.0322* 0.0219* -0.0088*** -0.2975* -0.0220* -0.0339* 1.0000 CSR -0.0551* 0.2422* 0.0138* 0.0772* -0.0089*** 0.0321* 0.0354* 0.0268* 1.0000 This table provides the pairwise correlation coefficients between the regression variables. See appendix A for the definition of the variables.
*Significant at 1% level **Significant at 5% level ***Significant at 10% level
12This variable is transformed due to extreme positive skewness. First, bad data is dropped. Second, the data is winsorized at the 1st and 99th percentile. Lastly, the data is transformed using natural logarithm.
13This variable is transformed due to extreme positive skewness. First, bad data is dropped. Second, the data is winsorized at the 1st and 99th percentile. Lastly, the data is transformed using the square root.
14Following prior studies, this variable is the natural logarithm of total assets.
5.2 Results of hypothesis tests
5.2.1 Regression results cost of public debt
Table 5 shows the OLS regression outputs of the cost of public debt capital, the cost of private debt capital, and the Chi² test. The output related to the cost of public debt capital displays a significant F-statistic of 121.67 (p-value 0.000) which indicates that at least one predictor variable significantly influences the cost of public debt capital. The regression output shows that IR is significantly (p-value 0.016) negatively associated with the cost of public debt capital, suggesting that there is a significant difference in the cost of public debt capital between firms who issue and firms who don’t issue IR.
The coefficient of the control variable LEV is significantly (p-value 0.000) negatively associated with the cost of public debt capital, indicating that firms who increase their leverage ratio experience a decrease in the cost of public debt. This result is in line with the result of Pittman and Fortin (2004). However, it is not in line with the results of Goss and Roberts (2011) and Lorca et al. (2011). An explanation of this result could be that a large part of the sample consists of firms with an investment strategy that causes external stakeholders to have a different view on the borrowing firm. For example, the borrowing firm could use the received cash to invest in (fixed) assets that could benefit operational activities and ultimately, increase profit. In the long-term, external stakeholders could see this as a benefit. Furthermore, the coefficient of the control variable SIZE is significantly (p-value 0.035) positively associated with the cost of public debt capital. The result suggests that large firms have a higher cost of debt capital compared to smaller firms. This result is not in line with the study of Goss and Roberts (2011) and Lorca et al. (2011) who argue that large firms have a lower cost of debt capital than smaller firms because they are better able to withstand negative economic shocks and thus, are seen as less risky. This difference with prior study could be explained that investors not only look at the size of a company but also its assets and performance. Both samples include companies that have a relatively low ROA and CF which could affect the cost of public and private debt capital. Furthermore, the coefficient of ROA is significantly (p-value 0.000) negatively associated with the cost of public debt capital. This is in line with the results of Lorca et al. (2011) who use ROA as a variable to control for a firm’s default risk. Compared to firms with low ROA, firms with high ROA perform better and are seen as less risky because they are more stable. The coefficient of CF is significantly (p-value 0.000) positively associated with the cost of public debt capital. This is not in line with the results of Pittman and Fortin (2004) who state that firms with a lot of cash flows from operational activities
are in a better position to service their debts and are therefore, seen as less risky because they are better able to withstand negative (economic) shocks than firms with lower level of cash flow from operational activities. Furthermore, the coefficient of the dummy variable BIG4 is insignificantly (p-value 0.165) negatively associated with the cost of public debt. In addition, the coefficient of LOSS is significantly (p-value 0.000) positively associated with the cost of public debt capital which indicates that firms that have a negative net income experience a higher cost of public debt capital. Furthermore, the coefficient of CSR is insignificantly (p-value 0.288) negatively associated with the cost of public debt capital. Moreover, the adjusted R² shows a value of 0.3178 which indicates that the model explains 31.78% of the variance of the cost of public debt capital.
5.2.2 Regression results cost of private debt
The output in table 5 regarding the cost of private debt capital shows a significant F-statistic of 67.94 (p-value 0.000). The value indicates that at least one predictor variable influences the cost of private debt capital. The output also shows that IR is significantly (p-value 0.001) negatively associated with the cost of private debt capital which indicates that there is a significant difference in the cost of private debt capital between firms who issue an IR and firms who don’t issue an IR. Furthermore, the table shows a significant (p-value 0.000) negative coefficient of LEV which is not in line with prior study (Lorca et al. 2011). This difference is possibly caused by a large percentage of firms that have a different type of investment strategy which affects the view of external stakeholders. The coefficient of SIZE is significantly (p-value 0.000) negatively associated with the cost of private debt capital which indicates that large firms, compared to smaller firms, experience a lower cost of private debt because they are less likely to default due to their ability to better withstand negative shocks. Therefore, these firms are seen as less risky. The result is in line with the results of Goss and Roberts (2011) and Lorca et al. (2011) and it may have economic significance in the private debt market because creditors have a larger amount to claim when a large firm goes bankrupt. The coefficient of ROA is significantly (p-value 0.000) negatively associated with the cost of private debt capital which suggests that firms with a high ROA experience a low cost of private debt in relation to firms with low ROA. The result is in line with the results of Pittman and Fortin (2004) and Lorca et al. (2011). Furthermore, the result has economic significance because ROA measures how a firm performs over the years and it is reliable for the investors. The coefficient of CF is insignificantly (p-value 0.672) positively associated with the cost of private debt capital. Moreover, the coefficient of BIG4 is insignificantly (p-value 0.151) negatively associated with the cost of private debt capital. The coefficient of LOSS is significantly
(p-value 0.000) positively associated with the cost of private debt which indicates that firms with a negative net income experience a higher cost of public debt capital than firms who have a positive net income. The result is in line with the results of Beatty et al. (2008) and Jiang (2008). The coefficient of the dummy variable CSR is insignificantly (p-value 0.145) negatively associated with the cost of private debt capital. Furthermore, the adjusted R² is 0.2192 which means that the model explains 21.92% of the variance of the cost of private debt capital.
