Does Legitimacy Matter?

(1)

- 0 -

Does Legitimacy Matter?

A Study of Corporate Social Responsibility Engagement

Among Dutch Royal Firms

Master Thesis M.T. Krijnen Student number: s2197332

Hekmeijerstraat 42 3572 WG Utrecht Tel.: +31(0)6-51694367 E-mail: m.t.krijnen@student.rug.nl

Program: MSc. Accountancy & Controlling: both tracks Supervisor: dr. N. Hussain

Second Assessor: dr. Y. Karaibrahimoglu Coordinator: dr. K. Linke

Word count: 11,234

ABSTRACT: This study explores the effects of firm legitimacy, in the form of royal status, on its corporate social responsibility (CSR) engagement and financial performance, from a legitimacy and stakeholder theory perspective. I hypothesize three statistically testable relationships. To test these, a unique CSR measure is used to operationalize the level of CSR engagement among public as well as non-public Dutch firms. This research shows that royal firm status exerts positive pressure for responsible corporate practices. Royal firm status lowers the amount of irresponsible corporate behaviour measured by the amount of tax avoidance. I conclude that putting emphasize on morality of firms is successful among Dutch royal firms. I argue that, this research has very useful policy implications for governments and policy makers to induce responsible corporate behaviour for the achievement of sustainable development goals.

(2)

1. INTRODUCTION

The general public considers Corporate Social Responsibility (CSR) as very important to businesses (Brooks, 2016). The literature considers legitimacy and stakeholder pressure to be determining factors in CSR engagement (Hoi, Wu & Zhang, 2013; Hooghiemstra, 2000; Margolis & Walsh, 2003; Yu & Choi, 2014). The CSR engagement of firms is part of the norms, values and beliefs of the environment they work in. Consequently, organisations will report about CSR according to these norms values and beliefs, this is the determining factor for firm legitimacy. It this signals that stakeholder and legitimacy pressure can provide reasons for firms to adopt CSR practices (Freeman, 2002; Suchman, 1995). At the same time, academics as well as governments and policy makers have largely ignored the role that legitimacy can play in stimulating firm engagement in CSR (Preuss, 2010).

Dutch companies that exist for at least 100 years, are of national importance, have high moral character, are financially stable, and avoid everything that could harm their reputation, can be qualified with the prestigious royal firm status (Predikaat Koninklijk, 2016). Royal firms have to satisfy high moral norms, values, and beliefs of stakeholders in order to not jeopardize their royal status and to achieve firm objectives. This signals the importance of legitimacy for royal firms and creates necessity to put additional emphasis on corporate social behaviour.

Firms use different tactics to attain legitimacy, they can make actual changes in their CSR engagement and inform stakeholders, they can use CSR as a window-dressing strategy to cover negative practices or they can manipulate stakeholders’ perceptions while not actually engaging in CSR (Hooghiemstra, 2000; Lindblom, 1994). There is a stream of literature that states that firms should not engage in CSR at all, since the only legitimate responsibility of business to their stakeholders is to increase profits (Friedman, 1970).

Knowledge about the tactics that royal firms use, covering positive and negative CSR engagement, provides evidence regarding the effects of legitimacy on CSR engagement. At the same time, evidence regarding the implications of CSR on firm performance can remove investor scepticism towards CSR and consequently resolve the tension that management accountants experience regarding the social and economic trade-offs in the company they manage (Devinney, 2009; Dowling, 2014; Kiessling et al., 2015; Margolis & Walsh, 2003; Zhu et al., 2014). Ultimately, this can provide useful policy implications for governments and policy makers to induce responsible corporate behaviour for the achievement of sustainable development goals. Taken together, the aim of this study is to investigate the effect of

(4)

legitimacy, in the form of royal firm status, on CSR engagement and firm performance, from a legitimacy and stakeholder theory perspective.

The aim of this study will be pursued by using a study of archival data, by utilizing a unique sample set of royal and non-royal Dutch firms. This sample contains data over the period 2001-2015 regarding positive CSR engagement, measured as the CSR score, negative CSR engagement measured by tax avoidance and firm performance. Considering the limitations of operationalized measures I use multiple measures of CSR, tax avoidance and firm performance to increase the generalizability of the results (Dyreng et al., 2008; Graham et al., 2014; Hoi, Wu & Zhang, 2013). Following Halbritter & Dorfleitner (2015) I combine different measures of CSR into a unique comprehensive CSR score that covers the ESG scores of ASSET4, Bloomberg and the Dutch Transparency benchmark. Following prior tax avoidance literature, tax avoidance will be measured by the ETR and Cash ETR for 1, 3 and 5 years (Dyreng et al., 2008; Hoi, Wu & Zhang, 2013). Firm performance will be captured by two measures, Return on Equity (ROE) and Return on Assets (ROA).

I find that Royal firm status exerts positive impact on responsible CSR practices and it lowers the amount of irresponsible CSR practices in terms of tax avoidance. The results for firm performance are against my expectations, CSR slightly lowers firm performance.

This can be explained by the argument that positive CSR associations might not always increase profitability immediately, the responsible CSR practices of the companies in my sample can currently go unnoticed by the general public, but they do impose costs. Only when the irresponsible CSR activities of competitors will be noticed, and their external legitimacy will be damaged, this will create a positive differential effect for companies with truly responsible CSR practices. The goal of the firm performance research, is to contribute to resolving the tension that management accountants experience regarding social and economic trade-offs, by showing that creating mutual value for all stakeholders leads to higher firm performance (Devinney, 2009; Dowling, 2014; Kiessling et al., 2015; Margolis & Walsh, 2003; Zhu et al., 2014). This study does partially contribute to the research gap, by providing a new point of view regarding this relation. The public attention for CSR practices is relatively new, and irresponsible CSR activities might not be noticed yet. This information can help management accountants in the ongoing debate regarding the mutual value that CSR creates for all stakeholders. I leave this research question for future studies to explore in more detail.

This study is related to a growing stream of literature on the success factors of CSR (Hoi, Wu & Zhang, 2013). It provides evidence regarding, the effects of legitimacy, in the form

(5)

of acknowledging firm status, on CSR behaviour and success. The call for further research about scepticism towards CSR and its manipulative purposes will be answered by the association between royal status and negative CSR behaviour, in this case, tax avoidance. The evidence regarding the negative association between royal status and negative CSR practices is useful for auditors, since it sheds a positive light their assurance practices. Auditors are expected to critically assess possible manipulation strategies of their clients and prevent negative CSR contributions in firms with CSR minded core values (Lindblom, 1994). The legitimacy of these CSR audits is constantly threatened by the misalignment of expectations about the auditing system. This study provides indications regarding the legitimacy of CSR reports, since firms that formally state they are of high moral character, show less irresponsible CSR practices.

Ultimately, I conclude that this study contributes to the current knowledge about the legitimacy effects on CSR and putting emphasize on the morality of firms has showed its success among Dutch royal firms. I argue that this research has very useful policy implications for governments and policy makers to induce responsible corporate behaviour for achieving sustainable development goals.

The structure of this paper is as follows. First, I provide the theoretical background of the research topic. Three hypotheses will be developed regarding legitimacy, Royal firm status, CSR, tax avoidance and firm performance. Second, I describe my research methodology, including the sample selection, variable measurement and the regression models. Third, the results will be described, including the summary statistics. Finally, the discussion and conclusion will be presented, together with the limitations and directions for future research.

