Opening the Black Box: The Effect on Institutional Distance of Corporate Social Irresponsibility

(1)

Opening the Black Box: The Effect on Institutional

Distance of Corporate Social Irresponsibility

By

Simone Payne

S2707004

s.e.payne@student.rug.nl

Thesis supervisor: Dr. C.H. Slager

Co-Assessor: Dr. J. Shin

University of Groningen

Faculty of Economics and Business

MSc International Business and Management

Nettelbosje 2

9747 AE Groningen

June 15

th

_{, 2020}

(2)

2

Abstract

This study endeavours to make fundamental contributions towards two different research fields, corporate social irresponsibility (CSiR) and institutional distances. This was done by studying the effect of the formal institutional distance, in terms of four World Governance Indicators, on MNEs’ CSiR performance score. A cross-sectional study was conducted using data from 2018 and a sample of 272 MNE operating in the extractive industry. Interestingly, the results for government effectiveness, rule of law and control of corruption turned out to be significant but negatively related to CSiR performance. This is in contrary from what was expected. Unfortunately, the results did not provide enough evidence to make inferences on the relationship between institutional distance in terms of regulative quality and CSiR performance. Nevertheless, this study makes important contribution by extending the literature on the antecedents of CSiR. More importantly, this study opens the black box to studying CSiR in the institutional distance context, since previous research has predominantly focused on researching social (ir)responsibility and the differences of institutional environments. In addition, by considering the between home and host country distance is asymmetric as an additional test, this study also contributes to this recent debate in institution-based literature which recognizes the asymmetry in distance.

Keywords: Corporate Social Irresponsibility, Corporate Social Responsibility, Institutional

(3)

3

Acknowledgements

(4)

4

3.2.1 Rule of law ... 22 3.2.2 Control of corruption ... 23 3.3 Conceptual model ... 25 4. Methodology ... 25 4.1 Research Philosophy ... 25 4.2 Data collection ... 26 4.3 Sample ... 26 4.4 Measurement of variables ... 27 4.4.1 Dependent variable ... 27 4.4.2 independent variables ... 28 4.4.3 control variables ... 29 4.5 Data analysis ... 32 4.6 Descriptive statistics ... 32 4.7 Correlation matrix ... 35 4.8 Robustness tests ... 35 5. Results ... 35

5.1 Results main model ... 36

5.2 Additional test ... 38

5.2.1 Logit regression ... 38

5.2.2 Directional Institutional Distance test ... 38

5.3 Robustness tests ... 39

6. Discussion ... 40

6.1 Implications ... 40

6.1.1 Theoretical and methodological implications ... 42

6.1.2 Implications for practice ... 44

(5)

5

7. Conclusion ... 46

8. References ... 48

Appendices ... 59

APPENDIX A: Debates and definitions in the literature of CSiR ... 59

APPENDIX B: Detail on the characteristics institutional environment qualities ... 62

APPENDIX C: Construction of ESG Controversy Category Score ... 63

APPENDIX D: Construction of the World Governance Indicators scores ... 65

APPENDIX E: Assumptions for parametric tests ... 70

APPENDIX F: Results additional tests ... 73

(6)

6

List of Figures

Figure 1: Conceptual model………25

List of Tables Table 1: Variables description, measurements and sources……….31

Table 2: Descriptive Statistics 2018……… 33

Table 3: Descriptive statistics industry dummy 2018………. 33

Table 4: Correlation matrix 2018……… 34

Table 5: Ordinary Least Squared regression analysis………. 37

List of Abbreviations

CC Control of Corruption

CSR Corporate Social Responsibility

CSiR Corporate Social Irresponsibility

EITI Extractive Industries Transparency Initiative

GDP Gross Domestic Product

GE Government Effectiveness

HDI Human Development Index

ID Institutional Distance

LOF Liability of Foreignness

MNE Multinational Enterprise

OLS Ordinary Least Squared

RL Rule of Law

ROA Return on Asset

RQ Regulative Quality

VIF Variance Inflation Factor

(7)

7

1. Introduction

“We must say now clearly that the age of irresponsibility must be ended” (Gordon Brown, UK Prime Minister to the UN General Assembly, 2008)

Early 2017 oil majors Shell and Eni were accused of bribery over a $1.3 billion payment, which was used to secure an exploration license for an oil block, back in 2011. However, there are valid claims that albeit the funds were paid to the Nigerian government, the capital actually went to Malabu Oil and Gas, a corporation linked to former oil minister and money-launderer Dan Etete. At first, both parties denied all allegations from the start, however later in 2017 Shell admitted being aware of the transactions to Malabu Oil and Gas. Correspondingly, campaigners indicate that this wrongdoing is one of the biggest corruption scandals in the history of the oil and gas sector (Gilblom, Browning and Albanese, 2019).

