The influence of corporate social responsibility ownership on corporate reputation

The influence of Corporate Social Responsibility ownership on Corporate reputation

Student: Alina Stroebele / Student No 11923997 MSc. Business Administration, Strategy Track

University of Amsterdam, Faculty of Economics and Business

Supervisor: Hesam Fasaei 22nd _{of June, 2018}

STATEMENT OF ORIGINALITY

This document is written by Alina Stroebele who declares to take full responsibility for the contents of this document.

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

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

TABLE OF CONTENTS

ABSTRACT ... 4

1. INTRODUCTION ... 5

2.1CORPORATE SOCIAL RESPONSIBILITY (CSR) ... 9

2.2OUTCOMES/CONSEQUENCES OF CSRREPORTING ... 11

2.3CSR AND SIGNALLING THEORY... 13

2.4CORPORATE REPUTATION AND SIGNALLING THEORY... 14

2.5 CSR AND CORPORATE REPUTATION FROM A SYSTEMIC VIEW ... 16

2.6CSR OWNERSHIP ... 18

3. METHODOLOGY ... 21

3.1METHOD ... 21

3.2EMPIRICAL SETTING ... 21

3.3DATA COLLECTION AND FINAL SAMPLE ... 22

3.4DEPENDENT VARIABLE AND EVENT STUDY METHODOLOGY ... 22

3.5INDEPENDENT VARIABLE... 26 3.6MODERATING VARIABLE ... 27 3.7CONTROL VARIABLES ... 27 3.8EMPIRICAL METHODOLOGY ... 28 4. RESULTS ... 29 4.1ANALYTICAL STRATEGY ... 29

4.2HYPOTHESIS TESTING –HIERARCHICAL REGRESSION ... 30

4.3HYPOTHESIS TESTING –TESTING THE MODERATING EFFECT ... 31

6. CONCLUSION ... 39

Abstract

The importance of this research lies in the relevance of Corporate Social

Responsibility (CSR) reporting and recent debates on the genuineness of in-house CSR reports as well as the doubted impartiality of third-party awarded CSR. Genuine CSR has become a factor on which firms compete for survival in the market. Consumers are showing less trust in large conglomerates who have become known for issuing voluntary firm-owned CSR disclosures. More than ever have consumers as well as investors become savvy towards supposed impartiality of third-parties and are questioning awarded CSR initiatives such as awards and certification labels. This study attempts to solve discrepancies in former literature surrounding the relationship between CSR and corporate reputation. This paper introduces ownership as a moderator to explain the complex systemic construct of how CSR evaluations impact corporate reputation perceived by investors. The main claim of this research is that CSR will have a positive effect on corporate reputation and that this will be moderated and thereby enhanced through CSR ownership. This study makes use of an event study to determine corporate reputation by computing cumulative abnormal returns (CAR) over an event period in which either an owned or awarded CSR initiative is announced to the public. Furthermore, a multiple linear regression is run to test whether CSR has a positive impact on corporate reputation (CAR). The interaction variable NewCSR*Ownership is regressed against CSR and corporate reputation (CAR) in order to detect a possible moderation effect. Results show insignificant values for the regressions between corporate reputation (CAR) and the control variables, for CSR and corporate reputation (CAR) and for the moderation effect of ownership. This research primarily contributes to future research from a theoretical and methodological point of view as it provides other authors with a starting point on how to address the systemic needs of the relationship between CSR and corporate reputation.

1. Introduction

“Corporate Social Responsibility is a hard-edged business decision. Not because it is a nice thing to do or because people are forcing us to do it because it is good for our

business”

– Niall Fitzerald, Former CEO, Unilever

Corporate Social Responsibility (CSR), is particularly essential for large stock listed corporates who seek to maintain a positive corporate reputation with the public and their investors. Many firms attempt to uphold their reputations through self-regulation, by creating company own CSR reports, henceforth known as “owned” CSR. Other corporations apply to third party “awarded” CSR such as certification labels.

Past research on the relationship between CSR and firm reputation has identified

corporate reputation as an outcome of CSR reporting (Pérez, 2014). CSR reporting entails the voluntary disclosure of social and environmental initiatives by companies to demonstrate their concerns for stakeholder interests, in pursuit of bettering their corporate reputation (Pérez, 2014). This stream of research, which concentrates on outcomes/consequences of CSR reporting, is the smallest in the field of CSR communication research (Pérez, 2014). Other streams of CSR communication literature focus on disclosure/accountability, process orientation, consumer-related and business-related subtopics (Pérez, 2014). In her literature review Peréz (2014) determines transparency and information quantity and quality as main themes in literature on the relationship of CSR reporting and corporate reputation. According to Piechocki (2004), transparency emphasizes the importance of open communication flows with stakeholders concerning CSR policies and activities to reinforce positive reputation (as cited in Pérez, 2014). Information quantity and quality signifies the “positive effect of

intensity/quantity and orientation of CSR reporting on corporate reputation. (Michelon, 2011; Othman et al., 2011; Bayoud and Kavanagh, 2012)” (Pérez, 2014, p.19). This theme

emphasizes the importance of signal credibility (Pérez, 2014). According to Golob et al. (2013) literature on the relationship between CSR reporting and corporate reputation is scarce, heterogeneous and inconclusive (as cited in Pérez, 2014). Most previous studies are purely theoretical in nature and thus, more empirical research is required to effectively quantify the relationship between CSR and corporate reputation (Pérez, 2014). In the past, literature dealing with the relationship between CSR and corporate reputation has primarily focused on legitimacy concerns regarding this association (Sen et al., 2006). Archel et al. (2009), criticize this limited perspective in that legitimacy theory is unable to address wider systemic views, for example, the response of business and capitalism to sustainable

development demands (Archel et al., 2009). This research offers an empirical perspective on stock listed US firms and their reactions, in form of CSR communications, on sustainable development demands by important stakeholders. In previous research CSR reporting is considered in very general terms, excluding the different dimensions of CSR reporting (Pérez, 2014). The concept of “owned” versus “awarded” CSR may make up for lack of different dimensions in this field of research and further explain different types of reactions to

sustainable developments demands, mentioned above. Former research has primarily focused on the direct relationship between CSR and reputation, which this study aims to expand by introducing the moderating effect of CSR “ownership” as elaborated in the preceding section. This research will empirically test the relationship between CSR and corporate reputation and the moderating effect of CSR “ownership” by means of quantitative data analysis which will contribute to already existing theoretical explanations. For this study a sample of 87 US stock listed firms is collected, which issued either owned or awarded CSR initiatives. An event study is conducted to measure cumulative abnormal returns (CAR) of company stocks throughout the event windows in which the CSR initiatives are publicly announced. A

multiple regression analysis is run to detect the direction of the relationship between CSR and corporate reputation (CAR) and to determine whether ownership has a moderating effect on the independent and dependent variable relationship.