(6)

2. LITERATURE REVIEW AND HYPOTHESIS

DEVELOPMENT

Prior studies consider legitimacy and stakeholder theory as key concepts in CSR literature (Preuss, 2010). In this section I provide a theoretical overview of these theories as well as a review of other relevant studies. This is used to lay out three arguments in detail that show in which way legitimacy, in the form of royal status, enhances CSR engagement and financial performance.

Legitimacy theory is explained by Talcott Parsons (1960) as ‘‘the appraisal of action in terms of shared or common values in the context of the involvement of the action in the social system’’(p.175). Suchman (1995) further specifies this in the context of business management. He describes legitimacy as the assumption that the actions of a firm are appropriate within the socially constructed system of norms, values and beliefs that the firm operates in. This induces firms to disclose information which enables the society to assess whether they are good corporate citizens (Guthrie & Parker, 1989).

Stakeholder theory identifies the internal and external stakeholders that are important to organisations and recommends methods to handle the interests of those groups (Freeman, 2002). Accordingly, a stakeholder is a group or individual that can effect or is effected by the achievement of firm objectives. Common stakeholder groups include investors, regulators, customers, employees, pressure groups, the natural environment and the community (Mishra & Suar, 2010). The achievement of firm objectives is a key issue to their survival, which depends on the ability of the organisation to sufficiently satisfy the stakeholders. This means that the difference between social and economic objectives is no longer relevant to the company, since these are both relevant to stakeholders (Yu & Choi, 2014). Stakeholders are a driving force in both stakeholder and legitimacy theory. Therefore, these theories may be viewed as overlapping perspectives (Hooghiemstra, 2000).

Following Caroll (1979) I define CSR as “The social responsibility of business encompasses the economic, legal, ethical and discretionary expectations that society has of organizations at a given point in time” (p. 500).

Prior literature identifies several reasons why firms should or should not adopt CSR practices, which can be explained by legitimacy and stakeholder theory. According to Suchman (1995), the CSR engagement of firms is part of the norms, values and beliefs of the environment they work in. Consequently, organisations will report about corporate social behaviour according to the expectations of their stakeholders to enable them to accomplish their

(7)

objectives. The above signals that stakeholder and legitimacy pressure can provide reasons for firms to adopt CSR practices (Freeman, 2002).

There is another stream of literature that is more sceptical about CSR engagement trough legitimacy perspectives. Friedman (1970) argues that social involvement is not a corporate responsibility at all and argues that the only social responsibility of business is to increase its profits. Although corporate managers can feel personally involved with social issues, it is never legitimate to spend someone else’s money. According to the principles of the free market, social cooperation of individuals should always be voluntary and corporate social practices should only be adopted when they benefit shareholders or the company objectives. Mishra & Suar (2010) provide a similar argument, they state that it is complicated to satisfy stakeholders by investments in CSR. Stakeholders have different interests regarding to CSR, which makes it very challenging to make legitimate decisions in which CSR practices to adopt. To give an example, investors are mainly concerned about governance aspects, while employees consider policies towards the elimination of child labour and participation in decision making, the community is interested in education and health, while environmental protection and CO2 reduction is of the main interest of environmental parties.

Gray, Owen & Maunders (1998) describe three reasons why firms should engage in CSR, it enhances corporate image, it discharges firms accountability with a social contract and provides additional information to investors. The social contract argumentation considers the society to be a provider of attributes and resources for firms to exist, and therefore, the societal benefits of the firm should exceed the societal costs. When society perceives the firm to be operating in accordance with their norms and values, legitimacy is accomplished (Deegan, 2002; Mäkelä & Näsi, 2010). When a firm breaks the social contract, a so called legitimacy gap exists. They will take action to retain and recover their legitimacy according to the level of pressure from the society to be legitimate, since the iron law of responsibility states that companies who are not perceived as legitimate will lose their position in the long run (Davis, 1973).

The above argumentation describes reasons to engage in CSR practices, in addition to this, Lindblom (1994) names four strategic options that describe the way in which legitimacy seeking firms adopt CSR practices. First, when a legitimacy gap arises, the firm will seek to educate and inform its stakeholders about important organizational changes, which can or cannot be actual changes. Second, when stakeholders misperceive the firm’s CSR engagement, the firm may seek to change the perceptions of the stakeholders, while not changing its actual

(8)

behaviour. Third, the firm may seek to manipulate the stakeholder perception by deflecting attention away from negative issues in terms of CSR. Fourth, when stakeholders have unrealistic expectations of the responsibilities of the firm, the firm may seek to change unrealistic aspects about expectations of its CSR performance.

Overall, many authors find support for the relationship between legitimacy theory, stakeholder theory and CSR (Freeman, 2002; Guthrie & Parker, 1989; Hooghiemstra, 2000; Lindblom, 1994; Mäkelä & Näsi, 2010; Mishra & Suar, 2010; Suchman, 1995; Yu & Choi, 2014). Previous findings from the literature indicate that both theories are important in explaining CSR behaviour. Therefore, these will guide my hypothesis development regarding the effects of Royal firm status on CSR.

2.1 Royal status and Corporate Social Responsibility Engagement

The King of the Netherlands may decide to qualify a corporation, association or foundation as Royal. To become qualified, a firm should exist for at least 100 years, be of national importance and have an important position in their field. In addition to that, royal firms must be of high moral character, everything that could harm the royal reputation must be avoided and the firm should be financially stable, otherwise the royal status will be removed (Predikaat Koninklijk, 2016). Together this forms a very prestigious status in the Netherlands.

Dowling & Pfeffer (1975) state that the importance of organizational legitimacy effects some firms more than others. It is especially important to firms that are considerably visible or depend relatively more on social and political support. Stieb (2009) argues that the norms and values of stakeholders will be important when their pressure has sufficient impact on the firm. As Royal firms are of national importance, have an important position in their field and cope with additional stakeholder pressure to keep their reputation immaculate, I expect legitimacy to be important to royal firms. The arguments of Suchman (1995) and Freeman (2002) support this statement. Royal firms have to satisfy high moral norms, values and beliefs of stakeholders in order to not jeopardize their status and to achieve firm objectives. This signals the importance of legitimacy for these firms and creates necessity to put additional emphasis on corporate social behaviour. According to the above stated argumentation, royal status is used to explore the effects of legitimacy on CSR engagement of firms.

The arguments of several authors can be applied to explain the CSR engagement of royal firms. In order to be faithful to the royal status, firms should always consider the long run social effects of their actions. According to the iron law of responsibility it is important for

(9)

royal companies to put additional emphasis on their social contract to not jeopardize their position in the long run (Davis, 1973). Lastly, a CSR-oriented culture as defined by Yu & Choi (2014) can enhance the level of CSR engagement. This culture is defined as “an organizational-wide consensus among all organizational members around a set of shared assumptions, values, and beliefs related to CSR” (p. 228). High moral character is expected to be an important core value of royal firms, accordingly a social responsible mind-set that enhances CSR engagement is expected to be embedded in their culture (Predikaat Koninklijk, 2016).

Following from the above, I expect Royal firm status to positively influence CSR engagement in the search for legitimacy. I draft the following hypothesis.

H 1: Legitimacy, measured as royal firm status, exerts a positive effect on the CSR score of a company.