The case of Shell and Eni is a relatively well-known example of MNEs being involved in controversies over irresponsible business practices or corporate social irresponsibility (CSiR). Other notable cases include the Enron fraud scandal (Rockness and Rockness, 2005) and Foxconn violating employment rights in China (Condliffe, 2018). Resulting from the numerous scandals committed by multinationals enterprises (MNEs), corporate social responsibility (CSR) is one of the most stressed concerns in the world of academic literature (Matten and Moon, 2005; McWilliams and Siegel, 2001; Pava and Krausz, 1996). On the contrary, little advancement has been done so far on CSiR,

(8)

8

Sun, 2017; Lange and Washburn, 2012). Consequently, CSiR has a negative effect on a corporations’ financial performance, in contrast to CSR (Salaiz, 2016; Frooman, 1997). Given the severe consequences of CSiR, it should be viewed equally important as its counterpart. Thus, there is a theoretical as well as a managerial relevance for studying CSiR.

Since CSiR has not been studied extensively in previous literature, little is known about the antecedents of CSiR. Indeed, even less is known about the determinants of CSiR in the institutional context, since most researchers focused on studying the effect of internal factors on CSiR (Pearce and Manz, 2011; Tang, Qian, Chen and Shen, 2015). Nevertheless, external factors of MNEs should also be considered to fully understand corporate wrongdoings. On the contrary, scholars have recognized the importance of the institutional environment and their effect on corporate activities and their outcome, such as CSR engagement (Marano and Kostova, 2016; Campbell, 2007; Ioannou and Serafeim, 2012). In general, formal institutions have the power to adheres corporations to act responsible, by implementing CSR related regulation and policies (Knuden and Moon, 2017). However, the quality of the formal institutions is not uniform around the globe, which causes CSR variation across border (Amaeshi, Adegnite and Adegbite, 2016). Most studies have focused on explaining the differences in institutions and the effect on CSR (Mattan and Moon, 2008; Campbell, 2007). While only a few have considered institutional distances and their effect on CSR (Campbell, Eden and Miller, 2012; Keig, Brouthers and Marshall, 2019). Indeed, even less is known what effect institutional distances will have on CSiR.

The theory of institutional distances has finitely been applied to research on CSiR. In fact, to the best of my knowledge, this thesis would be the first one to measure the effect of formal institutional distance on MNEs’ irresponsible behavior. This study focuses on formal institutional distances, since formal institutions are obligatory to comply with for corporations and therefore cannot be taken for granted (Shirodkar and Konara, 2017). Accordingly, from here onward institutional distance indicates the formal institutional distance between the home and host country, unless mentioned otherwise. To fill this research gap on institutional distances and CSiR, this study has built a bridge between two research fields: theoretical elements of institutional distances, and literature on the (institutional) antecedents of CSiR. To research this mechanism, this study uses the distances of four World Government Indicator (WGI). Thus, the distances in Governance Effectiveness, Regulative Quality, Rule of Law and

(9)

9

effect on CSiR. This study uses data from 2018. The sample consists of 272 MNEs operating in the extractive industry. Hereupon the research question in this paper is as follows:

What is the effect of the institutional distances on MNEs’ corporate social irresponsible behavior?

In contradiction to what was expected, the results revealed a negative and significant relationship between institutional distances in terms of government effectiveness, rule of law, control of corruption and CSiR performance. Although this thesis found no support for the hypotheses, this study still contributes significantly to literature in several ways. First of all, this study contributes to research on the antecedents of CSiR. Gaining a deeper understanding on the determinants of CSiR, also contributes to the conceptualization and the associated debates on this concept. Secondly, to the best of my knowledge, no research has been conducted on the effect of institutional distances and CSiR, this study makes a significant contribution to academic literature in the field of CSiR and institutional distances. Although the results show contradicting results to what was expected, this research still establishes that institutional environments and institutional distances affects CSiR. Accordingly, further research is needed to establish the direction of the relationship of this mechanism. In addition, this research also takes into account novel insights on institutional distances theory, namely directional institutional distances. Moreover, this research contributes to the methods of measuring corporate irresponsible behaviour, since this is still challenging in the field of CSiR. This thesis takes a critical stand on measuring CSiR behaviour by media reports, since corporations engage in CSiR when this behaviour can remain hidden to the public (Armstrong and Green, 2013; Lin-Hi and Müller, 2013). This research also offers practical insights for managers, since the management is responsible for pulling the strings in a corporation. Managers should be more aware of CSiR, its antecedents and its severe consequences. Moreover, activists should not scrutinize MNEs’ for operating in institutional distant environments, since the results illustrate that this does increase the overall CSiR performance.

