Change in Risk Reporting: The Effect of Stakeholders’ Influence

(1)

Change in Risk Reporting:

The Effect of Stakeholders’ Influence

Combined thesis for the master programs Accountancy and Controlling

June 16, 2017

Trudy Kamphof

Saffierstraat 104

9743 LK Groningen

Email: t.kamphof@student.rug.nl

Student number: 2344580

First supervisor: G.C. Helminck, Msc

Second supervisor: Prof. dr. J.A. Emanuels

Assessor: dr. V.A. Porumb

Word count: 11.958

(2)

ABSTRACT

This paper empirically examines the change in risk reporting, based on the risk reporting in obligatory SEC-filings. It was researched if the disclosure of a material weakness (MW) caused a significant change in risk reporting compared to a control sample. The next step would be to investigate whether the stakeholders influence this change, by looking into the change in the variables as given in Ullmann’s (1985) model on social disclosure and performance. These variables are stakeholder power, strategic posture, and past and current economic performance. However, the statistical tests revealed no significant difference in the change in risk reporting for organizations that disclosed an MW in comparison to the control sample. Therefore, the remainder of the study focused on the effect of the independent variables on the change in risk reporting in general. The results show a significant positive effect of the change in stakeholder power and a significant negative effect of the change in economic performance on the change in risk reporting. Concluding, these significant results offer a renewing insight into stakeholders’ influence on risk reporting, and provide an argument for the application of the stakeholder theory on risk reporting.

Keywords: risk reporting; internal control; internal control failure; stakeholder theory; stakeholder

(3)

1. INTRODUCTION

In September 2016 the news came out that Wells Fargo, one of the largest banks in the United States, was fined 185 million dollar for the opening of two million fake bank accounts (Blake, 2016). The failure had major effects on the organization: approximately 5,300 employees were fired, CEO Stumpf resigned, several lawsuits were filed, investigations into the fraud started, and Hillary Clinton even mentioned Wells Fargo as ‘an example of egregious corporate behaviour’ (Blake, 2016) in her campaign for the presidential elections. Wells Fargo also faced a decline in the opening of checking accounts and applications for credit cards. According to the CFO, this was partly due to bad publicity and irritated customers (Corkery, 2016). The new CEO Mr. Sloan said that his ‘immediate and highest priority is to restore trust’ (Corkery, 2016), and he also acknowledged problems concerning the culture of Wells Fargo.

1.1 Research Problem

Next to the example of Wells Fargo, there are numerous instances of internal control (IC) failures at other organizations that caused the same kind of problems. Previous research shows that the disclosure of material weakness (MW) in the IC has negative consequences for the organization. It leads to an increased cost of debt (Dhaliwal, Hogan, Trezevant, & Wilkins, 2011) and cost of equity (Ashbaugh-Skaife, Collins, Kinney Jr., & Lafond, 2009), and a higher audit fee (Hogan & Wilkins, 2008). The disclosure of an IC weakness also negatively affects the stock price (Beneish, Billings, & Hodder, 2008), which indicates that it increases investors’ risk perception (Rose, Rose, & Norman, 2016). The stakeholders will hold the management of the organization responsible for an IC failure (Erickson, Weber, & Segovia, 2011). Considering the importance of the stakeholders for the organization, the organization has to manage her reputation and restore the trust of the stakeholders (Erickson et al., 2011). This can be done by explaining the situation and by sharing details on the improvements in the IC system that were made in order to prevent recurrence of the failure (Erickson et al., 2011). Furthermore, the agency theory states that an efficiency loss due to increased information and agency problems can be reduced by supplying information, primarily to shareholders (An, Davey, & Eggleton, 2011; Deumes & Knechel, 2008). The stakeholder theory complements this by considering not only the accountability of the organization towards shareholders, but also towards other stakeholders (An et al., 2011). Therefore, it is expected that the risk reporting of the organization increases after an IC failure.

1.2 Contribution

It is still unknown if IC failures lead to a change in the subsequent risk reporting. IC disclosures have been a well-researched topic over the last decades (Deumes & Knechel, 2008). The interest in the topic

(6)

6 revived after the implementation of SOX section 302 (disclosure controls) and 404 (assessment of IC). Main themes among previous studies are the determinants of risk disclosure (Linsley & Shrives, 2006; Hooghiemstra, Hermes, & Emanuels, 2015) and the incentives of the organization to (voluntarily) report on IC and risk management (Ashbaugh-Skaife, Collins, & Kinney Jr., 2007; Dobler, 2008; Deumes & Knechel, 2008). An interesting research is the study of Hermanson (2000), who investigated the demand for reporting on IC among the users of financial statements. A research that is closer related to the topic of this study is the experimental research by Rose, Norman, and Rose (2010), which is about investors’ perceptions of the risk associated with ‘material control weaknesses’ and the influence of disclosure detail on their perceptions. However, the available research on IC and IC failures does not include research on the influence of IC failures on the subsequent risk reporting. Hence, this research focuses on the impact of IC failures on the subsequent risk reporting.

In addition, this study pays attention to determinants that are expected to influence this change. Since the organization is accountable towards stakeholders according to the stakeholder theory, it is interesting to research to what degree the change in risk reporting of the organization is influenced by the stakeholders. One of the fields in which the stakeholder theory is often applied is the research on corporate social accounting and disclosure (Gray, Kouhy, & Lavers, 1995; Chiu & Wang, 2015). A substantial part of this literature investigates the influence of stakeholders on the corporate social reporting (CSR) (Chiu & Wang, 2015; Soleimani, Schneper, & Newburry, 2014; Thijssens, Bollen, & Hassink, 2015; Clarkson, 1995). Frameworks that are used for these researches include Mitchell’s dynamic model of stakeholder relations for determining the salience of stakeholders (Mitchell, Agle, & Wood, 1997), and Ullmann’s contingency model (Ullmann, 1985). Ullmann’s model gives insights into stakeholders’ impact on corporate social disclosure (Ullmann, 1985). He introduced the model with the aim to reconcile the contradictory results of previous research on the relation between social disclosure, and social and economic performance. After its publication the contingency framework became important in the corporate social disclosure research (Chiu & Wang, 2015). The determinants of the change in risk reporting that are included in this research are derived from the framework of Ullmann (1985), which is renewing.

1.3 Relevance

This study has two main objectives that are relevant for both theory and practice. First, this research contributes to the research on IC and IC disclosures by looking into the change in risk reporting over time as a consequence of IC failures. The second objective is to improve the understanding of the expected change by looking into determinants, derived from the framework of Ullmann (1985), that are expected to influence this change. Ullmann’s framework takes stakeholder power, the strategic posture of the organization, and the past and current economic performance of the organization into account.

(7)

7 To summarize, this research aims to discover the influence of stakeholders on the change in risk reporting after an IC failure.