The aim of this research is to further the field of CSR in extending the explanation of its relationship to corporate reputation. According to Fombrun et al. (2000), the significance of this relationship lies in that activities creating Corporate Social Performance (CSP) only indirectly affect their bottom line through intangible ‘reputational capital’ rather than a direct impact on financial performance. Therefore, this study will contribute to previous studies by forming views on CSR ownership and its effect on corporate reputation through the eyes of shareholders. Firstly, this paper contributes to the, rather smaller, outcomes/consequences of CSR reporting field. Bebbington et al. (2007) remedy former, limited views on legitimacy theory by characterizing CSR as outcome and part of reputation risk management. The authors criticize literature which states that CSR contributes to firm accountability, as it lacks more in-depth analysis of why organizations chose to issue CSR reports (Bebbington et al., 2007). Instead, they explore contributions of reputation risk management (RRM) in

explaining corporate motives towards publishing CSR reports and conclude the theory’s usefulness for further investigation (Bebbington et al., 2007). The topic of this paper

compliments Bebbington et al.’s research (2007) in that it offers new empirical perspectives on the outcomes/consequences of CSR reporting. The differentiation between owned and awarded CSR initiatives helps to shed light on what motivates firms to issue own CSR reports as opposed to pursuing third part awarded CSR initiatives, for example certification labels. A possible explanation for the popularity of in house CSR reporting is the effect of reputation risk management on firm’s decisions towards CSR reporting ownership (Bebbington et al., 2007). Secondly, Archel et al. (2009), as mentioned previously, criticize legitimacy theory in that it is unable to address wider systemic views on how business responds to demands of

sustainable development. Golob et al. (2013) address the need for more of a systemic view by characterizing corporate reputation as a business-related outcome of CSR communication efforts (as cited by Archel et al., 2009). This perspective is sketched by examining the CSR communication responses, in form of CSR ownership decisions, towards sustainable development demands by shareholders. Through the lens of systems thinking this research offers multiple explanations for why firms chose to issue either owned CSR reports or pursue CSR initiatives that are awarded by third parties. These explanations are derived from theories such as reputational risk management and signalling theory all boiling down to corporate reputation. Thirdly, this research contributes to business practice in that it raises awareness towards firms’ behaviour in managing their corporate reputation through CSR ownership tactics. Often firms tailor their CSR reporting to their reputational needs. For example, many firms issue own CSR reports after suffering from reputational damage through corporate scandals. Grayson and Hodges (2004) highlight an example of such reputational risk

management, namely Shell’s strategic efforts towards sustainable development which resulted from increased political, economic, social and technological changes as well as heightened stakeholder expectations with regards to corporate behaviour (as cited by Ite, 2006).

This study will follow the subsequent structure: First, literature on CSR, signalling theory and corporate reputation is reviewed, leading to hypothesis one. Second, the

relationship between CSR and corporate reputation is examined from a systemic angle and the moderation effect of ownership is introduced, leading to the second hypothesis. Third, the methodological section introduces the sample, variables and methods employed in this research. Fourth, the statistical results and are analysed. Fifth, methodology and results to the hypothesis testing are discussed while considering literature as well as limitations to the current study. Sixth, the paper is concluded by presenting considerations for future research.

Throughout this paper the following research questions will be addressed: RQ1: How does the engagement in CSR initiatives affect corporate reputation? RQ2: How is the relationship between CSR and corporate reputation affected by CSR

“ownership”?

2. Literature review and theory development

2.1 Corporate Social Responsibility (CSR)

Before modern views on Corporate Social Responsibility (CSR) emerged in the 1970’s Bowen (1953) coined with the idea of the social responsibilities of the businessman (SRB). This laid the foundation for modern concepts of CSR and the role of business as responsible actors in society. The modern views on CSR emerged from the discourse between shareholder and stakeholder theory. “According to the shareholder approach, regarded by Quazi and O'Brien (2000) as the classical view on CSR, "the social responsibility of business is to increase its profits" (Friedman, 1962).” (Marrewijk, 2003). Contrastingly, Freeman (1987) poses stakeholder theory, which takes more of a societal approach in that companies are regarded as responsible to society as a whole (as cited in Marrewijk, 2003). Modern CSR research can be grouped into four fields of theories, according to Garriga and Melé (2004), instrumental theories, political theories, Integrative theories and ethical theories. First, Instrumental theories view CSR as a strategic means to create wealth (Garriga and Melé, 2004). This is in accordance to Friedman’s (1970) view of shareholder wealth maximization (as cited in Garriga and Melé, 2004). Instrumental theories consist of, shareholder value maximization, strategies of achieving competitive advantage, and cause-related marketing. Second, Political theories rather focus on the role of business in society and their CSR responsibilities ensuing from their power positions (Garriga and Melé, 2004). These theories

include, corporate constitutionalism, integrative social contract theory, and corporate citizenship (Garriga and Melé, 2004). Third, Integrative theories focus more specifically on how businesses take social demands upon themselves as their existence, continuity and growth depends on society (Garriga and Melé, 2004). Integrative theories include, issues management, principle of public responsibility, stakeholder management, corporate social performance, and normative stakeholder theory. Fourth, ethical theories concentrate on ethical specifications which dictate the relationship between business and society (Garriga and Melé, 2004). These theories include, normative stakeholder theory, universal rights, sustainable development, and the common good approach (Garriga and Melé, 2004). The

conceptualization of ethical theories derives from normative ethics which can be differentiated into three separate theories, utilitarianism, deontology, and virtue ethics (Garriga and Melé, 2004; Hasnas, 1998).

Within CSR research political theories, corporate citizenship specifically, address the participation of business in society and their sense of belonging to a community (Garriga and Melé, 2004). Matten et al. (2003) define three views of corporate citizenship, a limited view, a view equivalent to CSR, and their extended view of corporate citizenship. The authors argue,

“that because of elements of institutional failure, crucial to the functioning of the notion of liberal citizenship, corporate involvement in ‘citizenship’ moves from a voluntary form of behaviour to an unavoidable occurrence which ultimately results in a necessary reconceptualization of business-society relations.” (Matten et al., 2003, p.115).

Thereby, CSR can be seen as a way in which companies self-regulate the reporting of their social contributions, on one end of the spectrum, or as a binding behaviour in response to social pressures.

2.2 Outcomes/consequences of CSR Reporting

According to Golob et al. (2013), disclosure/accountability and process-oriented stream of literature makes up the largest proportion of literature focusing on CSR

communication and reporting, referred to as self-regulation in the above section (as cited in Pérez, 2014). Perez (2014) differentiates between two ways of defining and explaining the relevance of corporate reputation for firms, namely legitimacy and transparency. The former is discussed in conjunction with self-regulation and voluntary disclosure, while the latter is addressed below, focusing on information transparency towards stakeholders. Bebbington et al. (2007) discuss the, rather smaller, outcomes/consequences field of CSR reporting. The authors criticize limited views on legitimacy theory. CSR, in their point of view, acts as an outcome and a part of reputation risk management (Bebbington et al, 2007). Literature characterizing CSR as a contribution to firm accountability is heavily criticized by these authors, since, from their point of view, it lacks more in-depth explanations to why

organizations issue CSR reports (Bebbington et al., 2007). As alternative theory, Bebbington et al. suggest reputation risk management (RRM) as motivational factor for firms to publish CSR reports (Bebbington et al., 2007). Thereby, Bebbington et al. (2007) contribute to literature which specifies that CSR can also be used for firm serving tactics, rather than solving genuine societal needs.