2.2 Legitimacy and Negative CSR activities: Tax Avoidance

In the search for the effects of legitimacy on CSR behaviour of firms, not only positive CSR information is relevant, negative aspects provide information as well. Responsible CSR activities are sometimes considered to be too self-serving and tainted to provide legitimate information, while irresponsible CSR information contains additional explanatory power regarding the underlying CSR construct they intend to capture (Hoi, Wu & Zhang, 2013; Lindblom, 1994).

Legitimacy can be a resource that a firm can manipulate and it can help them to conceal their negative contribution to society (Lindblom, 1994; Mäkelä & Näsi, 2010). This can be achieved by CSR programs that are decoupled from their core business, which represent a form of window dressing and are solely implemented to give the appearance of legitimacy to satisfy external parties (Maclean & Benham, 2010). In that case, CSR information and publications can be used to cover negative contributions on the social or environmental domain (Deegan et al., 2000; Deegan et al., 2002; Gray et al., 1995a and Hooghiemstra, 2000). Irresponsible information is less vulnerable for legitimacy purposes.

Investigating irresponsible CSR information is interesting to external auditors. It is their goal to provide assurance on the accuracy, quality and credibility of CSR information (Goldberg et al., 2011). This includes a verification that the objectives and behaviour of the firm are in line with their overall strategy, in this way the auditor adds legitimacy to the reporting process of the firm (Power, 2003). Cohen & Simnett (2015) argue that verified CSR

(10)

reports are less labelled as ‘greenwashing’ by sceptics. In contrast, others argue that CSR assurance struggles for legitimacy (Gray, 1995; Power, 2003). The legitimacy of CSR audit is constantly threatened by the misalignment of expectations about the auditing system, through the absence clear definitions of audit quality. In fact audit quality is ‘unobservable’ therefore, Power (2003) suggests to focus audit research more on the credibility and common sense of CSR assurance, rather than on the procedures, structure or eﬃciency. Evidence regarding negative CSR engagement provides information about deviations between formal firm objectives and their actual CSR behaviour. Auditors are expected to critically assess possible manipulation strategies of their clients and prevent negative CSR contributions in firms with CSR minded core values. This can provide indications regarding the legitimacy of CSR reports, which can signal the quality of auditor’s services, while not using confidential client data, or unsubstantiated assessment criteria. Alternatively, it could highlight additional points or interests in the audit engagement.

Tax avoidance is according to Dyreng et al. (2008) broadly anything that reduces the firm’s cash effective tax rate over a long period of time. Taxes have an important community and societal implication, they fund public goods and services such as education, national defence, public health care, public transport and law enforcement (Lanis and Richardson, 2013; Freedman, 2003). Taken together, tax avoidance can be considered not right from the social contract concept because it imposes costs on society (Dowling, 2014). Tax payment is a clear measure of direct financial contribution by a firm to the government and to society. It is a fundamental and easily measured example of a firm’s citizenship behaviour. Taken together, tax avoidance is a clear example of irresponsible CSR activities and social responsible firms will pay their fair share of tax (Hoi, Wu & Zhang, 2013; Lanis & Richardson, 2013).

Hoi, Wu & Zhang (2013) find that corporate culture effects irresponsible CSR activities. Zeng (2016) also finds that socially responsible ﬁrms are less likely to avoid taxes. A manipulation strategy of providing responsible CSR information to cover for irresponsible CSR information does not fit the high moral character of a Royal firm. As stated before, I expect Royal firms to perceive additional pressure to retain their legitimacy. They are also expected to have a CSR minded corporate culture. Together with our expectation that CSR engagement of royal firms is part of the norms, values and beliefs of the environment they work in, I draft the following hypotheses.

(11)

2.3 Legitimacy and Firm Performance

The association between CSR and firm performance does currently receive considerable attention in CSR literature (Margolis & Walsh, 2003; Zeng, 2016). However, there is currently no unanimity regarding the direction of the association.

A Management Accountant or in Dutch, Controller, is someone who combines financial expertise and business acumen to guide critical business decisions and to achieve business success (CGMA, 2016). Management Accountants experience a tension when they have to make decisions which have consequences for the economic as well as for the social domain (Devinney, 2009; Dowling, 2014; Margolis & Walsh, 2003; Zhu et al., 2014). This is caused by an ambiguous and complex association between CSR and firm performance, as well as by divergent objectives of stakeholder groups (Crifo et al., 2016; Mishra & Suar, 2010; Zhu et al., 2014). High profits, long term sustainable growth and growing share prices are important to investors, while other stakeholders consider social and environmental issues to be more important. The disunity in literature can be illustrated by the views of Friedman (1970) and Mishra & Suar (2010), Friedman (1970) argues that the only social responsibility of business is to increase its profits. Mishra & Suar (2010) use an instrumental orientation towards CSR, to address the tension within stakeholder groups. They argue that CSR is an instrument that can be used to align the social goals of organisations with the business goals. The usage of CSR as a strategic tool can enhance the economic performance. This is explained by the statement that CSR induces customer goodwill and by the signalling theory that states that CSR signals higher organisation reliability. More evidence regarding a positive relationship between CSR and firm performance could resolve the tension that management accountants experience, and improve the acceptance within companies to invest in CSR.

Different arguments, derived from legitimacy and stakeholder theory, are used to explain the effects of CSR on firm performance. One of them states that CSR causes enhanced firm reputation through favourable stakeholder attitudes and strengthened stakeholder– company relationships, which causes increased firm performance (Du et al., 2010; Zeng, 2016). In addition, Kiessling et al. (2015) argue that firms who differentiate themselves from its competitors by means of CSR behaviour, can raise additional profits. However, positive CSR engagement has only marginal direct effects on reputation and consequently on firm performance, while negative CSR practices may lead to severe negative sanctions such as loss of ﬁrm reputation, potential ﬁnes, penalties, and even consumer boycott (Hoi, Wu & Zhang,

(12)

2013). Firms prevent themselves from these negative sanctions by using CSR as an insurance policy (Klein & Dawar, 2004). Positive CSR associations may be instrumental in reducing the risk of damage to the firm in case of negative publicity.

The CSR spill over effect can play a substantial role in the suggested association between CSR and firm performance as well. It explains the bias that CSR engagement causes that spills over to otherwise unrelated consumer attributions. Strong consumer beliefs about the social performance characteristics of a firm tend to spill over onto beliefs about its product quality and ultimately on firm performance (Klein & Dawar 2004). Finally, social identity theory emphasizes that one’s self-image is influenced by the reputation of organisations in which you are a stakeholder. Stakeholders prefer to be associated with socially responsible organisations and they avoid companies with irresponsible behaviour. Socially responsible organisations will therefore be more attractive to investors, which influences firm performance in a positive way (Mishra & Suar, 2010).

Despite the arguments above, there is currently no unanimity regarding the direction of the association. Enhanced firm reputation, differentiation strategies, insurance benefits, positive spill over effects, and social identity theory, indicate a positive association between CSR and firm performance. However, all arguments are based on the assumption that competitors behave in a different way, which will be noticed by the stakeholders. A sufficient timeframe of the study is critical to meet this assumption, since it takes time before observations of negative CSR practices are translated into altered firm performance. Based on the extensive time frame used in this study, I draft the following hypothesis.

H 3A: There is a positive association between CSR and firm performance.