The objectives of this study are relevant for several reasons. To start, the research offers a valuable addition to the literature on IC and IC disclosure. Almost no research has been published on the subject of change in risk reporting due to an IC failure. Since the research topic is renewing, this study contributes to the understanding of the topic. Furthermore, if the expected difference in risk reporting between failure and non-failure organizations is found, it is clear that a part of the change cannot be explained by time, and changes in regulations or obligations. As the change in risk reporting is less prescriptive than mandatory reporting, the underlying motivation for the disclosure is an interesting research topic (Chiu & Wang, 2015). If the expected relations are found, it could be valuable to reinforce the connection between the research areas of IC disclosures and corporate social disclosures, since it could offer renewing or different views on the motives for (voluntary) IC disclosure.

Besides the theoretical contribution of this study, the information about the motivation of the organization to disclose certain information is also relevant for the stakeholders of the firm. For example, investors are interested in risk reporting in order to improve their decisions (Solomon, Norton, & Joseph, 2000). The theory behind the expected relations are mainly based on the agency and stakeholder theory. However, if the expected change in risk reporting and the influence of stakeholders is found, the signaling and legitimacy theory also offer a relevant view on the motives for risk reporting. The reasoning behind the signaling theory is that organizations have incentives to show to the external environment that they deliver superior quality, which can be done by sending signals e.g. by voluntary disclosure (An et al., 2011). Another reason to send a signal is to distract the attention of the community and media from negative activities of the organization, in order to legitimize their operations (Deegan, 2006 as referred to in An et al., 2011). If the expected change in risk reporting as a result of the IC failure is found, it could be possible that an organization changes its voluntary disclosure behaviour before the identification of an MW, as the organization has a motive to distract the attention from potential negative information. This signal can be valuable to stakeholders, since it helps to improve the understanding of the organization, and more specifically the risk position of the organization. The improved understanding can influence the decision making of the stakeholders (Solomon et al., 2000).

1.4 Structure

The objectives of this study are translated into the following research question that guided this research:

To what extent does stakeholders’ influence explain the change in risk reporting?

(8)

8 1. Is there a significant difference between the change in risk reporting for organizations that

reported an MW in comparison to organizations that did not report an MW? 2. What is the effect of stakeholders' influence on the change in risk reporting?

In the next section the theoretical background of the research is construed and the hypotheses are introduced. Thereafter, the research method is explained. The results of the study are presented in the fourth section. The paper ends with the conclusions of the research and a discussion of the results.

2. LITERATURE AND BACKGROUND

2.1 Main Theories

The agency theory takes the separation of risk bearing and ownership, and the information asymmetry between the owners of the company and the management of the company into account. This offers managers the incentive and possibility to act in their own interest instead of the owners’ interests (Fama & Jensen, 1983). The behaviour of the management needs to be controlled in order to reduce the agency problems and the costs of managing those problems, which can be done by implementing accounting and auditing methods (Deumes & Knechel, 2008). One of these methods to reduce the information asymmetry is the disclosure of information (An et al., 2011), in this case more specifically IC disclosures.

The stakeholder theory expands the agency theory, by including other stakeholders to whom the organization has responsibilities on top of the shareholders (Freeman & Reed, 1983). As defined by Freeman (1984, as cited in Frooman, 1999), a stakeholder is ‘any group or individual who can affect or is affected by the achievement of an organization’s objectives’, which usually includes shareholders, employees, customers, creditors, suppliers, public interest groups, governmental bodies, and the community (Chiu & Wang, 2015). Organizations can discharge their accountability towards the stakeholders by disclosing information (An et al., 2011). In conclusion, the disclosure of IC reduces the information asymmetry between the management and the owners of an organization, and it helps the management to offer transparency to their stakeholders.

2.2 Risk Reporting

A part of the research on corporate disclosure is about risk reporting (Abraham & Cox, 2007), which is the central subject of this article. Before elaborating on risk reporting, risk and risk disclosures need to be defined. Risks are related to uncertainty about both the occurrence as the outcome of an event (Linsley & Shrives, 2006). A development in the concept of risk is the inclusion of positive risks, which

(9)

9 are events with uncertain outcomes that are expected to have a positive impact on the organization (Linsley & Shrives, 2006). Consequently, risk disclosures are defined as disclosures that inform the reader of ‘any opportunity or prospect, or of any hazard, danger, harm, threat or exposure, that has already impacted upon the company or may impact upon the company in the future or of the management of any such opportunity, prospect, hazard, harm, threat or exposure’ (Linsley & Shrives, 2006).

Based on the agency and stakeholder theory, it is expected that the stakeholders are interested in the risk disclosures of the company. Hermanson (2000) analysed the demand of nine different financial statement users for management reports on internal control (MRICs). The results of her research show that the respondents agree that voluntary and mandatory MRICs lead to improved controls and enhanced oversight. Moreover, voluntary MRICs are expected to improve the decision making, in contrast to the expected neutral usefulness of mandatory MRICs for decision making (Hermanson, 2000). Investors want information about the strategies of the company and the risks an organization takes to gain a profit (Beretta & Bozzolan, 2004). In the absence of disclosures of private IC information, it is unlikely that investors are well-informed about the IC (Deumes & Knechel, 2008). The research of Solomon, Norton, and Joseph (2000) shows that investors are interested in more corporate risk disclosures in order to improve their investment decisions. Hence, the organization has to be able to reassure the investors that the organization is in control of the risks (De Loach, 2000 as reffered to in Beretta & Bozzolan, 2004). Hence, besides the use of a risk management framework, organizations have to effectively inform their stakeholders on the risks affecting the organization, and the way in which the organization plans to handle those risks (Beretta & Bozzolan, 2004).

However, there is currently a gap between the demand for risk information by the stakeholders and the supply of risk information by the organization (Schrand & Elliott, 1998; Linsley & Shrives, 2006; Roulstone, 1999). The risk information that organizations provide nowadays is too short, too much focused on the short-term, and unsuitable for decision making (Abraham & Cox, 2007). This leads to difficulties for the shareholders and other stakeholders, who are unable to accurately determine the risk level of the organization (Linsley & Shrives, 2006; Miihkinen, 2012). Organizations can use narrative disclosures to reduce this gap (Khlif & Hussainey, 2016). The use of narrative disclosures is in line with the demand of the institutional investors, who agree that organizations have to provide more detailed and specific risk information (Linsley & Shrives, 2006). Knowing that managers make a conscious trade-off about the level of risk disclosures, the extent of disclosures increase as the economic incentives to disclose increase (Deumes & Knechel, 2008). Possible incentives to voluntary disclose are a reduction of the cost of capital, the estimation risk, the perceived riskiness, and an improvement of the market liquidity (Deumes & Knechel, 2008; Elshandidy & Shrives, 2016). Therefore, it is expected that the amount of voluntary disclosures increases as the stakeholders perceive an increase of the information gap (Deumes & Knechel, 2008). Next to the increase in voluntary disclosure, the

(10)

10 information gap also caused an increase in the attention addressed to risk reporting by accounting regulators (Deumes & Knechel, 2008; Khlif & Hussainey, 2016)). Understanding the dynamics of risk reporting is used by regulators to reduce the information gap (Khlif & Hussainey, 2016). Another reason for the increase in regulations is found in the number of corporate scandals, e.g. Enron and Worldcom (Agrawal & Chadha, 2005). The next paragraph elaborates on IC failures and their impact on risk reporting.