Several other authors extend Bebbington et al.’s (2007) view to CSR scepticism, faced by a multitude of stakeholders. This theory attributes a negative relationship to the CSR

reporting and firm reputation relationship depending on specific communication strategies. Kim (2014) concludes that when firms state both self- and society-serving motives for issuing CSR reports, rather than publishing only society-serving motives, stakeholders react less favourably to the latter communication strategy. This negative association with CSR stems from distrust towards self-published, non-transparent corporate communication. According to Kim (2014), corporate reputation has a predetermining effect on stakeholder scepticism. When a company already suffers from a poor reputation, due to corporate scandals, stakeholders are less likely to trust CSR and may view such programs as a public stunt for profit generation (Kim, 2014). This negative association with CSR programs, as public stunts (Kim, 2014), suggest a negative effect of CSR on corporate reputation due to firms’ non-transparent reputation risk management strategies (Bebbington et al., 2007). This negative relationship, specified above, can be primarily traced back to stakeholders’ perceptions of firms employing CSR as firm-serving tactics rather than genuinely contributing to societal wellbeing.

Other CSR reporting outcomes/consequences literature specify a positive relationship between genuine CSR and corporate reputation. Espinosa and Trombetta (2004) find that firms with better quality annual report disclosures score higher in terms of corporate

reputations (Espinosa and Trombetta, 2004). Furthermore, Toms (2002) also finds a positive relationship between genuine CSR and corporate reputation, specifically the author discovers that the disclosure of environmental efforts in annual reports positively contribute to the creation of environmental reputation (Toms, 2002). These researchers conclude that genuine CSR initiatives significantly influence stakeholder’s positive perceptions towards firm reputation.

Authors who are part of the outcome/consequences field of CSR reporting draw up negative influences of firm- serving CSR tactics and positive effects of genuine society-serving CSR initiatives on corporate reputation. Most importantly, these studies share a common perspective on the importance of analysing consequences/outcomes of CSR reporting and the central role of transparency rather than determining CSR legitimacy. CSR transparency is furthermore essential for firms to signal positive CSR performance to their investors. In case of too many information asymmetries regarding positive CSR conduct it may become more difficult for investors to distinguish between social corporate citizens and firms with faulty CSR policies.

2.3 CSR and Signalling theory

Past literature has identified CSR as a strong signal of firm values to numerous stakeholders, including investors. According to Jones and Murrel (2001), Corporate Social Performance (CSP) functions as an uncomplicated decision tool, in which investors, for example, form impressions about a firm, its values, direction and overall worth. Such

impressions can be made despite variable and incomplete information on firm practices (Jones and Murrel, 2001). Fombrun and Shanely (1990), see reputational status as a dimension on which firms compete in the market. Similarly, to Jones and Murrel (2001), Fombrun and Shanely (1990) address information asymmetries as the main reason firms signal key

characteristics to constituents, as for example, CSP (Fombrun and Shanely, 1990). Mahoney et al. (2012), examine CSR and signalling theory by comparing and contrasting signalling theory with greenwashing attempts. According to the authors, signalling theory suggests that “good” corporate citizens will be more likely to publish CSR reports, whereas greenwashing predicts that “bad” corporate citizens will employ CSR disclosures more often (Mahoney et al, 2012). This comparison is important to consider in the relationship between CSR and corporate reputation, as according to signalling theory, more often firms which publish CSR

reports rightfully claim a positive corporate reputation. In contrast, according to greenwashing theory, more deceiving firms will benefit from publishing greenwashing CSR reports

(Mahoney et al, 2012), thereby increasing their likelihood to attain a fake positive reputation among shareholders. The authors contribute to literature on the positive genuine CSR

corporate reputation link by taking the signalling theory perspective (Mahoney et al, 2012). Mahoney et al.’s (2012) research on CSR and signalling theory links to research on voluntary disclosure theory as mentioned in a previous section (2.1Corporate Social Responsibility). Signalling theory theorizes that firms weigh costs of CSR reporting against benefits gained, however when firms perform poorly in terms of CSR performance they will tend to avoid punishment costs from publishing falsified, positive CSR reports, thereby foregoing social disclosure (Mahoney et al., 2012). According to Prado-Lorenzo and Garcia-Sanchez (2010) firms with inferior CSR avoid distributing CSR information that would negatively affect their reputations (as cited in Mahoney et al., 2012). The authors mentioned above, conclude that CSR is a strong signal of firm values, to stakeholders, due to information asymmetries, and thereby determines corporate reputation, which is controlled for by market mechanisms.

2.4 Corporate Reputation and Signalling theory

Signalling theorists explain corporate reputation as a phenomenon stemming from judgements made by outside observers with regards to prior resource allocations managers make to first-order activities (Fombrun and Riel, 1997). These in turn signal a perception of reliability and predictability to, for example, investors (Fombrun and Riel, 1997). Info asymmetry plays a major role in signalling theory and the formation of company reputations, as outside observers do not possess insider knowledge on firm features and products

(Fombrun and Riel, 1997). As a result of info asymmetries, managers can make use of their company’s reputation to signal attractiveness to its shareholders (Fombrun and Riel, 1997).

This may be achieved by, for example smoothing quarterly earnings to keep dividend pay-outs high and fixed or by employing renowned auditors, to signal high and stable earnings to investors (Fombrun and Riel, 1997). Applying signalling theory to corporate reputation and CSR, firms with above average CSR performance may incur an opportunity loss, when dealing with information asymmetries, since their corporate reputation could be valued more highly if investors were aware of their superior performance. In contrast, firms with inferior social performance can make an opportunity gain by receiving a weighted average of

investors general perceptions which may result in a higher reputation gain than if shareholders were fully aware of their faulty CSR initiatives (Morris, 1987). Adapting Morris’ (1987) view on signalling theory to the CSR and corporate reputation relationship, in order to succeed, firms employing genuine CSR initiatives must signal their positive CSR contributions to their shareholders in an effective way. The author suggests for signals to be inimitable by

competitors with low CSR performance. Furthermore, since signalling costs are inversely related to CSR performance, superior societal commitments must be observable after investors have committed to investing in the firm (Morris, 2012). On the one hand, firm-serving CSR initiatives can act as a negative signal (Bebbington et al., 2007; Kim, 2014), whereas on the other hand, genuine CSR proves as positive signal (Espinosa and Trombetta, 2004; Toms, 2002), towards investor’s perceptions of corporate reputation, depending on their proper execution. Assuming market efficiency, the genuine CSR initiatives will have a

positive effect on corporate reputation as falsified CSR will be punished by the market.

In conclusion, within the field of CSR research, authors examining outcomes

consequences of CSR reporting identify genuineness of CSR initiatives as the leading factor for positive shareholder reactions. As opposed to legitimacy research the outcomes

consequences CSR field does not focus on reasons for firms to employ CSR reports, yet looks from the angle where companies have already issued such initiatives. Accordingly, corporate

reputation proves as part of this stream of literature as it can help detect the

outcomes/consequences of CSR reporting. Corporate reputation is a tool used by investors to judge investment potential in corporations. This phenomenon is derived from signals sent by firms to overcome information asymmetries faced by their shareholders. Signalling theory thereby facilitates an explanation of the relationship between CSR and corporate reputation. Additionally, signals sent out by companies to their investors are regulated by market

mechanisms. These ensure that only genuine CSR initiatives, as opposed to firm-serving CSR tactics, receive reputational benefits. Therefore, the following hypothesis ensues,

H1: Genuine CSR initiatives, as opposed to firm-serving CSR tactics, will have a positive effect on corporate reputation.