The above stated arguments are widely supported in research, nevertheless there are researchers that are sceptical about the relationship between CSR and firm performance. Klein and Dawar (2004) assert that stakeholders are increasingly sceptical and sophisticated about corporate behaviour that appears to be socially responsible. This makes a positive association between CSR and firm performance dependent on the believe in the right motivations to engage in CSR (Maignan & Ralston, 2002). The findings of Yu & Choi (2014) complement this argumentation, they state that a CSR minded firm culture is critical for raising additional profits from CSR.

(13)

Royal firms have an excellent reputation of high moral character, this additional trust can remove scepticism about reasons to engage in CSR. Following Maigan & Ralston (2002) removed scepticism can positively influence the relation between CSR and firm performance. Royal firms differentiate from their competitors in the fact that their behaviour leads to the acquisition of royal status, and the royal family wants to connect their name to the brand. This is expected to have a positive effect on the CSR, firm performance relationship (Kiessling et al., 2015). As mentioned before, I expect royal firm status to result in a responsible mind-set which is expected to be embedded in their culture. Therefore, I expect Royal firm status to have a positive influence on the relationship between CSR and firm performance. This is translated into the following hypothesis.

(14)

3. RESEARCH MEHTODOLOGY

3.1 Sample Selection

This study investigates the effects of firm legitimacy, in the form of royal Dutch firm status, on its CSR engagement and financial performance. The sample comprises of all Dutch companies in three CSR databases. The period of the study is restricted to the availability of ESG data of Dutch companies. The data collected range from 2001-2015 and have been extracted from five different sources. The ESG data are gathered from Bloomberg, Thomson Reuters ASSET4 or the Dutch Transparency benchmark. The Thomson Reuters ASSET4 database is accessed through Compustat Global. Data of the Dutch Transparency benchmark is published on their website1. Data about the Royal status of the firms in this sample is gathered from the list Royal predicate (koninklijkhuis.nl). Financial data is retrieved from Bloomberg and bureau van Dijk ORBIS. Together, this forms a unique sample that captures the level of CSR engagement among public as well as non-public Dutch firms. Following prior literature financial and utility companies matching the SIC descriptions of SIC codes 4000-4999; 6000-6599 and 6700-6799 are excluded from some analyses of the data, for testing the robustness of the findings (Hoi, Wu & Zhang, 2013). Appendix B provides a detailed overview of the SIC descriptions.

3.2 Variable Measurement

The following paragraphs will elaborate on the variables used for testing the three hypotheses. See Appendix A for an overview of the definitions of the variables used. Since every measure or proxy of a variable has its limitations, multiple measures of each dependent variable are used for testing the hypotheses, in order to test the generalizability over various operationalisations.

Measures of CSR

Since the availability of CSR data of Dutch companies is limited, following Halbritter & Dorfleitner (2015) I combine different measures of Environmental, Social and Governance data (ESG), namely data of ASSET4 and Bloomberg as a proxy for CSR. Joseph et al. (2016) use GRI compliance as a measure of CSR performance, as the Dutch transparency benchmark

1_{2016 scores represent the scores that belong to financial year 2015 and are retrieved from the following link:}

https://www.transparantiebenchmark.nl/scores#/survey/3 Scores from years prior to 2015 are retrieved from the

(15)

is the most widely used benchmark of Dutch firms, including GRI compliance, I use transparency scores in my CSR measure.

Bloomberg ESG is a proprietary Bloomberg score which currently contains information of most publicly traded companies worldwide between 2005-2015. This ESG measure is widely acknowledged in research, being consulted among more than 12.000 customers. The data collected by Bloomberg contains 219 metrics covering a wide range of information about sustainability and CSR issues. The ESG matrix in Bloomberg is based upon the underlying principles of the widely used GRI framework for triple bottom line firm performance.

ASSET4 is part of Thomson Reuters and contains data between 2001-2015. This comprehensive ESG database contains information on more than 400 metrics, including all known criteria and aspects of sustainability performance. Apart from KLD, Bloomberg and ASSET4 are the largest ESG rating agencies of public companies worldwide (Halbritter & Dorfleitner, 2015).

The Dutch Transparency Benchmark is an initiative of the Ministry of Economic Affairs intended to provide insight into reporting manners of large Dutch companies regarding their CSR activities (Transparency Benchmark, 2016). The benchmark is performed every year and results are available for the period between 2003-2015. Participation is obligatory according to the standards of the participations protocol2. The ultimate goal is to encourage organizations to be transparent about their CSR performance, helping stakeholder dialogues by providing more specific information on CSR performance (Transparency Benchmark, 2016). Among the assessment criteria are the “guidelines of GRI (G4: 4th generation), IIRC and OECD-guidelines for multinational enterprises and the EU legislative proposal and focus on materiality, value creation and impact” (Transparency Benchmark, 2016).

None of the Bloomberg ESG data are based on estimations and all data are retrieved from annual reports, CSR reports, and surveys taken by Bloomberg. Research shows that there are no reliability differences between Bloomberg and ASSET4 and Bloomberg has the highest data availability rate in my sample (Heuvel, 2012; Wheelan, 2009). Taken together, Bloomberg ESG will be used as the primary measure of CSR. I prefer Bloomberg and ASSET4 scores

2_{Participation is mandatory for PIEs with 500 employees or more, for all companies listed on the Amsterdam}

Stock exchange, for companies with operations within the Netherlands with a substantial turnover and number of employees, all companies in which the government participates in the risk capital for at least 10%, Universities and University Medical Centres and large companies (more than 250 employees) that voluntarily joined the research group. The participation protocol can be retrieved from:

https://www.transparantiebenchmark.nl/sites/transparantiebenchmark.nl/files/afbeeldingen/participant_protocol_ tb2016_0.pdf

(16)

above the Transparency benchmark, based the fact that they are the largest ESG rating agencies

worldwide. CSR_SCOREit is determined in the following way. When the CSR scores of all 3

databases are available for a firm (216 firm-year observations), the Bloomberg score is used. When the Bloomberg score is missing (63 firm-year observations), the ASSET4 score is used, when the ASSET4 score and the Bloomberg score are missing (1320 firm-year observations), the Dutch Transparency benchmark score is used.

The three benchmarks have different ranking methods, following Hoi, Wu & Zhang (2013) as well as Crifo et al. (2016), the CSR measures are transformed into ordinal scores ranging between 1-4 for CSR_SCOREit and 0-1 for CSR_DUMMYit.. CSR_SCOREit equals 0

when the score is below the first quartile of the relevant CSR measure, it equals 1 when it is between the first quartile and the median, it equals 3 when it is between the median and the third quartile and it equals 4 when the score is above the third quartile of the relevant CSR score. CSR_DUMMYit equals 0 when the CSR score is below median and equals 1 when the

CSR score is at, or above median. The criteria for the Transparency Benchmark have been revised several times during my sample period, therefore its ordinal score is based on the per year median and quartile scores. CSR_SCOREit and CSR_DUMMYit are calculated for firm i

and year t.