2.3 Internal Control Failure

The most widespread framework for IC management is the Internal Control – Integrated Framework, which has been developed by the Committee of Sponsoring Organizations of the Treadway Commission (COSO) (Power, 2007 as referred to in Power, 2009). The relevance of the COSO report is also acknowledged by regulators, as evidenced by the fact that the Public Company Accounting Oversight Board (PCAOB) used the report as basis for their regulations on the internal control over financial reporting (ICFR) (PCAOB, 2007a). The PCAOB defines ICFR as follows:

‘A process designed by, or under the supervision of, the company’s principal executive and principal financial officers, or persons performing similar functions, and effected by the company’s board of directors, management, and other personnel, to provide reasonable assurance regarding the reliability of financial reporting and the preparation of financial statements for external purposes in accordance with generally accepted accounting principles (…)’ (PCAOB, 2007a).

Next to this definition the auditing standard gives a description of the policies and procedures that cover the ICFR. An interesting topic both in research and practice is the information about the limitations of the IC of an organization. Based on the Auditing Standard No. 2, these limitations are categorized in control deficiencies, significant deficiencies, and MWs (PCAOB, 2007a). These categories are often used in the research on IC weaknesses (e.g. Ashbaugh-Skaife et al., 2007). The scope of IC failures is in this research limited to the financial reporting and is operationalized as the reporting of an MW in the IC. An MW is defined as ‘a deficiency, or a combination of deficiencies, in internal control over financial reporting, such that there is a reasonable possibility that a material misstatement of the company’s annual or interim financial statements will not be prevented or detected on a timely basis’ (PCAOB, 2007b). A control deficiency and a significant deficiency are less severe in comparison with an MW. However, the corporate governance mechanisms still have to pay attention to these deficiencies issue (PCAOB, 2007b).

An IC failure has negative consequences for the organization. When stakeholders realize that the organization has governance and IC problems, the trust in the organization decreases (Gertsen, van Riel,

(11)

11 & Berens, 2006). The reputation of the firm is damaged and stakeholders have less confidence in the credibility of the management in relation to the financial reporting (Mercer, 2005; Erickson et al., 2011). This leads among stakeholders to more uncertainty and lower expectations with respect to the firm fulfilling its commitments (Chakravarthy, deHaan, & Rajgopal, 2014). Other research confirms that the perceived risk increases in response to the disclosure of an IC weakness (Rose et al., 2010; Beneish et al., 2008; Hammersley, Myers, & Shakespeare, 2008). The reduced trust in the organization and the increased risk-perception causes among other things an increase in the monitoring and bonding costs (Jensen & Meckling, 1976).

Knowing that stakeholders are essential for the organization, the organization has to regain the support and approval of the stakeholders (Erickson et al., 2011). This offers incentives to recover the reputation and reduce the uncertainty among the stakeholders (Erickson et al., 2011). To restore the trust in the relationship between the stakeholder and the organization, transparency is required (Jahansoozi, 2006). The disclosure of more risk information provides transparency, and also reduces the perceived risk (Hassan, 2009). In line with this, Deumes and Knechel (2008) show that increased information and agency problems increase the economic incentives to voluntary disclose. Hence, it is expected that the risk reporting of an organization increases after an IC failure.

2.4 Determinants of the Change in Risk Reporting

The determinants that are expected to influence the change in risk reporting are based on the stakeholder theory. It is likely that the stakeholders are able to influence risk reporting, as organizations need their support to survive (An et al., 2011; Clarkson, 1995). Several studies evidence that organizations take the stakeholders into account in their financial reporting policies (Graham, Harvey, & Rajgopal, 2005). Since the stakeholder theory takes next to the shareholders other stakeholders into account (An et al., 2011), the stakeholder theory offers more explanations for the changes in risk reporting than the agency theory.

One of the fields in which the application of the stakeholder theory is dominant, is the research on CSR (Gray et al., 1995; Ullmann, 1985). CSR can be defined as the reporting on subjects related to the environment, and ethical and social issues (Chiu & Wang, 2015). These disclosures can be made in any medium, e.g. annual reports, press releases, social reports, or company websites (Chiu & Wang, 2015; Gray et al., 1995). Since the regulations for CSR are less prescriptive and mandatory, the organization’s motivation for disclosure is an interesting research topic (Chiu & Wang, 2015). However, the conclusions of those studies are lacking systematic relationships, which had adverse effects on the concept of CSR (Gray et al., 1995; Ullmann, 1985; Mathews, 1987).

(12)

12 In response to the ambiguous results of previous research, Ullmann suggested that the used models are incomplete (Ullmann, 1985). In his opinion the missing element for the research into the relation between social disclosure, and social and economic performance was strategy. To overcome this shortcoming, Ullmann (1985) suggested a three-dimensional model including stakeholder power, the strategic posture of the organization, and the past and current economic performance of the organization. The model of Ullmann is supported by several researches (e.g. Roberts, 1992; Magness, 2006). In this study, the influence of these three factors on the change in risk reporting after an IC failure are studied. Translating Ullmann’s model to risk reporting can be justified, since the basic logic derived from the stakeholder theory that the disclosure of certain information can be used to manage the stakeholders of the organization is also applicable to risk reporting (Deegan & Blomquist, 2006). Subsequently, the expected relationships and hypotheses are further explained.

2.4.1 Stakeholder power

The basis for the model of Ullmann is stakeholder power, which is defined as the extent to which the stakeholders influence the organization to do something the organization otherwise would not have done (Mitchell et al., 1997). Ullmann (1985) founded his model among others on Thompson, who stressed that organizations are selective in which stakeholders they take into consideration (Thompson, 1967 as referred to in Ullmann, 1985). Pfeffer and Salancik (1978, as referred to in Ullmann, 1985) complement this view by stressing that the importance of the stakeholders derives from the resources they possess and the relevancy of those resources. In the area of CSR, several studies have found evidence for the relationship between stakeholder power and CSR (e.g. Deegan & Blomquist, 2006; Elijido-Ten, Kloot, & Clarkson, 2010).