5.2.3 Seemingly Unrelated Estimation regression results
The hypothesis tests whether the effect of IR on the cost of debt is more negative in the public debt market than in the private debt market. In order to test the hypothesis, the results of the two regressions need to be compared (using SUEST and the Chi² test in STATA).
The regression result regarding the public debt sample document a significantly (p-value 0.016) negative coefficient of IR on the cost of public debt capital which indicates that there is a difference in effect on the cost of public debt capital between firms that issue IR and firms that don’t issue IR. The regression result regarding the private debt sample document a significantly (p-value 0.001) negative coefficient for the cost of private debt capital. This result indicates that there is a significant difference in effect on the cost of private debt capital between firms that issue IR and firms that don’t issue IR. The result indicates that IR is relevant for the investors in the debt market. This empirical evidence has economic significance since managers in leveraged firms could use this information for their debt placement decisions. Furthermore, the coefficients in table 5 show a clear difference in the effect of IR on the cost of public and private debt capital. The coefficient of IR in the public debt sample is stronger than the coefficient of IR in the private debt sample. However, the Chi² test displays a very insignificant p-value (0.908) which indicates that the effect of IR between the cost of public debt capital and the cost of private debt capital is not significantly different. Based on this result, my hypothesis is not supported.
Table 5: Regression results of both cost of public and private debt (1) Cost of public debt (2) Cost of private debt Chi² test16 (1) & (2) IR -0.094 (0.016) (0.001) -0.089 (0.908) 0.01 LEV17 -2.938 (0.000) (0.000) -0.730 SIZE18 0.019 (0.035) (0.000) -0.041 ROA -0.978 (0.000) (0.000) -0.473 CF19 1.941 (0.000) (0.672) 0.113 BIG4 -0.100 (0.165) (0.151) -0.093 LOSS 0.452 (0.000) (0.000) 0.197 CSR -0.040 (0.288) -0.0400 (0.145) CONSTANT -2.149 (0.000) (0.000) -2.567
Year effects Yes Yes
Industry effects Yes Yes
Country effects Yes Yes
Observations 39,629 35,763 Adjusted-R² 0.3178 0.2192
F-statistic 121.67 67.94
p(F) 0.000 0.000
This table provides the regression results of the cost of public and private debt capital and the Chi² test related to IR. The regression results are robust and clustered by firm. The cost of public debt sample consists of 39,629 observations of 3,976 distinct firms. The sample of cost of private debt consists of 35,763 observations of 3,741 distinct firms.
16The p-value related to the Chi2 test of IR is based on one-tailed tests.
17This variable is transformed due to extreme positive skewness. First, bad data is dropped. Second, the data is winsorized at the 1st and 99th percentile. Lastly, the data is transformed using the square root.
18Following prior studies, this variable is the natural logarithm of total assets.
6 Conclusion
This study examines the effect of IR on the cost of public and private debt capital. Specifically, I examine whether the negative effect of IR is stronger on the cost of public debt capital than on the cost of private debt capital. My study is motivated by the growing interest of firms, investors and regulators in IR, the important role of debt markets, the limited research on and limited knowledge of IR, as well as calls from academics to further investigate IR. Furthermore, this study is motivated to research the claimed benefits of IR.
My study consists of two regression models with different samples. The first regression model examines the effect of IR on the cost of public debt capital. The sample of the first regression consists of 39,629 observations of 3,976 distinct firms. The second regression model examines the effect of IR on the cost of private debt capital and it consists of 35,763 of 3,741 distinct firms. After running separate OLS regressions for the models, I use a SUEST test in order to store the estimates of the two regressions into two separate variables. These variables are used for the Chi² test to measure whether the effect of IR on the cost of public and private debt is significantly different from each other.
The findings in this study indicate that firms who do IR, experience a low cost of debt capital. Specifically, I find that the negative effect of IR is stronger on the cost of public debt capital than on the cost of private debt capital. However, the Chi² test shows that this difference is not significant. Therefore, I conclude that the negative effect of IR on the cost of public debt capital is not stronger than the negative effect of IR on the cost of private debt capital. Thus, my prediction is not supported. My prediction is based upon prior researches that argue that investors in the public and private debt market are fundamentally different from other stakeholders. Private debtholders are, compared to public debtholders, quasi-insiders because they have a close relationship with the borrower. In addition, the private debt market consists of a small group of institutional investors that have better tools to acquire and process information from the debtor. As a result, the private debtholders are less affected by IR than public debtholders because these private debtholders rely less on publicly disclosed information such as IR.
My study contributes to the existing literature by extending the scope of IR research by comparing the cost of public and private debt and its association with IR. To my knowledge there is only one academic study that examines IR and its relation to the cost of capital. Specifically, the current literature only concentrates on the cost of equity capital. However, since the debt market is also economically important, the examination of this market and its association with IR is yet to be seen. The results of my study show a significant negative association between IR and the cost