Measures of tax avoidance

The literature on tax avoidance mainly uses three measures, book-tax differences, GAAP effective tax rate (ETR) and Cash effective tax rate (CETR) (Dyreng et al., 2008; Graham et al., 2014; Hoi, Wu & Zhang, 2013; Kim et al., 2011). Book-tax differences are the differences between incomes reported to the capital market and the incomes reported to the tax authorities by U.S. public firms (Kim et al., 2011). This measure cannot be used for my sample of Dutch firms, since their book-tax differences are not publicly available. Firms that comply to GAAP, have to disclose the ETR annually. This results in the fact that the ETR is a widely available measure, since GAAP is widely used in the Netherlands. Following Dyreng et al. (2008) ETRit for firm i in year t is defined as:

𝐸𝑇𝑅_𝑖𝑡 =𝐼𝑛𝑐𝑜𝑚𝑒 𝑡𝑎𝑥 𝑒𝑥𝑝𝑒𝑛𝑠𝑒𝑖𝑡 𝑃𝑟𝑒𝑡𝑎𝑥 𝑖𝑛𝑐𝑜𝑚𝑒_𝑖𝑡

Prior literature names the following limitations to ETRit as measure of tax avoidance. It

(17)

income is negative, the ETRit cannot be calculated because of the negative denominator. As the

ETRit includes deferred taxes, tax avoidance using deferred taxes cannot be detected with the

ETRit.Tax avoidance by the use of accruals or employee stock options can also go undetected

in the case of ETRit..

Dyreng et al. (2008) show that the one-year tax rates are not accurate measures for long-run tax avoidance, and longer term measures should be used to accurately measure a cash effective tax rate. The CETRit will avoid the use of negative denominators and the long term

CETRi avoids significant impact of year-to-year variation in the tax rates and the tax benefits

of employee stock options (Dyreng et al., 2008). The CETRi for firm i and period t=1 to

measurement period N is defined as:

𝐶𝐸𝑇𝑅𝑖 =

∑𝑁 𝐶𝑎𝑠ℎ 𝑇𝑎𝑥 𝑃𝑎𝑖𝑑_𝑖𝑡

𝑡=1

∑𝑁𝑡=1(Pretax Income𝑖𝑡− 𝑆𝑝𝑒𝑐𝑖𝑎𝑙 𝐼𝑡𝑒𝑚𝑠𝑖𝑡 )

The measurement period N will vary by 1, 3 and 5 years for ETRi as well as for CETRi.

Following Graham et al. (2014), Hoi, Wu & Zhang (2013) and Kim et al. (2011) all ETR scores are transformed into ordinal measures ranging between 1-4 according to the quartile scores of the relevant measure. ETRi and CETRi are reversed measures of tax avoidance, therefore an

ETR score of 1 indicates a high level of tax avoidance, where 4 indicates a low level of tax avoidance. Considering the limitations of different tax avoidance measures, ETRit, CETRi,

ETR3i, CETR3i,ETR5i and CETR5i are all used in this thesis to test the generalizability over

various operationalisations (Dyreng et al., 2008; Graham et al., 2014; Hoi, Wu & Zhang, 2013).

Measures of Firm performance

ROA is used as a measure of firm performance, since it is an authentic measure of firm performance that is not effected by leverage (Mishra & Suar, 2010). ROA is a measure that addresses earnings generated from invested capital (assets) independent from firm size. It represents firms’ profitability in respect to the total set of resources, which consists of all assets in its control. ROA has been widely used by CSR researchers to measure the impact of CSR on firm performance (Cheng et al., 2015; Kiessling, 2014). Following Cheng et al. (2015) I define ROAit for firm i and year t as follows:

𝑅𝑂𝐴𝑖𝑡 = 𝐿𝑜𝑔 (

𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠_𝑖𝑡 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠𝑖𝑡)

(18)

ROEit is another widely used measure of firm performance in CSR literature (Cheng et

al., 2015; Kiessling, 2014). It is a profitability indicator which measures the degree in which a company deploys its capital. This is calculated for firm i and year t as follows:

𝑅𝑂𝐸_𝑖𝑡 = 𝐿𝑜𝑔 ( 𝑂𝑝𝑒𝑟𝑎𝑡𝑖𝑛𝑔 𝐼𝑛𝑐𝑜𝑚𝑒𝑖𝑡

(𝑇𝑜𝑡𝑎𝑙 𝐶𝑜𝑚𝑚𝑜𝑛 𝐸𝑞𝑢𝑖𝑡𝑦𝑡+ 𝑇𝑜𝑡𝑎𝑙 𝐶𝑜𝑚𝑚𝑜𝑛 𝐸𝑞𝑢𝑖𝑡𝑦𝑡−1

2 )

)

CSR*ROYALit represents the interaction term used in hypothesis 3B, to test the

attenuation of Royal firm status on the association between CSR and firm performance. The interaction term CSR*ROYALit is given by CSR_SCOREit multiplied by ROYALi..

3.3 Regression Models

This study investigates the effects of firm legitimacy, in the form of royal status, on its CSR engagement and financial performance for a given set of firms over a time span of 15 years. Given the fact that the sample consists of panel data, regression models will be used that are suitable for data of entities across time. The fixed or random effects model are preferred to analyse panel data, as they control for entity specific factors or factors that change over time but apply to all entities (Torres-Reyna, 2007). Time invariant variables cannot be included in fixed effects analysis, since ROYALi is a time invariant variable in this sample, application of

the Hausman test shows that the random effects model is suitable for the analyses of my hypotheses (Hausman, 1978). In addition, the The Breusch and Pagan Lagrangian multiplier test for random effects indicated that the Random effects model is preferred above OLS regression for hypotheses 1 and 3. The data of hypothesis 2 did not meet the requirements to calculate a F-statistic using the random effects model. Therefore, in line with prior tax avoidance literature, OLS regression is used (Hoi, Wu & Zhang, 2013). The variance inflation factor (VIF) will be calculated for each regression model to test for multicollinearity.

Based on the theory depicted above, a positive association is expected between ROYALi

and CSR_SCOREit and CSR_DUMMYit. In the equations in this study, i represents firms and

t represents time. The vertical bar (|) separates the different dependent variables for which the regression analyses are performed. The regression model for the first hypothesis is as follows:

(19)

Based on the above depicted theory, I expect a positive association between ROYALi

and C|ETR measures. The regression model for the second hypothesis is as follows:

ETRit| CETRi| ETR3i| CETR3i|ETR5i||CETR5i= β0 + β1 ROYALi + β2 SIZEit + β3 ΔSALEit +

β4PERFit + β5 DE_RATIOit

Based on the theory depicted above, a positive association is expected between CSR_SCOREit and ROAit|ROEit. The regression model for hypothesis 3A is as follows:

ROAit|ROEit = β0 + β1 CSR_SCOREit + β2 SIZEit + β3 ΔSALEit + β4 DE_RATIOit +β4 PPEit

The regression model for hypothesis 3B is as follows:

ROAit|ROEit = β0 + β1 CSR_SCOREit + β2 SIZEit + β3 ΔSALEit + β4 DE_RATIOit +β5 CSR*ROYALit

The definitions and abbreviations of the dependent and independent variables are depicted above in the variable measurement chapter. Appendix A displays an overview of all variables. Following prior literature,next to the dependent and independent variables, a number of control variables are used in the regression model to control for firm specific characteristics (Halbritter & Dorfleitner, 2015; Hoi, Wu & Zhang, 2013; Kiessling, 2014; Lanis & Richardson, 2012; Zeng, 2016).

Following prior literature, SIZEit is used as a control variable in, to control for firm

specific characteristics, it equals the natural logarithm of the total assets of firm i in year t. Others argue that size is related to higher CSR scores, more opportunities for tax avoidance practices and higher firm performance, therefore SIZEit is used as a control variable for all

hypotheses (Crifo et al., 2016; Dyreng et al., 2008; Kim et al., 2015; Kiessling et al., 2015).