In the following, the concept of stakeholder power is translated to the domain of risk reporting. Primary stakeholders, who are essential for the survival of the organization, and secondary stakeholders can hold the organization accountable according to the stakeholder theory (An et al., 2011; Clarkson, 1995). The organization can discard this accountability among other things by disclosing information (An et al., 2011), in this case more specifically risk disclosures. In case of an IC failure it is even more important for the organization to disclose information, as increased transparency helps to restore the reputation of an organization (Erickson et al., 2011). That is why it is expected that the risk reporting changes over time for organizations that reported an IC failure. However, the extent of the change in risk reporting is expected to depend on the level of stakeholder power. This is in line with the results of Boesso and Kumar (2007), who found that voluntary disclosure is used by the management to manage the stakeholders that are important and can influence the organization. If the organization has a higher dependency on the stakeholders, the organization has a stronger incentive to disclose more information and to be more transparent. Hence, it is expected that an increase of the stakeholder power results in an increase of risk reporting, which leads to the following hypothesis:

(13)

13

H1: The change in the power of the stakeholders is positively related to the change in risk reporting.

2.4.2 Strategic posture of the organization

The second dimension of Ullmann’s model is strategic posture, which is ‘the mode of response of an organization’s key decision makers towards social demands’ (Ullmann, 1985). It is possible for an organization to have an active or a passive strategic posture. With a passive posture, the organization does not attempt to monitor and manage the demands of the stakeholders. An active posture is characterized by intentionally managing stakeholders’ perceptions of the organization (Chiu & Wang, 2015). It comes forth from the results of studies on CSR that an active strategic posture leads to higher levels or a higher quality of social disclosure (e.g. Roberts, 1992; Chiu & Wang, 2015). When the organization is more actively managing the stakeholders’ demands and expectations (active posture), the organization is more sensitive for changes in the stakeholders’ demands. If the organization reports an MW, it is probable that the stakeholders’ demands are more pressing. Due to the increase in the activeness of the posture the organization, it is more likely that the organization responds to these demands by disclosing more risk information. This leads to the formulation of the following hypothesis:

H2: The change in the activeness of the strategic posture towards risk management is positively

related to the change in risk reporting.

2.4.3 Past and current economic performance

The third dimension of Ullmann’s model is the past and current economic performance of the organization, which is important in two ways (Ullmann, 1985). First, the economic performance of the organization influences the relative importance of CSR (Ullmann, 1985). Second, the economic performance influences the amount of resources that are available within the organization that can be used for CSR (Ullmann, 1985). The findings of Roberts (1992) show that organizations that performed relatively well in past periods have higher levels of social disclosure.

The same reasoning is applied to the change in risk reporting. The better the past and current economic performance, the more the organization’s attention and resources can be devoted to the risk management. Therefore, an increase in the past and current economic performance is expected to lead to a positive change in risk reporting. Hence, the following relationship is hypothesized:

H3: The change in the past and current economic performance is positively related to the change in risk reporting.

Next to this direct relation between the change in economic performance and the change in risk reporting, it is expected that the change in economic performance influences the relationship between the change in the activeness of the strategic posture of the organization and the change in risk reporting. This is in line with the reasoning of Magness (2006), who contrary to previous research into the model

(14)

14 of Ullmann, considered ‘the interactive impact of profit and strategic posture’. The financial performance of the organization influences the ability and willingness of the organization to meet the stakeholders’ demands regarding social disclosure (Magness, 2006). Translated to the change in risk reporting, the following relation is expected:

H4

:

The change in the past and current economic performance strengthens the relation between the activeness of the strategic posture towards risk management and the change in risk reporting. A schematic overview of the before mentioned hypotheses is presented in figure 1, the conceptual model. The next chapter elaborates on the research method that is used to test the hypotheses.

FIGURE 1

Conceptual model

3. METHOD

3.1 Sample Selection and Data Gathering

The sample is selected out of a data file retrieved from the database Audit Analytics. There are two general selection criteria, namely the disclosure of an MW, and the selected year has to be after 2004 due to the effects of the new SOX sections. Other sample selection criteria arise from the research design of other research group members, which makes it impossible to randomly select the sample. Next to the research sample, a control sample is needed to prove that there is a difference between the changes in risk reporting for organizations that disclosed an IC failure respective to organizations that

Change in stakeholder power

Change in activeness of the strategic posture

Change in past and current economic performance Change in risk reporting

+

(15)

15 did not.The control sample is randomly selected from the data file retrieved from Audit Analytics. The only prerequisite is that the organization did not disclose an MW in the selected year.

The total sample size, including the control group, is determined by taking time restrictions into account. Since the control sample is only used to make an inference about the difference in risk reporting between organizations that did or did not disclose an MW, the size of the control sample is limited to 65. This sample size is determined based on calculations of the required effect size, which is the strength of a relation between variables (Field, 2013). The necessary input to calculate the effect size is the statistical power (the ability of a test to find an effect; 0,8), the level of significance (α = 0,05), and the sample size. An effect size of 0,2 constitutes a small effect and an effect size of 0,5 constitutes a medium effect (Cohen, 1988, 1992 as referred to in Field, 2013). The effect size of the selected size of the control sample is 0,39, which means that a small to medium effect size can be proven with a significance level of 0,05.

After the selection of the sample, the data for the different variables are collected from the obligatory SEC-filings of the organizations in question, and are retrieved from the database BoardEx. In the following paragraph the data sources that are used for the data collection are mentioned explicitly.

3.2 Operationalization of Variables

3.2.1 Dependent variable: Change in risk reporting

The dependent variable of the research is the change in risk reporting (Chan_RR). This change is measured based on the number of words, since the amount of disclosures is one of the dimensions of disclosure quality (Beattie, McInnes, & Fearnley, 2004), and the amount of risk reporting is sensitive for change and easy to measure. The risk reporting is measured by using the risk information published in obligatory SEC-filings, e.g. 10-K or 20-F filings. To measure the amount of risk reporting the number of words in the sections ‘risk factors’ and ‘quantitative and qualitative disclosures about market risk’ is counted. Next, the number of words of risk reporting in the report of the year before the disclosure of the MW is detracted from the number of words of risk reporting after the disclosure of the MW. For the control sample, the selected year and the year before are used. To determine if there is a significant difference between the amount of risk reporting for organizations that disclosed an MW and organizations that did not, an independent samples t-test between means is performed.

3.2.2 Independent variables

3.2.2.1 Change in stakeholder power

In line with Elijido-Ten et al. (2010), this study assumes that the power of stakeholders is related to the relevance of certain stakeholders for the organization. Different measures for determining stakeholder power were used in previous studies (e.g. Soleimani et al., 2014; Thijssens et al., 2015). For this research

(16)

16 it is preferred that a broad array of stakeholders is included. Therefore, the idea of counting the times the stakeholders are mentioned in the risk reporting (Thijssens et al., 2015) is used. To determine which stakeholders and words need to be included, a categorization of stakeholders based on Fassin’s (2009) stakemodel of the firm is developed.