2.5 CSR and corporate reputation from a systemic view

As effects between genuine CSR initiatives, firm-serving CSR tactics, and corporate reputation proves plentiful, Archel et al. (2009), address the need for a wider, systemic view on how firms react to sustainable development demands. The above-mentioned literature specifying positive (Espinosa and Trombetta, 2004; Toms, 2000) and negative (Bebbington et al., 2007; Kim, 2014) effects can be highlighted from a systemic thinking angle. Firms practicing CSR as reputational risk management (Bebbington et al., 2007) or firms managing their CSR as a profit maximizing stunt (Kim, 2014) are rather complying to sustainable development demands than exceeding shareholder expectations. In contrast, firms issuing quality CSR reports (Espinosa and Trombetta, 2004; Toms, 2000) can significantly boost their corporate reputation by fully taking into account shareholder’s sustainable development demands and integrating them into firm strategy.

Along the lines of the above discussion, that attaining a favourable corporate reputation requires additional efforts towards exceeding shareholder expectations, and criticism of legitimacy theory from previous sections, Deephouse and Carter (2005), specify differences between legitimacy and corporate reputation. The authors establish two

viewpoints by which the two concepts differ namely, in their definition and by methods by which they may be accessed. Firstly, according to Andrew (1984) and Parsons (1960),

legitimacy is defined as complying with social expectations, norms, values rules etc. (as cited in Deephouse and Carter., 2005). Shareholders often take legitimate organizations’ actions for granted. In contrast, gaining positive reputation requires firms to exceed shareholder

expectations. Reputation is characterized as the relative standing or desirability of a company in comparison to others (Deephouse and Carter, 2005). Secondly, the two theories differ in terms of how they are assessed. According to Deephouse and Carter (2005), legitimacy is primarily accessed through regulative, normative or cognitive dimensions. However, as stated by Ruef and Scott (1998) reputation can additionally be assessed on ‘virtually any attribute along which organizations may vary that can serve as status comparison’ (Deephouse and Carter, 2005, p.332). Ownership of CSR initiatives may prove as such an attribute along which firms differ. Furthermore, CSR ownership can serve as a status comparison tool for investors, as they may form opinions towards the genuineness of the firms CSR efforts, as demonstrated in the section below (2. CSR ownership). Thereby, gaining a favourable reputation is a lot more complex and time sensitive than achieving a legitimate position with shareholders. As specified by Archel et al. (2009), corporate reputation must be analysed from a systemic point of view to capture these complexities.

Wyland et al. (2012) take the systemic view one step further and predict both positive and negative CSR corporate reputation relationships occurring in different time sequences. According to the authors, firms engaging in philanthropic activities can signal a highly

positive image to their investors, however once they have gained this trust some organizations and individuals may abuse this gained legitimacy (Wyland et al., 2012). Thereby, legitimacy may be a catalyst towards aiding dubious individuals within the organization to commit organizational illegal behaviours (Wyland et al., 2012). As a result, a primarily positive, philanthropic signal, towards stakeholders, can change to a negative reputation over time, in case individuals within the organization abuse their privileges related to increased legitimacy (Wyland et al., 2012). Firms employing inhouse CSR reporting are more likely to suffer from these time sensitive reputational fluctuations, as dubious managers are more likely to have leverage over in-house CSR initiatives than when these are awarded by third-parties.

However, in case third-parties are not so impartial, as mentioned by Jahn et al. (2005) in the section below, this relationship could over time also result in dubious behaviours, negatively affecting an initially positive reputation. Consequentially, the relationship between CSR and corporate reputation must be seen from a systemic point of view to capture the complexity of all positive and negative signals over time and the intricate moderating effect of ownership.

2.6 CSR ownership

As mentioned above, the complex moderating effect of ownership on the relationship between CSR and corporate reputation adds to the systemic point of view of this phenomenon by attempting to capture all its positive and negative signalling effects over time. CSR efforts can be seen a mechanism of self-regulation, in which companies voluntarily disclose

information on social initiatives, in for example their annual report, or by other means of communication (Gray et al., 1995). According to Gray et al. (1995), “CSR, at its broadest may embrace: both self-reporting by organizations and reporting about organizations by third parties” (p. 47). The authors thereby address the dimension of CSR ownership, which this paper seeks to explore, by differentiating between self-reporting and corporate CSR

statements issued by a third party. Other authors criticize this upcoming trend in self-reported (“owned”) CSR. Gray and Herremans (2011), express that, instead of enhancing

accountability, such company owned reports only increase the appearance of accountability. Alternatively, they encourage the development of independent reports namely, ‘external social audits’ (Gray and Herremans, 2011). Other authors examine the transparency of verification statements which appear in published corporate environmental reports and to what extent external parties conducting them are empowered (Ball et al., 2000). Ball et al. (2000) conclude that external verification statements do not add to a company’s CSR report, but solely adds as management tool. As a result, they conclude that external CSR reporting by third parties may not always ensure transparent reporting practices (Ball et al., 2000). When this lack of transparency becomes known publicly, due to the publishing of research such as that of Ball et al. (2000), shareholders will learn to mistrust the empowerment of third parties. Thereby, corporate reputation will not be visibly enhanced for firms employing third party auditors. Furthermore, Jahn et al. (2005) establish “that certification systems are susceptible to opportunistic behaviour” (Jahn et al., 2005, p. 54). The authors draw attention to the fact that companies to be supervised by third parties can choose their own auditor, which may lead to incentives that are misleading and therefore impartiality of outside parties can no longer be guaranteed (Jahn et al., 2005). The paper illustrates the act of receiving a certification label as an act of bargaining, which appears to be irrespective of impartiality.

In contrast to these findings, Birkey et al. (2016), discover a positive relation between assurance and environmental reputation, independent from assurer type. These authors establish that assurance by third parties can enhance corporate reputation and therefore prove as a beneficial addition to company own CSR reporting (Birkey et al, 2016). Researchers have been in disagreement about whether CSR initiatives issued by third parties have a more

most authors agree that ownership acts as a moderator between CSR and corporate reputation, thereby amplifying the effect of genuine CSR on firm reputation.

In conclusion, the relationship between CSR and corporate reputation is far too complex to be seen as a unidirectional linear relationship between only two variables. As a result, authors have argued that in order to observe all complexities and time sensitivities of the relationship researchers must take a systemic perspective on the construct. Ruef and Scott (1998) argue that reputation can be assessed on ‘virtually any attribute along which

organizations may vary that can serve as status comparison’(p.879) (Deephouse and Carter, 2005). Thereby, ownership proves as such an organizational attribute that can be used by investors to judge firms based on their corporate reputation. When firms issue in-house CSR reports or when firms decide on their own third-party auditors, thereby forgoing impartial reputational judgement, there is more opportunities for dubious managers to abuse their leverage (Wyland et al., 2012; Jahn et al., 2005). In contrast, Birkey et al. (2016) argue that assurance by third parties can enhance corporate environmental reputation. In both these cases authors argue that the relationship between CSR and corporate reputation is moderated by ownership. Furthermore, the authors specify differences between genuine and firm-serving CSR initiatives. Therefore, the following hypothesis (H2) proceeds Hypothesis 1:

H2: Ownership moderates and amplifies the positive relationship between genuine CSR and corporate reputation.

Figure 1. Conceptual Model.