ΔSALEit is used as a variable to control for sales growth, it is given by the change in

sales for firm i and year t. Prior literature identifies ΔSALEit as an important firm characteristic

which can also be seen as a measure of reputational aspects (Graham et al., 2014; Hoi, Wu & Zhang, 2013). Reputational aspects are important to royal companies, CSR and to firm status, ΔSALEit is expected to have a positive association with CSR and firm performance, but it is

(20)

corresponding legitimacy concerns. ΔSALEit is included as a control variable in all regression

models (Graham et al., 2014).

Following prior literature, DE_RATIOit is used to control for leverage as firm

characteristic. DE_RATIOit equals total debt divided by total equity for firm i and year t. Others

argue that DE_RATIOit influences firm performance, tax avoidance and CSR engagement

(Cheng et al., 2015; Dyreng et al., 2008; Graham et al., 2014; Hoi, Wu & Zhang, 2013; Kim et al., 2015; Lanis & Richardson, 2012). It is expected to be negatively associated with royal status and firm performance since financially stable firms tend to prefer equity over debt when they finance their activities. Consequently, DE_RATIOit is used as a control variables in all

hypotheses.

In hypothesis 2, I control for the effect of PERFit which is given the return on assets

being calculated as the net income divided by the average total assets for firm i and year t. PERFit is a firm performance measure that is commonly used in tax avoidance research, as in

general the annual paid taxes should be driven by profitability, a positive association between PERFit andC|ETRit. is expected (Dyreng et al., 2008; Graham et al., 2014; Hoi, Wu & Zhang,

2013; Lanis & Richardson; Kim et al., 2015).

Following prior literature, for all hypothesis additional robustness tests will be performed in which finance and utility companies will be excluded from the sample (Cheng et al., 2015; Hoi, Wu & Zhang, 2013). The industries of the companies in my sample have been identified using Bureau van Dijk ORBIS, I exclude utility and finance companies with SIC codes 4000-4999 and 6000-6999, companies that fit the description of these codes but do not have a SIC code in ORBIS are also excluded. Appendix B provides a detailed overview of the SIC descriptions.

(21)

4. RESULTS

4.1 Descriptive Statistics of the Sample

Table 1 gives descriptive statistics for the measures that for the hypotheses described above. The dataset contains measurements of 280 companies among which 25 have royal status giving a total of 375 royal firm-year observations between 2001 and 2015. Together, the dataset contains 4200 firm-year observations for the variable ROYALi that indicates whether or not a

firm has royal status. The comprised CSR measure provides data of 2,133 firm-year observations, resulting from 233 companies, of which 21 are royal. The sample sizes vary tremendously, CETRit provides the lowest amount of observations with 1,247.This results in

lower numbers of observations for hypothesis 2 in comparison to hypothesis 1 and 3.

The descriptive statistics show that the mean CSR_SCOREit and CSR_DUMMYit score

as expected within their range of 1-4 and 0-1 respectively. The C|ETRit measures give larger

means for the measures over longer periods of time. This is consistent with prior literature as it is difficult to sustain low effective tax rates for longer periods of time (Dyreng et al., 2008). The small median for ROYALi is in line with my expectation as only 25 companies of my

sample have Royal status.

The descriptive statistics for the sample of non-financial and non-utility companies are enclosed in Appendix C, they do not show any unexpected values.

Table 2 gives an overview of the Pearson correlations of the variables used for the hypotheses. Small but positive and significant correlations are found between ROYALi and

CSR_SCOREit |CSR_DUMMYit (0.17 and 0.11). Which implies that Royal companies perform

better in terms of CSR than non-Royal companies. According to the expectation, CSR_ SCOREit and CSR_DUMMYit show a high mutual correlation (0.70).

The different measures of tax avoidance, ETRit, CETRi, ETR3i, CETR3i, ETR5i, and

CETR5i show high mutual correlations (0.68-0.86). The tax avoidance measures show a

significant positive correlation with ROYALi in accordance to the expectations of the second

hypothesis. The correlations range between 0.18- 0.15, this indicates that Royal companies have a higher C|ETRit and thus tend to avoid less taxes.

In contrast to my expectation based on the theory section, the correlations between CSR_ SCOREit and firm performance measures ROAit |ROEit are negative (-0.16 and -0.05).

This implies that companies with high CSR scores are less profitable. ROAit and ROEit which

(22)

All control variables show mutual correlations of 0.19 or below. The Variance of Inflation Factors(VIF) shows that multicollinearity is no concern, because all VIFs are smaller than 1.1. According to the expectation, SIZEit shows positive and significant correlations with

CSR_SCOREit |CSR_DUMMYit and a negative significant correlation with ETRit |CETRi

|ETR3i |CETR3i |ETR5i |CETR5i.. This finding indicates that larger firms might have more

opportunities to avoid taxes. The correlation between CSR_SCOREit |CSR_DUMMYit and

ROAit |ROEit is significant negative, which is contrary to my expectation. ΔSALEit does not

significantly correlate with any other variable. DE_RATIOit does not show any correlations that

confirm my expectations, it was expected to show a negative correlation with the measures of CSR, C|ETRit and firm performance, but it shows a positive correlation with CETRit, ETRit,

ETR3it and ETR5it. The remaining correlations are non-significant. CETRit, ETRit, CETR3it

and CETR5it show positive significant correlations with PERFit which confirms my

expectations.

The Pearson correlations of the sample of non-financial and non-utility companies are enclosed in Appendix D. The following deviations from the above stated findings are observed. The correlation between ROYALi and CSR_SCOREit |CSR_DUMMYit is slightly higher (0.21

and 0.13), the correlation between ΔSALEit and ROAit |ROEit is according to the expectations

significant positive and contrary to the expectation, DE_RATIOit shows a positive significant

(23)

TABLE 1 Descriptive Statistics Count Mean Stand. Dev. P25 Median P75 Dependent Variables CSR_SCOREit 2,133 2.40 1.11 1 2 3 CSR_DUMMYit _2,133 _.55 _.50 ₀ ₁ ₁ ETRit _1,389 _1.63 _1.04 ₁ ₁ ₂ CETRit _1,247 _1.76 _1.09 ₁ ₁ ₃ ETR3it 1,488 1.73 1.08 1 1 2 CETR3it 1,254 1.80 1.11 1 1 3 ETR5it 1,462 1.75 1.09 1 1 2 CETR5it 1,275 1.83 1.12 1 1 3 ROAit 1,942 1.38 1.28 .81 1.65 2.17 ROEit 1,983 2.89 1.16 .23 2.98 3.54 Independent Variables ROYALi 4,200 .09 .29 0 0 0 Control Variables SIZEit 2,454 16.75 5.86 10.1 19.40 20.93 ΔSALEit _2,398 _3,167 _155,060 _-.06 _.01 _.12 DE_RATIOit 2,443 13,429 660,707 23.06 40.27 59.02 PERFit 2,362 4.75 9.57 .65 3.93 7.70

Variable definitions are included in Appendix A. Count = number of firm-year observations.