Fassin (2009) introduced this model in response to the criticism on the stakeholder theory by the introduction of a renewed model, The stake model of the firm. His model is strongly based on the original model of Freeman (1984) that was meant to give an overview of the strategic implications of the stakeholders for the organization (Fassin, 2009). He introduced three terms to describe the triangular relation between the stakeholders, namely (1) the stakeholder, who holds a concrete stake in the organization, (2) the stakewatcher, who champions the interests of the stakeholders, and (3) the stakekeeper, who has no stake in the firm but can influence the organization (e.g. regulators). For a complete overview of the categorization of Fassin (2009), see appendix A.

The focus of this research lies for two reasons on (most of) the stakeholders that fall into Fassin’s ‘stakeholders’ category. First, it is necessary to reduce the number of stakeholders to be able to gather the necessary data in the available time. All stakeholder models simplify the reality (Pesqueux & Damak-Ayadi, 2005), which increases the clarity of the model (Fassin, 2009). Second, the selected stakeholders are the most relevant and general stakeholders, and they are present in organizations operating in different countries and industries. Fassin’s ‘stakeholder’ category is the most basic category, and includes the stakeholders that hold a concrete stake in the organization. Next to that, the selection includes to a large extent the stakeholders who are part of the original stakeholder model of Freeman (1984) and the internal stakeholders in his revised model.

The groups that Fassin (2009) identified as part of the category ‘stakeholder’ are the financiers (including shareholders), employees, customers, business, communities, and the wider world. The ‘wider world’ is excluded from the categorization, because of the broad scope of stakeholders that fall into this category, which makes it difficult to determine for a large number of organizations. Next to the stakeholder groups identified by Fassin (2009), several synonyms that might be present in the obligatory SEC-filings of the selected organizations are identified. These synonyms, together with the stakeholder terms as identified by Fassin (2009), are included in the measurement index (appendix B). The count of these predefined stakeholder terms is applied to the sections ‘risk factors’ and ‘quantitative and qualitative disclosures about market risk’ of the obligatory SEC-filing in the selected year and the year before. Thereafter, the total count of the stakeholder terms is divided by the total number of words on risk reporting for both years separately. The final data on the change in stakeholder power (Sta_Pow) is calculated by detracting the relative stakeholder count in the year before the selected year from the relative stakeholder count in the selected year. To summarize, the measurement of the change in

(17)

17 stakeholder power depends on the change in density of the stakeholder terms in the risk reporting of organizations.

3.2.2.2 Change in activeness of the strategic posture

As discussed before, the organization’s sensitivity towards the stakeholders’ demands is expected to depend on the change in the activeness of the strategic posture. Chiu and Wang (2015) measured the strategic posture by looking at the presence of a formal department or a special-purpose team that works on the reporting that is demanded by the stakeholders. This is a sign that the organization has an active posture towards the relevant stakeholders. If this idea is applied to risk reporting, strategic posture could be measured by the presence of a CRO, the application of a risk management framework, and the number of employees in the risk department. However, for this research the strategic posture is solely measured by the presence of a person fulfilling a risk function in the executive board (referred to as ‘CRO’). The number of employees working in the risk department is excluded, since the necessary information is not publicly available. The application of a risk management framework is also excluded, since the organizations have to comply with regulations that are influenced by COSO (PCAOB, 2004), which is a wide-spread best practice ERM control framework (Power, 2007 as referred to in Power, 2009). On top of that, there is a direct link between the adoption of ERM and the appointment of a CRO according to Pagach and Warr (2011).

The presence of a CRO in one of the two years constitutes a change in the activeness of the strategic posture. If there is only a CRO in the selected year, the activeness increases. In this case, the change in the activeness of the strategic posture (Stra_Pos) is scored a ‘1’. If there is only a CRO in the year before the selected year there is a decrease in the activeness, which is scored a ‘-1’. If there is a CRO in both years, or if there is no CRO in both years, the change in activeness of the strategic posture is scored a ‘0’. The necessary data is retrieved from BoardEx.

3.2.2.3 Change in past and current economic performance

Ullmann (1985) identified two reasons for the importance of the past and current economic performance as a determinant of CSR. The impact of the change in current economic performance on risk reporting is determined by change the return on assets (ROA), in which ROA is calculated by dividing the net income by the total assets. ROA is a proper measure, since this ratio offers insight into the profitability of the firm. Furthermore, the ROA is also used by other studies on Ullmann’s model (e.g. Chiu & Wang, 2015; Magness, 2006).

Next to ROA, the leverage of the organizations is also measured. The leverage offers insight into the debt position of the organization, which is relevant for the relative importance attached to risk reporting (Ullmann, 1985). It also gives some information about the past performance of the organization as the leverage is likely to be higher for firms with a more negative past and current performance. The leverage

(18)

18 is calculated for both years by dividing the current and long-term debt by the total assets at the end of each year (Beneish et al., 2008).

The total score for the economic performance (Ec_Perf) is determined as follows. First, for the selected organizations the ROA and leverage are calculated for both years. The necessary data is derived from the relevant obligatory SEC-filings. Next, the organizations are ranked from the highest to lowest ROA and the lowest to highest leverage, in both the selected year and the year before. The organization with the highest (lowest) ROA (leverage) receives the highest score (=N). The organization with the lowest (highest) ROA (leverage) scores a one. Then, the total ranking score is determined for both years by adding the ranking for the ROA and the leverage. Finally, the change in economic performance is determined for each case by detracting the total ranking score of the year before the selected year from the total ranking score of the selected year.

3.2.3 Control variables

Next to the independent variables of this research, other factors might influence the change in risk reporting. Hence, three variables are deployed as control variables.

Previous research concluded that the industry impacts the voluntary corporate social responsibility disclosure (Gamerschlag, Möller, & Verbeeten, 2011). This is also the case for risk reporting, and especially for organizations operating in the financial industry, since the financial industry falls under relatively strict regulations (Khlif & Hussainey, 2016). Firms operating in the financial industry are expected to make different risk disclosures as they are seen as ‘risk management entities’ (Bessis, 2002 as referred to in Linsley & Shrives, 2006). To control for this effect an industry dummy (Ind) is included. The necessary data is extracted from Audit Analytics. Another factor that is found to influence the level of disclosure is the size of the organization, as mentioned by Hooghiemstra et al. (2015), and Thijssens et al. (2015). Therefore, the results of this study are controlled for the size (Size) by including the natural logarithm of the total assets of the organization in the research model. This operationalisation is in line with other researches (e.g. Hooghiemstra et al., 2015; Ashbaugh-Skaife et al., 2008). Finally, the change in the total length of the filing is included as control variable.

An overview of the variables, the attached labels, and the operationalization of the variables is given in table 1. Herein, the abbreviation ‘aft’ refers to the selected year and ‘bef’ refers to the year before.