CSR Corporate _reputation

3. Methodology

3.1 Method

This research will be conducted by means of quantitative data analysis, specifically an event study and regression analysis, to determine the impact of CSR reporting on corporate reputation moderated by ownership. The event study analysis focuses on the effect a specific corporate CSR event, will have on the corporate reputation of a firm, measured by cumulative abnormal return (CAR). This study employs event study methodology to compute cumulative average daily abnormal returns for firms that announced either owned and awarded CSR initiatives between 1999 and 2016. CAR computations will be performed using the Wharton Research Data Services (WRDS) eventus software. Furthermore, this research makes use of regression analysis to test the direction of the relationship between CSR and corporate reputation (CAR) and test the validity of the moderating effect of ownership on the two variables. The two methodologies will be discussed in more detail in the following sections.

3.2 Empirical Setting

Publicly listed firms in the US, which announced either “owned” or “awarded” CSR initiatives, prove as an ideal setting for this research. According to Carrol (1999) US firms are the obvious choice for CSR research as the majority of research in this field has been

conducted in the states. This data collection focuses on New York Stock Exchange (NYSE) listed companies, in order to make use of WRDS’ event study tool (eventus) and to measure abnormal returns based on stock price. Furthermore, to examine the effect of “ownership” on the relationship between CSR and corporate reputation, firm’s must either qualify for the “owned” or “awarded” initiatives category. This can be achieved by either announcing own CSR initiatives or being “awarded” CSR by a third party. “Owned” CSR initiatives can come

in the form of voluntary CSR reports or in-house certificates, whereas “awarded” initiatives are here classified as environmental or social awards or certifications presented to a

corporation which has fulfilled the unbiased awards criteria.

3.3 Data Collection and Final Sample

Data for this research will be collected manually, as no such data base already exists. In order to conduct the event study (in WRDS eventus), both CUSIP numbers and event dates of each company’s “owned” or “awarded” CSR announcements must be collected. This data will be accumulated by means of research on certification label and corporate websites. The timing of the event dates collected will span from the year 2000 to 2015. After the event study is run, in WRDS eventus, cumulative abnormal return data of each firm will be documented to measure the signalling effect of the “owned” or “awarded” CSR announcement. To collect data for the independent variable this research makes use of KLD “concern” ratings via WRDS which provide a numerical score based on firm corporate social performance.

According to Chatterji et al. (2009), KLD ratings are fairly good at summarizing for example past environmental performance of firms. KLD ratings will be collected for every company which this research has previously performed an event study for. The final data sample will include 87 publicly listed US firms which have either announced “owned” or “awarded” CSR initiatives. Out of this final data sample this research aims at a balanced distribution between companies with “owned” and “awarded” CSR initiatives of about 50/50.

3.4 Dependent variable and event study methodology

Corporate reputation will be measured through cumulative abnormal return,

be used to detect differences in how the stock market responds to the announcement (Jones and Murrell, 2001) of “owned” versus “awarded” CSR initiatives.

The statistical method of an event study helps classify the influence of an event (e.g. announcements of owned/ awarded CSR initiatives) on firm value. Event studies are often run for financial research purposes (Peterson, 1989). Cumulative abnormal returns, gained through the event study methodology proves as a good measure for corporate reputation, as it assumes that capital markets are efficient (Fama, 1998). According to the market efficiency hypothesis all relevant, available information on the market is reflected in stock prices. This is in line with the way by which investors evaluate corporate reputation according to information that is openly available to them. In contrast it is through

information asymmetries that managers, in turn, can use their corporate reputation to their advantage, thereby sending positive signals to their shareholders (Fombrun and Riel, 1997). According to Fombrun and Shanely (1990), firms compete in the market for reputational status, as when newly revealed information is readily available to investors, the stock price instantaneously reacts. This research conducts an event study to measure corporate reputation through cumulative abnormal returns, related to the announcement of owned/awarded CSR initiatives (the event), while adjusting for stock price fluctuations in the market (Jarrel and Poulsen, 1988). In case of a negative abnormal return the

owned/awarded CSR initiative did not create shareholder wealth. A positive abnormal return would signal a positive influence of the CSR initiative on corporate reputation.

Several event study models exist, such as the Constant Mean Return Model, the Control Portfolio Return Model, the Risk-Adjusted Return Model and the Market Model. All these models can be applied to compute a company’s abnormal returns of their stock surrounding the event date of public CSR announcements, the issuance of company-own

CSR reports, or the declaration of CSR awards by third parties. The above-mentioned models differ with regard to market return benchmarks implemented and estimated intervals. Most commonly used is the OLS market model, thereby also implemented in this study. For the OLS market model this study firstly determines and computes normal returns for every firm CSR initiative in the sample. To calculate normal returns the following parameters are estimated over the estimation- also known as clean period (T-200; T-30, proceeding the event window (T-1; T1). The before-mentioned estimation period is a time frame excluding the event. T0 specifies the date of the event. Below, a graphical representation of the timeline around the event (Figure 2):

Figure 2. Event Study timeline

(Estimation Window) (Event Window)

T-200 T-30 T-1 T0 T1

OLS market model normal returns are computed followingly:

NRiτ = αi + βi* Rmτ + εiτ

With: NRiτ being normal return on day τ of the common stock of firm i.

αi and βi represent market model parameters estimated in accordance with an OLS regression analysis.

Rmτ depicts market return in accordance with the market index of the company’s home market.

The abnormal returns of 20 days prior to the announcement of owned/awarded CSR initiatives up until 7 days after the CSR announcement is computed to take into consideration possible information leakage effects. For every firm’s owned/awarded CSR initiative in the sample, abnormal returns are calculated during the (-20,7) event period window:

AR iτ = Riτ – NRiτ = Riτ – (αi – βi* Rmτ)

With: AR iτ being abnormal return of company i’s common stock on day τ Riτ: represents the actual return of firm i’s common stock on day τ

For each day situated in the event period (-20,7), the average abnormal return is calculated by finding the mean of the abnormal returns across the 87 firms in the sample. Abnormal returns of the separate firms are averaged together with the underlying assumption of independence and normal distribution. Finding the mean of abnormal returns reduces noise effects produced by individual stock return outliers in the sample data. Average abnormal returns which are highly different to zero indicate either especially high or low abnormal performances. The formula for the un-weighted cross-sectional average of the abnormal returns in period t reads as follows:

𝐴𝑅_#= 𝐴𝐴𝑅_# = _'& ∑' 𝐴𝑅𝑖𝜏 +,&

The purpose of conducting an event study does not solely concentrate on the stock performance during the event date, however also considers the period surrounding the event.

window, as necessitated by this research. Abnormal returns are summed up from the beginning (τ1) up until the end of the event window period (τ2). The formula for the Cumulative Abnormal Return (CAR) over the event period τ1 to τ2 for company i reads as follows:

𝐶𝐴𝑅₊(𝜏1, 𝜏2) = 𝐴𝑅_{+3 &}+ ⋯ + 𝐴𝑅₊₃₆ = 7 𝐴𝑅𝑖 𝜏 36

3,3&

To ensure reliability and validity of the event study, statistical test(s) must support the study. Thereby, this research takes advantage of Eventus Software by WRDS, to circumvent invalid, unreliable results.