ROYALi is the starting point of this research therefore, the initial sample of firms for which data

(24)

- 23 - TABLE 2 Pearson Correlations CSR_ SCOREit CSR_

DUMMYit ROYALi SIZEit ΔSALEit

DE_

RATIOit CETRit ETRit ETR3it CETR3it CETR5it ETR5it PERFit ROAit ROEit

CSR_SCOREit 1.00 CSR_DUMMYit 0.70*** 1.00 ROYALi 0.17*** 0.11*** 1.00 SIZEit 0.10*** 0.18*** -0.20*** 1.00 ΔSALEit 0.01 0.02 0.05*** 0.01 1.00 DE_RATIOit 0.07*** 0.03 0.05*** -0.00 -0.00 1.00 CETRit 0.07** -0.04 0.17*** -0.80*** 0.01 0.28*** 1.00 ETRit 0.01 -0.08* 0.17*** -0.70*** -0.01 0.26*** 0.76*** 1.00 ETR3it _0.02 _-0.10*** _0.17*** _-0.69*** _-0.02 _0.20*** _0.75*** _0.90*** _1.00 CETR3it 0.12*** -0.01 0.18*** -0.78*** 0.02 -0.02 0.90*** 0.70*** 0.70*** 1.00 CETR5it _0.13 _-0.01 _0.17*** _-0.75*** _-0.02 _-0.02 _0.86*** _0.68*** _0.65*** _0.93*** _1.00 ETR5it 0.00*** -0.09*** 0.15*** -0.69*** -0.02 0.15*** 0.73*** 0.86*** 0.85*** 0.70*** 0.68*** 1.00 PERFit _-0.03 _-0.04* _-0.04* _-0.07*** _-0.01 _0.05** _0.08*** _0.14*** _0.07 _0.06** _0.05* _0.04 _1.00 ROAit -0.16*** -0.14*** 0.02 -0.17*** 0.03 0.03 0.06* 0.11*** 0.10*** 0.05 0.03 0.11*** 0.71*** 1.00 ROEit _-0.05* _-0.07** _-0.03 _-0.19*** _-0.04 _0.12*** _0.14*** _0.18*** _0.14*** _0.15*** _0.12*** _0.17*** _0.46*** _0.57*** _1.00

*, **, *** indicate 10 percent, 5 percent and 1 percent significance. Variable definitions are included in Appendix A. Hypothesis 1 measures the relation between ROYALi and CSR_SCOREit |

SIZEit ΔSALEit PERFit and DE_RATIOit. Hypothesis 3A measures the relation between CSR_SCOREit and ROAit|ROEit and controls for SIZEit ΔSALEit and DE_RATIOit. Hypothesis 3B

(25)

- 24 -

4.2 Regression Results

Results of the first hypothesis

Table 3 presents the results of the first hypothesis. To check for the generalizability of the findings with respect to the CSR measures, CSR_SCOREit and CSR_DUMMYit are both

tested in two separate regression analyses. The coefficient of ROYALi is positive and highly

significant in both models, for the whole sample as well as for the non-financial and non-utility companies. This confirms the first hypothesis and the robustness of the findings. SIZEit is the

only control variable that proofs to be a relevant for CSR.

Although the results are robust and highly significant for both CSR measures, the R2_of

the model is relatively low. However, the squared correlation shows that this model has higher explanatory power compared to a single independent variable correlation. The results show the within, between and overall R2. The within R2 refers to the variance of CSR_SCOREit

|CSR_DUMMYit within one firm over time that is explained by the model. The between R2

refers to the variance of CSR_SCOREit |CSR_DUMMYit between firms over time that is

explained by the model. The overall R2 refers to the overall variance of CSR_SCOREit

|CSR_DUMMYit over time that is explained by the model. The results indicate that the models

explains at least 6 times more variance between firms and at least 5 times more overall variance compared to the variance explained within firms. In this context I find that ROYALi, which is

a constant factor within firms, is a small but reliable factor in the prediction of CSR_SCOREit

and CSR_DUMMYit.

The results show that Royal firm status positively impacts the CSR score of a company, which confirms the first hypothesis.

(26)

Results of the second hypothesis

Table 4 shows the results of the testing of the second hypothesis. As every measure of tax avoidance has its own limitations, I measure tax avoidance using ETRit |CETRi, |ETR3i|

CETR3i |ETR5i |CETR5i to improve the generalizability of the findings. The coefficient

ROYALi is positive for all tax avoidance measures. The tax avoidance measures give a higher

ETR score when the level of tax avoidance is lower. These findings are in line with the second hypothesis and shows that Royal firm status is negatively associated with tax avoidance. The results are significant for ETRit, CETRi, ETR3i, CETR3i, and ETR5i. These results are robust

for the sample of non-financial and non-utility companies for ETRit, CETRi, ETR3i, and ETR5i.

Following the theory, the different levels of significance can be explained by the fact that every tax avoidance measure has its limitations in terms of their operationalisations.

SIZEit is a control variable which is negatively associated with ETR and is highly

significant for all ETR measures. This suggests economies of scale on tax avoidance (Dyreng et al., 2008). This means that larger sized firms are paying lower amounts of taxes. DE_RATIOit

TABLE 3

Hypothesis 1, random effects model analysis of CSR_SCORE and CSR_DUMMY

CSR_SCOREit CSR_DUMMYit

Panel Data Random Effects

Model Panel Data Random Effects Model

Independent Variables All Firms

Non-Financial &

Non-Utility _{All Firms}

Non-Financial & Non-Utility ROYALi .77*** .81*** _.27*** .29*** _(3.65) _(3.79) _(3.18) _(3.26) Control Variables SIZEit .04*** .03*** _.02*** .02*** (3.57) (2.66) _(4.21) (4.04) ΔSALEit .00 .00 _.00 .00 (0.36) (0.32) _(1.26) (1.24) DE_RATIOit .00 -.00 _.00 -.00 (0.46) (-0.44) _(0.46) (-0.45) Intercept 1.50 1.57*** _.16* .12 (7.00) (7.06) _(1.77) (1.28) R2_Within _0.01 _0.01 0.00 0.01 R2_Between _0.06 _0.08 _0.08 _0.10 R2_Overall _0.05 _0.06 _0.07 _0.08 Number of firms 229 188 ₂₂₉ 188 Number of Observations 1,517 1,253 1,517 1,253

*, **, *** indicate 10 percent, 5 percent and 1 percent significance. Definitions of financial and utility companies are included in Appendix B. Variable definitions are included in Appendix A.

The regression model of hypothesis 1 is given by:

(27)

is a positive and significant control variable for ETRit, CETRi and ETR3i. This indicates that

more leveraged firms are paying slightly higher amounts of taxes. PERFit is a negative and

significant control variable for CETR3i, ETR5i and CETR5i., this is consistent with Graham et

al. (2014).

The R2 indicates that more than half of the variance of ETRit |CETRi, |ETR3i| CETR3i

|ETR5i |CETR5i is explained by this model. The results show that Royal firm status positively

effects the ETR of a firm in my sample, this confirms the second hypothesis.

Results of the third hypothesis

Table 5 shows the findings of the tests associated with hypothesis 3A and 3B. The negative and significant coefficient CSR_SCOREit indicates a negative association between the

level of CSR and firm performance, which has a higher level of significance for ROAit than it

has for ROEit. The coefficient of CSR_SCOREit is for ROEit half the size of the coefficient for

ROAit. The size of the effect of CSR_SCOREit on ROEit is 5 to 10 times smaller than the effect

of ΔSALEit on ROEit. In this context it can be concluded that CSR_SCOREit is a small but

reliable factor in the prediction of the relationship with ROAit. The results are robust for the

non-financial and non-utility sample. They are in contrary direction compared to my expectation as formulated in hypothesis 3A.