3.3 Statistical Analysis

The first test that will be performed is an independent samples t-test between means, to determine if an IC failure causes a significant difference in the change of risk reporting. If the results show a significant difference between the means of the two groups, the proposed analysis can be executed.

(19)

19 TABLE 1

Variables and operationalization

Label Operationalization

Dependent variable

Change in risk reporting Chan_RR = # 𝑤𝑜𝑟𝑑𝑠 𝑅𝑅_𝐴𝑓𝑡 − # 𝑤𝑜𝑟𝑑𝑠 𝑅𝑅_𝐵𝑒𝑓

Independent variables

Change in stakeholder power Sta_Pow

= (# 𝑆𝑡𝑎𝑘𝑒ℎ𝑜𝑙𝑑𝑒𝑟𝑠_𝐴𝑓𝑡 # 𝑤𝑜𝑟𝑑𝑠 𝑅𝑅_𝐴𝑓𝑡 ) − (

# 𝑆𝑡𝑎𝑘𝑒ℎ𝑜𝑙𝑑𝑒𝑟𝑠_𝐵𝑒𝑓 # 𝑤𝑜𝑟𝑑𝑠 𝑅𝑅_𝐵𝑒𝑓 )

Change in strategic posture Stra_Pos ‘0’ if there is no executive officer fulfilling a risk position, or an executive officer fulfilling a risk position in both years;

‘1’ if there is only an executive officer fulfilling a risk position in the selected year; ‘-1’ if there is only an executive officer in the year before the selected year

Change in economic performance

Ec_Perf = (𝑅𝑎𝑛𝑘_𝑅𝑂𝐴_𝐴𝑓𝑡 + 𝑅𝑎𝑛𝑘_𝐿𝑒𝑣_𝐴𝑓𝑡) − (𝑅𝑎𝑛𝑘_𝑅𝑂𝐴_𝐵𝑒𝑓 + 𝑅𝑎𝑛𝑘_𝐿𝑒𝑣_𝐵𝑒𝑓)

Control variables

Industry Ind ‘1’ if the organization operates in the financial industry; ‘0’ if the organization operates in a non-financial industry

Company size Size = 𝐿𝑛(𝑇𝑜𝑡_𝐴𝑠𝑠𝑒𝑡𝑠_𝐴𝑓𝑡)

(20)

20 Before the multiple regression analysis is conducted, the outliers and missing values will be identified. The effect of outliers in continuous variables will be eliminated with the help of winsorizing, which adjusts all values that depart more than three standard deviations from the mean to a value of three standard deviations from the mean. Furthermore, a correlation analysis of the variables will be presented, since a high correlation can be a sign of multi-collinearity among the independent variables and control variables. Next to the correlation analysis, the level of multi-collinearity will be determined by the variance inflation factor (VIF). In case the VIF exceeds 10, the variable has to be excluded from the regression model and separate tests have to be performed.

After the before mentioned adjustments and tests have been performed, the descriptive statistics of the variables will be presented. Thereafter, the multiple regression will be performed in order to test the hypotheses. The statistical model that has been composed is as follows:

𝐶ℎ𝑎𝑛_𝑅𝑅 = 𝛽0+ 𝛽1∗ 𝑆𝑡𝑎_𝑃𝑜𝑤 + 𝛽2∗ 𝑆𝑡𝑟𝑎_𝑃𝑜𝑠 + 𝛽3∗ 𝐹𝑖𝑛_𝑃𝑒𝑟𝑓 + 𝛽4∗ (𝑆𝑡𝑟𝑎_𝑃𝑜𝑠

∗ 𝐹𝑖𝑛_𝑃𝑒𝑟𝑓) + 𝛽5∗ 𝐼𝑛𝑑 + 𝛽6∗ 𝑆𝑖𝑧𝑒 + 𝛽7∗ 𝐿𝑒𝑛𝑔𝑡ℎ + 𝜀,

in which βi denotes the coefficient of the independent variable and ε represents the error of the model. Different levels of significance will be used for the assessment of the relations (i.e. 0,10; 0,05; and 0,01).

4. RESULTS

The results of the data analysis are presented in this chapter, starting with the results on the change in risk reporting. Thereafter, the results on the hypothesized relations are presented. The chapter concludes with several additional analyses.

4.1 Change in Risk Reporting

4.1.1 Descriptive statistics

The total sample of firms that disclosed an MW consisted out of 235 cases. Four items were excluded from the sample, since there was no risk reporting before or after the disclosure of an MW. The control sample consisted of 83 case, which is larger than was determined in advance, because the original sample did only include one case with a CRO. The researcher chose to include several hand-picked cases in which there was a CRO in place. Before running the tests, the values that deviated more than three standard deviations were winsorized. This resulted in the descriptive statistics as presented in table 2, along with the results of the conducted tests.

(21)

21

4.1.2 Results

In order to analyze the difference in the change in risk reporting for respectively organizations that disclosed an MW and organizations that did not, three different independent sample t-tests are conducted. The first test included the complete sample and tested if there was a significant difference in the mean change in risk reporting between the sample and the control sample. The independent samples t-test was insignificant, t(314) = -0,381, p = 0,703. The average change in risk reporting for organizations that disclosed an MW (M = 568, SD = 1.540) does not differ from the average change in risk reporting of the control organizations (M = 642; SD = 1.437).

For the second test, the change in risk reporting was operationalized as the difference between the reporting on ‘risk factors’ in the year before the MW was disclosed and the year in which the MW was disclosed. This measure excludes the paragraph concerning the quantitative and qualitative market risks, since not all organizations in the sample are obliged to report on the market risks, and the changes in the reporting on the market risks are relatively stable compared to the reporting on the risk factors. Nevertheless, the results of this independent samples t-test were also insignificant, t(306) = -0,0278, p = 0,781. So, the average change in risk reporting on the risk factors does not differ between organizations that faced an MW (M = 521; SD = 1.502) and the control organizations (M = 564; SD = 1.250).

TABLE 2

Descriptive statistics and results of the independent samples t-tests

Test Groups N Mean SD Min Max T Sign.

1 MW 231 568 1.540 -4.624 5.833 -0,381 0,703 No MW 83 642 1.437 -3.028 5.831 2 MW 224 521 1.502 -4.596 5.695 -0,278 0,781 No MW 82 564 1.250 -4.596 4.502 3 MW 77 753 2.104 -4.123 7.503 0,395 0,693 No MW 83 642 1.437 -3.028 10.542

The final t-test that was executed excluded the companies that also reported an MW in the year before the selected year. In line with the expectation that an MW causes a change in the subsequent risk reporting, it is likely that this change is weakened if the organization also disclosed an MW in the year before the selected year. Hence, the third test includes only the cases which did not disclose an MW in the previous year. For this test the risk reporting on both the risk factors and the quantitative and qualitative market risks are included. The result of the test was insignificant, t(160) = 0,395, p = 0,693. Concluding, the average change in risk reporting for organizations that disclosed an MW in the selected

(22)

22 year and none in the year before (M = 753; SD = 2.104) does not differ from the average change in risk reporting of the control organizations (M = 642; SD = 1.437).