3.5 Independent variable

The independent variable, CSR reporting (CSR) will be measured by means of KLD “total concern” and “total strength” ratings via WRDS. Instead of summing up all positive total strengths and negative total concern scores to achieve one overall KLD score sum for each company (Hull and Rothenburg, 2008), this research will only take three score categories into account. The overall KLD score for each company will be computed by summing up total community performance-, environmental-, as well as employee- strengths and concerns (Brammer and Pavelin, 2006). Bramer and Pavelin (2006), refer to a seminal empirical study by Fombrun and Shanley (1990) in which they measure the impact of social responsiveness on corporate reputation. The authors regard CSR as synonymous to social responsiveness, which they measure through, community, environmental, and employee performance (Bramer and Pavelin, 2006).

3.6 Moderating variable

Ownership will be classified by means of valid, reliable coding. Wang and Berens (2014) differentiate between CSR initiatives otherwise, however their approach of using binary kld ratings, “with a ‘1’ indicating the presence of a particular social action, while a ‘0’ indicates the absence of such an action” (Wang and Barens, 2014), can be translated to this study. A dummy variable will help signify whether a CSR initiative is “owned” 1 or 0 if otherwise (“awarded”). Whether or not a CSR initiative is “owned” or “awarded” will be classified based on whether or not a firm has issued an in-house CSR report or certification label. In case the firm has been awarded a certification label or CSR award by an impartial third party, the initiative is considered to be “awarded”. To test the moderating effect of the ownership dummy on the CSR corporate reputation relationship, this study necessitates the definition of an interaction term (Wang and Barens, 2014), NewCSR*Ownership. This interaction term will help quantify the moderating effect of the categorical moderating variable (“ownership”).

3.7 Control Variables

On basis of previous literature, the following control variables must be taken into account when measuring the relationship between CSR initiatives and corporate reputation, return on assets (ROA), leverage, book to market ratio, idiosyncratic risk, and the corporate governance (CG) Kld score of the respective firms (Cahan et al., 2014) at the time of CSR initiative announcement. These control variables are included, specifically to mitigate the reputational effects caused by more financial evaluations of investors with regards to firm performance. As in Kim et al. (2012) and Guili and Kostovetsky (2014), the CG kld score is incorporated as a control variable instead of being used as a measure of CSR, thereby

the financial control variables were acquired through WRDs Compustat software, such as current share price, tot outstanding shares, net income, tot assets, daily stock prices within the event window, and tot debt. Book to market ratio was calculated by multiplying current share price with tot outstanding shares, while ROA is computed by dividing net income by tot assets. Idiosyncratic risk is calculated by computing the standard deviation of the stock prices within the respective event window:

𝑆𝐷 = :∑(𝑥 − 𝑥)6 𝑁

Leverage calculations are made by dividing tot debt by tot assets. CG kld score were attained by summing up both tot CG strengths and concerns from WRDS kld.

3.8 Empirical Methodology

Data will be analysed by means of a moderated multiple regression analysis. This will result from adding a linear interaction term ‘ownership’ in a multiple regression analysis between CSR and corporate reputation, as seen in Fernández Sánchez et al (2015). This will help determine whether ownership moderates the relationship between the aforementioned dependent and independent variables. The multiple regression should look as follows:

4. Results

4.1 Analytical strategy

First, frequencies have been consulted to exclude any errors existent in the data set. No errors were found while checking frequencies. Missing values were left blank and cases were excluded pairwise as all data collected on the main relationship between the independent (CSR) and dependent variable (Corporate reputation; CAR) were complete. Since data was collected by means of database research, recoding data to account for counter-indicative items does not apply. Second, descriptive statistics, normality tests, skewness and kurtosis were run. Both data for CSR (NewCSR) and corporate reputation (CAR) were normally distributed. In terms of skewness NewCSR lay between 7.89 and 8.37 while CAR skewness was between -56.15 and 57,24. For kurtosis the confidence interval for NewCSR lay between -7.47 and 8.41 and for CAR between -54.83 and 58.47. Thereby, for both NewCSR and CAR skewness and kurtosis lower and higher confidence intervals did not share the same sign resulting in a normal distribution of values for both variables. A reliability test was not necessary to

perform for this research, as data was not collected by means of a survey, but instead through database research. However, from the four statistically significant correlations in the matrix (Table1) one correlation proves as having a medium effect between the two respective variables, namely NewCSR and CG (.398**<0.01), while the rest represent large effects between ROA and booktomarket (.989**<0.01), leverage and booktomarket (.496**<0.01), as well as leverage and ROA (.590**<0.01). Therefore, apart from idiosyncratic risk all the financial control measures are highly correlated with each other. The correlation between NewCSR and CG results from their shared source, namely kld scores.

Table 1 Means, Standard Deviations and Correlations Mean SD (1) (2) (3) (4) (5) (6) (7) 1. CG -.24 1.107 2. booktomarket 1.499 6.800 .025 3. ROA .304 1.472 .023 .989** 4. leverage .476 1.608 -.083 .496** .590** 5. idiosyncraticrisk 6.507 8.658 -.055 -.061 -.072 -.087 6. NewCSR 2.99 4.051 .396** -.123 -.120 -.122 .079 7. CAR -.319 3 28.905 -.141 .021 .028 -.105 .032 -.111 8. ownership .58 .496 .028 .081 .062 -.142 .064 -.008 -.003 ** p < .01

4.2 Hypothesis testing – Hierarchical regression

A Hierarchical regression analysis (Table 2) is performed to test the relationship between the independent variable (NewCSR) and the dependent variable (CAR). This test will determine the ability of NewCSR to predict CAR, and accordingly Corporate reputation, after controlling for corporate governance kld score (CG), booktomarket ratio, return on assets (ROA), leverage and idiosyncratic risk (Cahan et al., 2014). In step one of running the hierarchical regression this research controls for the effects of CG, booktomarket, ROA, leverage and idiosyncratic risks on the dependent variable (CAR) (Cahan et al., 2014). Thereby, the relationship effect of CSR on corporate reputation should be independent of the effects of the three control variables.

The model in step 1 was statistically insignificant F(5, 45) = 1.106; p >.05, therefore we cannot conclude confidently that it explained 10.9 % variance in CAR. Due to statistical insignificance one cannot be sure that most variance of the variable corporate reputation (CAR) is necessarily explained by other factors. In step 2 NewCSR was entered as predictor. This test also proved to be statistically insignificant F(6, 44) = 1.040; p >.05. As a result, it cannot be concluded with certainty that total variance explained by the entire model was 12.4 %. Thus, adding NewCSR and CSRownership to the model only explained an additional 1.5 % in CAR, after controlling for CG, booktomarket, ROA, leverage and idiosyncratic risks if the data would have been significant.

None out of the seven predictor variables were statistically significant in the final model. Therefore, one cannot conclude with confidence that ROA explains by far the most predictive power on corporate reputation (CAR).