The attenuating effect of the CSR*ROYAL coefficient shows to be insignificant for ROAit but is significant for ROEit. For ROEit the effect of CSR on firm performance is

insignificant. CSR*ROYAL has a negative effect on the negative relationship between CSR and firm performance, which can be interpreted as follows. For Royal firms the insignificant relationship between CSR and firms performance is more negative than for non-royal firms. The limited explanatory power of the attenuating effect can be explained by the fact that it only has explanatory power for royal firms. When a firm is not royal the value of CSR*ROYAL is 0. The limited amount of royal firms can provide an explanation for the lack of results in these tests. These findings are in contrast to hypothesis 3B

The results of hypothesis 3A and 3B are both consistent for non-financial and non-utility companies. SIZEit, ΔSALEit and DE_RATIOit are relevant control variables for firm

performance. SIZEit shows a small negative effect on firm performance, sales growth has a

positive effect and leverage increases ROAit but decreases ROEit.

The R2 of all tests are relatively low, which is expected in this context through the complexity of CSR measurement. The R2 indicates that the models explain more overall

(28)

variance of ROAit and between variance of ROAit compared to the variance of ROAit explained

within firms. These findings are not consistent with the findings of ROEit, the model explains

less variance between firms in these tests. However, the squared correlation shows that the models for ROAit. andROEit. Both have higher explanatory power compared to the single

independent variable correlations.

Overall, the findings in table 5 indicate a direct significant negative effect of CSR_SCOREit on ROAit. The size of this effects is the largest of all effects observed in this

table. There is no attenuation effect observed of CSR*ROYAL on ROAit. There is a weakly

significant attenuation effect observed of CSR*ROYAL on ROEit. The directions of the results

(29)

- 28 - TABLE 4

Hypothesis 2, Royal status and Tax Avoidance Regression Results

ETRit CETRit ETR3it

OLS Regression OLS Regression OLS Regression

All Firms

Non-Financial &

Non-Utility All Firms

Non-Financial &

Non-Financial & Non-Utility Independent Variables ROYALi .11** .12* .14*** .13** .18*** .22*** (1.98) (1.86) (2.63) (2.19) (3.21) (3.33) Control Variables SIZEit -.11*** -.12*** -.14*** -.14*** -.11*** -.12*** _(-32.16) _(-29.89) _(-43.33) _(-41.64) _(-32.63) _(-29.65) ΔSALEit -.00 .08* -.06 -.05 -.00 -.08 _(-0.22) _(1.75) _(-1.39) _(-1.13) _(-0.23) _(1.13) PERFit -.00 -.00 -.01*** -.01*** -.00 -.00 (5.76) (-1.21) (-3.94) (-3.60) (-1.02) (-1.04) DE_RATIOit .00*** .00*** .00*** .00*** .00** .00** (4.22) (3.69) (4.40) (3.20) (2.35) (1.97) Intercept 3.34*** 3.35*** 3.94*** 3.95*** 3.36*** 3.35*** (48.59) (45.37) (62.11) (60.03) (52.50) (48.11) Adj. R2 _0.50 _0.50 _0.65 _0.65 _0.48 _0.46 Number of Observations 1,301 1,124 1,191 1,074 1,373 1,182

(30)

TABLE 4

Hypothesis 2, Royal status and tax avoidance regression results (Continued)

CETR3it ETR5it CETR5it

OLS Regression OLS Regression OLS Regression

All Firms

Non-Financial &

Non-Financial & Non-Utility Independent Variables ROYALi .11 ** .072 .17 *** .15** .09 .01 (2.00) (1.24) (3.08) (2.30) (1.51) (0.25) Control Variables SIZEit -.14*** -.14*** -.12*** -.12*** -.13*** -.13*** (-43.64) (-42.41) (-33.55) (-30.34) (-38.32) (-36.87) ΔSALEit -.01 -.10** -.00 .037 .00 -.00 (-0.30) (-2.28) (-0.22) (0.72) (0.11) (-0.07) PERFit -.01*** -.01*** -.01** -.01** -.01*** -.01*** (-4.04) (-3.21) (-2.14) (-2.37) (-3.43) (-3.31)

DE_RATIOit -1.43e-08 -1.43e-08 -.00 -.00 -1.45e-08 -1.18e-08

(-0.69) (-0.69) (-0.38) (-0.56) (-0.65) (-0.53)

Intercept 4.03*** 4.02*** 3.45*** 3.48*** 3.84*** 3.84***

(66.00) (64.47) (53.89) (49.49) (61.14) (59.04)

Adj. R2 _0.62 _0.63 _0.48 _0.46 _0.56 _0.57

Number of Observations 1,192 1,080 1,377 1,172 1,202 1,072

*, **, *** indicate 10 percent, 5 percent and 1 percent significance levels. Definitions of financial and utility companies are included in Appendix B. Variable definitions are included in Appendix A. The regression model of hypothesis 2 is given by:

(31)

TABLE 5

Hypothesis 3, CSR and firm performance regression results

ROAit ROEit

Panel Data Random Effects Model Panel Data Random Effects Model

Independent Variables All Firms All Firms

Non-Financial & Non-Utility

Non-Financial &

Non-Utility _{All Firms} All Firms

Non-Financial & Non-Utility Non-Financial & Non-Utility CSR_SCOREit -.13*** -.12*** -.13*** -.11*** _-.06* -.04 -.06* -.03 (-3.55) (-2.96) (-3.57) (-2.69) _(-1.95) (-1.06) (-1.88) (-0.87) Control Variables SIZEit -.05*** -.05*** -.04*** -.04*** _-.04*** -.04*** -.03** -.03*** _(-3.67) _(-3.72) _(-2.85) _(-2.98) _(-3.50) _(-3.75) _(-2.46) _(-2.68) ΔSALEit .08* .080* .23*** .23*** .30*** .30*** .29*** .29*** (1.83) (4.54) (4.10) (4.12) (5.89) (5.91) (5.49) (5.51) DE_RATIOit -.00** -.00** -.00 -.00 .00*** .00*** .00*** .00*** (-2.00) (-2.00) (-1.37) (-1.43) (3.23) (3.23) (3.43) (3.38) CSR*ROYAL -.040 -.08 -.09* -.12** (-0.65) (-1.36) (-1.76) (-2.07) Intercept 2.37*** 2.38*** 2.31*** 2.33*** _3.51*** 3.54*** 3.45*** 3.47*** (9.65) (9.65) (9.43) (9.45) _(17.01) (17.09) (15.30) (15.46) R2 Within 0.01 0.01 0.03 0.04 _0.05 0.06 0.06 0.06 R2 Between 0.09 0.09 0.07 0.07 _0.05 0.06 0.03 0.04 R2 Overall 0.07 0.07 0.06 0.06 _0.07 0.07 0.05 0.06 Firms 213 213 174 174 ₂₁₆ 216 177 177 Number of Observations 1,250 1,250 1,034 1,034 1,288 1,288 1,068 1,068

*, **, *** indicate 10 percent, 5 percent and 1 percent significance levels. Definitions of financial and utility companies are included in Appendix B. Variable definitions are

included in Appendix A. The regression model of hypothesis 3A is given by: ROAit|ROEit = β0 + β1 CSR_SCOREit + β2 SIZEit + β3 ΔSALEit + β4 DE_RATIOit + β4 PPEit

Does Legitimacy Matter?