4.1.3 Conclusions and follow-up

Based on the before mentioned results, it can be concluded that the disclosure of an MW does not cause a change in the risk reporting. Since the dependent variable of the proposed research model is insignificant, and in order to make use of the collected data, the proposed research model is adjusted. The dependent variable of the research is changed into the change in risk reporting in general. In previous research there has been paid attention to risk reporting and factors that influence this (e.g. Linsley & Shrives, 2006; Hooghiemstra et al., 2015). Therefore, it is more renewing to look into the change in risk reporting between two years and search for variables that influence this change. The operationalization of this adjusted dependent variable remains the same as the one presented in table 1. Since there is no significant difference between the change in risk reporting for organizations that disclosed an MW and organizations that did not, the control sample is combined with the original sample.

The adjustment of the dependent variable has only small complications for the theoretical underlying of the hypothesized relations. For all hypothesized relations in chapter two it holds that the hypothesized effect of the independent variables on the dependent variable is strengthened by the effect of IC failures. Now that the dependent variable is no longer specific for organizations that had an IC failure, this strengthening effect of IC failures has to be ignored. This is only a small complication, since the idea of the relation remains the same.

4.2 Results of Adjusted Model

4.3.1 Descriptive statistics and correlation analysis

The combined sample consists out of 318 cases. Four cases that did not report on risks in one year or both years were treated as missing values for the change in risk reporting and the change in stakeholder power, since the measurement of these variables was based on the risk reporting. Due to the missing values for the assets at the end of the selected year, a fifth case was treated as missing values for the variables past and current economic performance. After the missing values were picked out, the data was winsorized. The descriptive statistics of the variables are presented in table 3.

Hereto, two remarks need to be made. First, the sample of 318 cases included 26 banks. For these cases the term ‘bank’ is excluded from the stakeholder power count. Second, it must be noted that there was limited data available on the presence of a CRO. There was one organization in the original sample that had CRO at the time of the selected year. To collect the necessary data on this variable the researcher hand-picked twenty organizations out of the BoardEx file, which also filed a 10-K or 20-F report. For

(23)

23 two out of these twenty cases the information was unavailable, and for another one it turned out that there was no CRO present. This resulted in a total sample of eighteen cases which had a CRO. Out of these cases, there were nine cases which had a change in the strategic posture. All eighteen cases are active in the financial industry, most of them (N = 14) are banks. It has to be concluded that the results on this variable need to be handled with caution.

Next to the descriptive statistics, a correlation analysis was conducted (table 4). There are several significant correlations between the variables, which can be a sign of a significant causal relation between the variables. The correlation analysis shows that multi-collinearity is not an issue, since the highest correlation among independent variables is 0,337. To further assure that multi-collinearity is not present, the highest VIF of each model is included in the table presenting the results of the multiple regression (table 5).

4.3.2 Results

4.3.2.1 Control variables

The first regression model only tested the control variables (table 5). This model is significant (p < 0,01) and the adjusted R-Square shows that the control variables together explain 22,1% of the change in risk reporting. Furthermore, there is a significant positive relation between the change of the total length and the change in risk reporting (β = 0,073; p < 0,01). Despite the changes in the Beta throughout the different models, this relationship remains significant at the p < 0,01 level.

4.3.2.2 Change in stakeholder power

In the second model the relation between the change in stakeholder power and the change in risk reporting is tested. The model is significant at the p < 0,01 level and the adjusted R-Square shows that 23,9% of the change in risk reporting is explained by the change in stakeholder power and the control variables. Next to the significant model, there is also a significant positive relation between the change in stakeholder power and the change in risk reporting (β = 143.047,417; p < 0,01). This means that an increase in stakeholder power causes an increase in risk reporting. Since the relation is conform the expectation and significant, and multi-collinearity (VIF = 1,010) is not an issue, it can be concluded that the first hypothesis is supported.

4.3.2.3 Change in strategic posture

The third model includes the control variables and the change of the strategic posture. The complete model is significant (p < 0,01) and accounts for 21,8% of the change in risk reporting. Based on the results it is concluded that the relationship between the change of the strategic posture and the risk reporting is not significant (β = -46,411; p > 0,10). Therefore, the second hypothesis is not supported.

(24)

24 TABLE 3

Descriptive statistics of independent and control variables

Variable N Mean SD Minimum Maximum

Chan_RR 314 587,27 1.510,36 -4.568 5.826 Sta_Pow 314 0,00012 0,00153 -0,00566 0,00596 Stra_Pos 318 0,03 0,166 0 1 Ec_Perf 317 1,06 120,883 -375 377 Stra_Pos x Ec_Perf* 317 0,0059 0,66065 -10,12 3,58 Ind 318 0,20 0,404 0 1 Size 317 19,06 3,037 9 28 Length 318 3.943 10.067 -4.4806 51.581

(25)

25 TABLE 4 Correlation analysis (1) (2) (3) (4) (5) (6) (7) (8) (1) Chan_RR 1 (2) Sta_Pow 0,148*** 1 (3) Stra_Pos 0,004 -0,033 1 (4) Ec_Perf -0,119** -0,095* 0,006 1 (5) Stra_Pos X Ec_Perf -0,131** 0,225*** 0,051 -0,149*** 1 (6) Ind 0,079 0,089 0,337*** -0,035 0,027 1 (7) Size 0,110* -0,031 0,133** 0,091 -0,002 0,096 1 (8) Length 0,478*** 0,006 0,010 -0,015 -0,087 0,115** 0,256*** 1

*

p < 0,10; ** p < 0,05; *** p < 0,01

(26)

26

4.3.2.4 Change in past and current economic performance

The fourth model includes, next to the control variables, the change in the economic performance. Overall, the model explains 22,7% of the change in risk reporting and is significant (p < 0,01). It was expected that there would be a positive relation between the change in economic performance and the change in risk reporting. However, there is a significant negative relation between the variables (β = -1,221; p < 0,10). Hence, the third hypothesis is not supported.

4.3.2.5 Direct relations

The fifth model includes all variables, except the moderating variable. Model five is significant at the

p < 0,01 level and explains 24,1% of the change in risk reporting. This is slightly higher than the

explanatory power of model 2. The relation between the change in stakeholder power and the change in risk reporting (β = 135.833,744; p < 0,01), and the relation between the change in past and current economic performance and the change in risk reporting (β = -1,061; p < 0,10) remain significant. The relation between the change of the strategic posture and the change in risk reporting is still insignificant (β = 37,377; p > 0,10).