4.3 Hypothesis testing – Testing the moderating effect

Finally, the moderating of effect of ownership on the relationship between CSR and Table 2 Hierarchical Regression Model of corporate reputation (CAR)

R R2 R2 Change B SE b t Step 1 .331 .109 .109 1. CG -1.505 3.685 -.059 -.408 2. booktomarket -8.416 5.010 -2.228 -1.680 3. ROA 45.927 25.972 2.523 1.768 4. leverage -8.236 4.159 -.494 -1.980 5. idiosycraticrisk .922 .811 .162 1.137 Step 2 .352 .124 .015 1. CG -3.061 4.115 -.119 -.744 2. booktomarket -8.573 5.028 -2.270 -1.705 3. ROA 47.057 26.081 2.585 1.804 4. leverage -8.290 4.172 -.497 -1.987 5. idiosyncraticrisk .884 .815 .155 1.085 6. NewCSR 1.069 1.244 .137 .860

Table 3 and 4. Due to an insignificant p-value this study cannot confirm that ownership as a moderator explains 1.26 % of the variance in CAR. One also cannot judge over the fact that this model explains far less variance than the previous one. In addition, none of the interaction variables prove statistically significant, whereby this research cannot conclude on any

moderating effects. For a level of confidence of 95% there an insignificant (P =.8837>0.05) positive moderating effect of ownership (P =1.228, p>0.05) on the relationship between CSR and corporate reputation. Due to insignificant statistical results this research cannot conclude an increase of R2 by 0.0003. Both owned and awarded CSR ownership models demonstrate insignificant p-values, thereby this study cannot conclude a significant moderating effect on the relationship between CSR and corporate reputation.

Table 3 Analysis of the moderating Effect of ownership

Coefficient SE t p Intercept i1 -.185 4.874 -.038 .970 NewCSR -4.108 6.184 -.664 .508 Ownership -.228 6.392 -.035 .972 NewCSR*Ownership 1.23 7.204 .171 .865 R2=.013 p >0.05 F(3,82)= .363

Table 4 Conditional effect of NewCSR (X) on CAR (Y) at levels of Ownership (M)

Effect SE t p

Owned -2.880 3.262 -.883 .380

Awarded -4.108 7.702 -.533 .595

In conclusion, as all regression analysis resulted in insignificant statistical findings this research failed to reject both H1 and H2. In order to exclude a type two error, this research proceeds with a test of statistical power (Cohen, 1992). Type two error may occur in a

detecting the true variance of the overall model (Cohen, 1992). For all three regressions both a post hoc and a priori power analysis was performed. The post hoc power analysis helps to determine the influence power, reflecting the probability to detect the true difference of the variable relationship. The a priori power analysis specifies a power probability of 80%, in this case, and determines the appropriate sample size to achieve the prespecified power

probability. This research makes use of the G*Power calculator to perform said power analysis.

For the first Model (Model 1), a regression was run to control for the effect of five control variables. The post hoc power analysis determined an influence power of 67.47 %, meaning that there is a 67,47% chance the researchers will be able to detect the true effects of the control variables on the dependent variable, in a sample with 87 observations (Cohen, 1992). Thereby, there is a fair chance (67.47 %), that if results were significant, the H0 could be rejected and therefore, this study could detect a significant influence of the control

variables on the dependent variable. In future research, a sample of 111 observations should be collected to reach an influence power of 80%, as determined by the a priori power analysis.

The second model determined a post hoc influence power of 71.17 %, slightly higher than for model 1. This means that the probability of determining the true effect of the

relationship between CSR and corporate reputation (CAR), after controlling for the 5 control variables, is at 71.17% (Cohen, 1992). Although statistically insignificant, results collected in this sample may be of use to predict the true effect of CSR on corporate reputation, thereby rejecting the H0 and suggesting an effect of CSR on corporate reputation after controlling for the 5 control variables. However, in order to achieve an 80% probability of determining the correct effect, future research should collect at least 103 observations. For both models 1 and

2 there are low probability of type 2 error, which allows more liberal interpretations of insignificant data rejecting the H0.

Contrastingly, for the final regression model, testing the moderation effect of

ownership, the post hoc power test only computed an influence power of 13.91%. Therefore, there is only a 13.91% (1-b) chance of detecting the true moderation effect of ownership on CSR and corporate reputation in a sample of only 87 (Cohen, 1992). Since there is a high chance of type two error (b) it is highly likely that data analysis would falsely presume that there is no moderation effect of ownership on CSR and corporate reputation (type two error). As a result, one must be careful when interpreting insignificant results not to falsely retain the H0 that there is no moderating effect. In this instance researchers must be weary of falsely interpreting data and are best advised to increase the sample size.

5. Discussion

This research has examined the effect of CSR on corporate reputation, from a systemic point of view, through the moderation effect of ownership. A sample of 87 US firms was collected which either launched in-house CSR reports/certification labels (owned) or were awarded certification labels/CSR awards by impartial third-parties. The independent variable, CSR, was measured through kld score ratings retrieved from wrds. Specifically, this study concentrated on social responsiveness measured by community-, environmental-, and

employee- performance. A single aggregate social responsiveness, kld score was achieved by summing up the respective strength totals and subtracting social responsiveness total concern ratings (Bramer and Pavelin, 2006). The dependent variable, corporate reputation, is

measured through cumulative abnormal return (CAR) which has been computed by means of an event study. This research took advantage of wrds’ eventus eventstudy software which

measured how the stock market responds to announcements of owned/awarded CSR

initiatives. The following variables were controlled for in this setting, return on assets (ROA), leverage, book to market ratio, idiosyncratic risk, and the corporate governance (CG) Kld score of the respective firms (Cahan et al., 2014). These control variables helped ensure that financial evaluations of investors, with regards to firm performance, did not influence reputational effects too heavily. Hypothesis testing was done by means of hierarchical regression analysis. First, control variables were regressed against corporate reputation to determine their effect on corporate reputation. Second, a regression was run between CSR and corporate reputation (CAR), controlling for control variables, mentioned above. Third, an interaction term was added to the regression equation, NewCSR*Ownership to establish the moderating effect of ownership on the relationship between CSR and corporate reputation. Fourth, due to statistical insignificance a post hoc power analysis was performed to determine the faults in the research design.

This study has not brought forth many significant findings, however it spurs on discussions about systemic viewpoints of corporate reputation as well as provides methodological

direction for future research. The correlation matrix (Table 1) has brought forth four statistically significant correlations, namely a medium effect between NewCSR and CG (.398**<0.01) and large effects between ROA and booktomarket (.989**<0.01), leverage and booktomarket (.496**<0.01), as well as leverage and ROA(.590**<0.01). These findings demonstrate that all financial control measures, apart from idiosyncratic risk are highly correlated with each other. NewCSR and CG are derived from a shared source, namely kld scores, which explains their moderate correlation. The correlation between NewCSR and CAR, however, proves statistically insignificant, from which we cannot conclude a negative correlation coefficient (-.111>0.01). In model one (Table 2), of the hierarchical regression analysis, this study could not find statistically significant results F(5,45) = 1.106; p >.05,

therefore no conclusions could be drawn on whether and by how much the control variables explained significant variance in CAR. Similarly, adding NewCSR and CSRownership as predictors into multiple regression model two (Table 2), did not result in significant statistics F(6,44) = 1.040; p >.05. Thereby, no statistical patterns in CAR may be detected with regards to the explanation power of NewCSR and CSRownership, after controlling for CG,

booktomarket, ROA, leverage and idiosyncratic risk. None of the seven predictor variables proved statistically significant in the third model (Table 3). No predictions could be made with regards to which variable had the most predictive power on corporate reputation (CAR). After testing the moderation effect of ownership on the CSR corporate reputation (CAR) relationship, this study concluded statistical insignificance. As a result, no moderation effect can be detected. Neither owned, nor awarded CSR initiatives have a significant effect on corporate reputation in this study. After the majority of results were proven statistically insignificant, the post hoc power analysis proved that for model one, there was an influence power of 67,47. This translates to a 67,47% chance, for a sample of 87 firms, that researchers will be able to detect the true effects of the control variables on the dependent variable