4.3.2.6 Moderating effect economic performance

Model six and seven are based on standardized data, because these models include the moderating effect. The sixth model explains 22,7% of the change in risk reporting and is significant at the p < 0,01 level. The results of the moderating effect of the change in economic performance on the relation between the change of the strategic posture and the change in risk reporting is significant (β = -0142; p < 0,10). However, the direction of the relation is contrary to the expectation. Hence, hypothesis 4 is not supported.

4.3.2.7 Complete model

The final model includes all independent variables, the moderating variable and the control variables. In total, the variables explain 25,9% of the change in risk reporting (p < 0,01). Furthermore, the change in stakeholder power is significant and positive related to the change in risk reporting (β = 0,169; p < 0,01), which is in accordance with hypothesis 1. Next, the relation between the change in strategic posture and the change in risk reporting is insignificant related to the change in risk reporting (β = 0,013;

p > 0,10). The relation between the change in economic performance and the change in risk reporting

is significant, but the direction is contrary to the expectation (β = -0,104; p < 0,05). Finally, the moderating effect of the change in economic performance on the relationship between the change of the strategic posture and the change in risk reporting is significant (β = -0,222; p < 0,01). The direction of this relation is also not in correspondence with the expectations.

(27)

27 TABLE 5

Results of the regression analysis

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6**** Model 7****

Intercept 389,134 323,830 383,805 308,575 261,429 0,001 0,002 Control variables Ind 54,952 3,854 61,535 49,749 -3,391 0,018 -0,002 Size -5,216 -2,142 -4,925 -0,608 1,472 -0,008 0,008 Length 0,073*** 0,073*** 0,073*** 0,072*** 0,072*** 0,476*** 0,464*** Independent variables Stak_Pow - 143.047,417*** - - 135.833,744*** - 0,169*** Stra_Pos - - -46,411 - 37,377 - 0,013 Ec_Perf - - - -1,221* -1,061* - -0,104** Stra_Pos x Ec_Perf - - - - - -0,142* -0,222*** Adjusted R-Square 0,221 0,239 0,218 0,227 0,241 0,227 0,259 F-Value 30,458*** 25,516*** 22,772*** 23,945*** 17,516*** 23,913*** 16,582*** VIF 1,082 1,082 1,160 1,088 1,174 1,091 1,174 N 313 313 313 313 313 313 313

(28)

28

4.4 Additional Analysis

An additional analysis was performed on two interesting findings from the regression analysis, namely the change in stakeholder power and the past and current economic performance. The goal of the additional analysis was to deepen the understanding of the relationships and to improve the interpretation of the results.

4.4.1 A closer look at stakeholder power

The stakeholders that were included in the measurement index consist out of five groups, namely financiers, employees, customers, business, and community. An analysis was performed to see the effect of the change in power of these groups on the change in risk reporting. The change in stakeholder power of each group was determined in the same way as the change in stakeholder power in general. After the data was prepared, a correlation analysis was conducted (table 6). From this stems that there is a significant negative correlation between the change in risk reporting and the change in power of financiers. There are significant positive correlations between respectively the change in risk reporting and the change in power of employees, customers, and business. The correlation between the change in risk reporting and the change in power of the community is not significant.

TABLE 6

Correlation analysis of the stakeholder groups

Chan_Fin Chan_Empl Chan_Cust Chan_Bus Chan_Com

Correlation -0,153 0,199 0,241 0,162 0,010

Significance 0,007 0,000 0,000 0,004 0,865

The next step was to perform a linear regression, of which the results are presented in table 7. It appears from model (A) that the change in power of the financiers is significant and negative related to the change in risk reporting, which is contrary to the expectation. Further, the change in power of employees, customers and business is significantly positive related to the change in risk reporting. The relation between the change in power of the community and the change in risk reporting is not significant. If all sub-variables are put together in a model (model (F)), it becomes clear that the explanatory power of that model is the highest in comparison with other models tested in this paper. In this model, the relation between the change in risk reporting and respectively the change in business power and community power are not significant.

(29)

29 TABLE 7

Multiple regression analysis on the changes in stakeholder power

(A) (B) (C) (D) (E) (F) Intercept 440,227 226,281 490,600 471,600 442,518 412,231 Control variables Ind 86,446 9,739 58,937 57,653 48,647 40,556 Size -7,515 2,962 -10,220 -9,702 -7,070 -5,818 Length 0,071*** 0,070*** 0,072*** 0,072*** 0,073*** 0,068*** Sub-variables Chan_Fin -125.433,603** - - - - -118.896,449** Chan_empl - 279.926,979** - - - 282.749,811*** Chan_Cust - - 475.964,850*** - - 378.053,451*** Chan_Bus - - - 281.761,604** - 117.128,959 Chan_comm - - - - 247.042,215 112.391,282 Adjusted R-Square 0,229 0,241 0,272 0,239 0,219 0,293 F 24,120*** 25,713*** 30,171*** 25,442*** 22,875*** 17,185*** VIF 1,095 1,098 1,083 1,083 1,089 1,239 N 313 313 313 313 313 313 * p < 0,10; ** p < 0,05; *** p <0,01

(30)

30

4.4.2 Change in ROA and leverage

The results of the relation between the change in economic performance and the change in risk reporting are significant, but contradictory to the expectation. Therefore, it is interesting to further investigate this relation by splitting up the variable in the past performance, as measured by the leverage, and the current performance, as measured by the ROA. The leverage is operationalized as the difference between the total debt divided by the total assets for the year before the selected year and the selected year. The ROA is determined as the difference between the net income or net loss divided by the total assets for the year before the selected year and the selected year. A high (low) leverage corresponds with a bad (good) past performance, and a high (low) ROA with a good (low) current performance.

First, a correlation analysis between the two independent variables and the change in risk reporting was performed. The change in leverage did not significantly correlate with the change in risk reporting (r = 0,070; p = 0,217). The change in the ROA had also no significant correlation with the change in risk reporting (r = 0,009; p = 0,875). Next, a regression analysis was performed (table 8). The results show that the change in leverage has no significant effect on the change in risk reporting (β = 54,829, p = 0,180). The relation between the change in ROA and risk reporting is also not significant (β = -47,939;

p = 0,545). If both variables are included in the same model, the relations are still insignificant. Despite

the insignificance of the relations, the sign of the regression coefficients of the variables is remarkable since the signs are contrary to the expectation.

TABLE 8

Multiple regression analysis on the change in past and current economic performance

(A) (B) (C) Intercept 78,374 384,347 96,612 Ind 54,531 68,818 86,082 Size 11,365 -5,336 9,732 Length 0,068*** 0,074*** 0,069*** Chan_Lev 54,829 - 46,815 Chan_ROA - -47,939 -72,282 Adjusted R-Square 0,208 0,214 0,199 F 21,211*** 21,801*** 15,969*** Vif 1,188 1,081 1,208 N 309 306 302 * p < 0,10; ** p < 0,05; *** p <0,01

Change in Risk Reporting: The Effect of Stakeholders’ Influence