(Cohen, 1992). This means that there is a fair chance, although these results deem

insignificant, that the control variables chosen in this research are appropriate for the model, since they are likely to have a significant influence on the relationship between CSR and corporate reputation (CAR). For the second model the post hoc power analysis predicted a 71.17% probability of determining the true effect of the relationship between CSR and corporate reputation (CAR), after controlling for the five control variables (Cohen, 1992). This tendency suggests that there is a likelihood that in a different sample CSR may have a positive effect on corporate reputation, since although insignificant the slope of NewCSR (b =0.137>0.05) proves positive. For the final regression model, the post hoc power test only computed a 13.91% influence power. Accordingly, there is a rather low chance of detecting the true moderation effect of ownership on CSR and corporate reputation (CAR). It is very

likely for data analysis to falsely accept the H0 that there is no moderation effect (type two error). In conclusion, to correctly determine this relationship researchers would be best advised to increase the number of observations substantially.

To analyse the occurrence of statistically insignificant results, this study performs both post hoc and a priori power analysis. The first analyses the probability of attaining statistical results portraying the correct relationship between the observed variables (influence power), while the second determines the correct sample size for future research by specifying an 80% influence power. For models one and two the post hoc power test computed chances of attaining the correct relationship results, when sampling a different set of 87 company

observations, at 67.47% and 71.17% respectively. These probabilities of observing the correct relationships between corporate reputation (CAR) and the control variables, as well as the influence of CSR on CAR when controlling for the control variables, deemed relatively high. Furthermore, the a priori tests for the two models concluded that in order to gain an 80% influence power future research must collect 111 observations for model one and 103 for model two. Since, the post hoc tests showed relatively high influence power results and the a priori power test, at 80% influence power, only asked for up to 24 more observations (28% more) insignificance is more likely traced back to missing control variables rather than lack of observations. Possibly, more financial measurements could add to the impact of control variables on the relationship between CSR and corporate reputation (CAR). Furthermore, other measures influencing investor perceptions of firm reputation would add to the study. Model three, however did not score very high on influence power. There is only a 13.91% chance of collecting the correct results for the moderation effect of ownership on the relationship between CSR and corporate reputation. Therefore, model three suffers from a high chance of type two error, namely, researchers are likely to falsely keep the H0. This would mean that it is likely for this study to conclude that there is no moderation effect of

ownership on the relationship between CSR and corporate reputation (CAR), although this effect does in fact persist. In order to reduce the chance of suffering from a type two error future research must collect a data set with at least 308 observations to ensure an 80% chance of observing the correct moderation effect. For the third model it deems advisable to collect substantially more data to enhance the probability of capturing the true moderation effect of ownership on the CSR corporate reputation (CAR) relationship. The limitations of this study are the predominantly insignificant statistical results which obstruct the researcher from interpreting the results from the three regression models. This statistical insignificance resulted from the small sample size and missing control variables which might further influence effects of CSR on corporate reputation (CAR). Findings may only be analysed through the lens of power analysis as it gives the observer an indication of how likely the correct effect could be observed if future research were to collect 87 new observations under similar methodological conditions. As a result, findings should not be seen as significant results, yet merely as directional gauges for future research. Limitations of this research have hugely impacted the results, as lacking control variables and missing observations heavily influenced the statistical validity of this study.

Findings of this study, although primarily insignificant, help to path the way for researchers looking into examining the relationship between genuine CSR and corporate reputation (CAR) in the future. The ownership moderation effect provides an interesting viewpoint for further research in this field as it adds a more systemic point of view to research on the relationship between CSR and corporate reputation. However, with this systemic view comes complexities regarding the addition of multiple control variables to account for multiple indirect influences within the system. Therefore, this study provides researchers with a starting point by reviewing related literature, providing a methodological framework for

future studies and by illustrating the complexity of obtaining significant results in such a systemic field of research.

6. Conclusion

This research set out to discover the effects of CSR on corporate reputation as well as the moderating impact of ownership. A positive relationship between the independent and

dependent variables was hypothesized, meaning that if CSR scores increase positive corporate reputation evaluations will follow. Whereas firms with low CSR levels are more likely to suffer from a negative firm reputation. Furthermore, the moderating variable, ownership, was expected to strengthen this positive relationship. CSR was measured through kld social responsiveness scores, which were computed by summing up total community,

environmental, and employee performance strengths and subtracting their respective total concerns. Corporate reputation was observed through cumulative abnormal returns (CAR), computed by means of an event study. The event study, conducted in wrds’ eventus software calculates cumulate abnormal returns by comparing market returns with abnormal returns from the announcement day of the owned or awarded CSR initiative. Thereby, the CAR detects the differences in how the stock market responds to the announcement (Jones and Murrel, 2001) of owned/awarded CSR initiatives. In accordance with the market efficiency hypothesis (Fama, 1998), CAR is a sufficient measure of corporate reputation. The market self regulates itself whereby firms that don’t perform with regards to CSR suffer from lacklustre stock returns. Investors evaluate firms based on openly available market

information. In turn, these evaluations of corporate reputation are reflected in stock prices, regulated by the market efficiency hypothesis (Fama, 1998). The following control variables are controlled for when statistically testing the relationship between CSR and corporate reputation, return on assets (ROA), leverage, book to market ratio, idiosyncratic risk, and the

corporate governance (CG) Kld score of the respective firms (Cahan et al., 2014). Financial measures of firm performance are controlled for in this setting to avoid reputational

evaluations based on financial performance. Furthermore, Cahan et al. (2014), suggest controlling for CG scores from kld. Reputation is influenced through many firm characteristics and available public information about the firm’s actions. It is deemed important to account for multiple influences of corporate reputation in this model, which could affect the relationship between CSR and corporate reputation. Correlations between independent, dependent, and control variables were observed. Except for idiosyncratic risk all other financial measures were significantly correlated to each other. Furthermore, the only other statistically significant correlation deemed CG and CSR which are related on basis of a common source, namely kld scores via wrds. Thereby these correlations do not come to a surprise. However, the major relationship in this research, namely CSR and corporate

reputation (CAR), did not show significant correlation results. A regression analysis was run with three separate models. Firstly, the control variables were regressed against CAR to specify whether the financial measures and CG sufficiently influence corporate reputation (CAR). The results of this first model prove as insignificant F(5,45) = 1.106; p >.05, whereby no judgement may be passed over the relationship between the control variables and CAR. Secondly, a regression analysis was run between CSR and CAR, while controlling for the control variables. Similarly, this regression analysis was statistically insignificant F(6,44) = 1.040; p >.05. Thirdly, the regression model proved insignificant, whereby the moderation effect of ownership was introduced by adding the interaction term CSR*ownership. The results of this third model once again demonstrated insignificant results. At the 95% confidence level there is an insignificant (P = .8837>0.05) positive moderating effect of ownership (P =1.228, p>0.05) on the relationship between CSR and corporate reputation. Since the majority of the results from this research demonstrated insignificant results, this study performed a post hoc power analysis to determine the probability of attaining