(10)

10

2. Literature Review

2.1 Corporate Social Responsibility

Throughout the past decades, CSR has become an extensively debated topic within the academic world. McWilliams and Siegel (2001) define CSR as “an action that appear to further some social good, beyond the interest of the firm and that which is required by law”. Their definition is one of the most frequently cited definitions of CSR. Hence, CSR goes beyond the pure financial gains (Ioannou and Serafeim, 2012), in contradiction to its sister term corporate social performance (CSP). CSP is the evaluation of corporations’ responsible and irresponsible activities (Liou and Lamb, 2018). In the rich stream of literature on CSR, the antecedents and the outcomes of CSR have been thoroughly investigated. This has been done at the institutional level and at the organisational level (Aguinis and Glavas, 2012). But especially the mechanism between CSR and financial performance has been studied extensively (i.e. McWilliams and Siegel, 2000; Margolis, Elfenbein and Walsh, 2007). These studies show contradicting outcomes, and theretofore this relationship has not been consistently established yet. Unfortunately, CSR is often associated with organizations that invest in CSR for the wrong reasons. For example, to compensate for negative outcomes (Gond and Moon, 2011). Popa and Salanță (2014) argue that this particular statement gives a hint towards using CSR in an irresponsible way, in which “good deeds” are used to cover up “bad deeds”. However, since CSR is subjective, a single act can appear to be responsible toward some and irresponsible towards others (Wood, 2010). Similarly, a corporation can have widely different records in both responsible and irresponsible acts. Indeed, Fry and Hock (1976) argue that a lot of corporations that score high on social disclosures are among the worst violators and polluters. Thus, “the current debate about the social responsibility of corporations has a serious blind spot: the research on the notion of CSiR.” (Greenwood, 2007: 325) Accordingly, there is a strong need for elaborating on CSiR.

2.2 Corporate Social Irresponsibility

(11)

11

receive much academic interest until four decades later, when society started to question the responsibilities of firms after the financial crisis of 2007-2008. There is no doubt that the negative consequences of CSiR, such as loss of legitimacy and damage of firm reputation, have played a major role in reintroducing this concept (Herzig and Moon, 2013; Armstrong and Green, 2013). In his first article on CSiR Armstrong (1977:185) describes the concept as “an immoral decision taken by the corporation’s executives, which will result into a gain by one party at the expense of the total system.” Thereafter, only a few academics have attempted to conceptualize CSiR. Evidently, Lin-Hi and Muller (2013) conducted an extended literature review where they found 22 academic articles covering CSiR. Yet only seven of these articles gave an explicit definition of CSiR. Seen the novelty of this concept, no consensus has been reached so far on a unanimous definition. Nevertheless, it appears all CSiR definitions share a common denominator, namely “CSiR is seen as immoral and/or illegal corporate actions with negative consequences for other stakeholders” (Lin-Hi and Müller, 2013: 1932). Correspondingly, this definition will be used through this study. Based on this definition CSiR is viewed as negatively affecting stakeholders, which may involve violating the law. However, CSiR can also occur without violating the law due to matters such as incomplete contracts (Hart and Hölmstrom, 1987) or lack of legal regulations on the global scale (Scherer and Palazzo, 2011).

Moreover, literature has revealed three different ongoing debates conceptualizing CSiR, all characterized by different positions and approaches (Riera and Iborra 2017). The two

most important debates1_{for this study are (1) the matter of deliberate vs. accidental CSiR and}

(2) the matter of self-contained vs. continuous conceptualizing of CSiR. This research takes the stand that CSiR can be done deliberately as well as accidentally. Accidental CSiR is viewed as “actions that disadvantage and/or bring harm to others are not inflected deliberately by a corporation.” (Lin-Hi and Müller, 2013: 1932). In other words, CSiR can result from unforeseen side effects of certain activities. For example, CSiR activities can occur along the supply chain. In this way, the corporation can be unaware of these deeds. CSiR is likely to occur when MNEs have a complex corporate value creation, due to a high level of internationalization (Strike, Gao and Bansal, 2006). To the contrary, “deliberate CSiR implies that corporations deliberately perform actions that disadvantage and/or harm others” (Lin-Hi

1_{Partial observer vs. impartial observer is the third ongoing debate on conceptualizing CSiR, however this}

(12)

12

and Muller, 2013: 1932). Most scholars agree on the notion that CSiR is used as a strategy (Child, 1972; Pearce and Manz, 2011; Lange and Washburn, 2012; Surroca, Tribó and Zahra, 2013; Keig, Brouthers and Marshall, 2019). Corporations engage in CSiR activities to reduce production costs or increase their profits. Activities such as child labour and illegal waste dumping can be an effective way of reducing costs. Notably, firms can only reap the benefits from these irresponsible activities when they remain hidden from stakeholders; when revealed, deliberate CSiR can become even more destructive to the corporation (Armstrong and Green, 2013; Lin-Hi and Müller, 2013).

The second debate refers to the relationship between CSR and CSiR. This research views CSiR as an independent mutually exclusive concept from CSR, meaning CSR and CSiR can occur coincidentally, as well as increase or decrease at the same time (Strike et al., 2006). Most scholars agree on this view (Prince and Sun, 2017; Muller and Kraussl, 2011; Shea and Hawn, 2018; Surroca et al., 2013). For example, Surroca et al. (2013) demonstrate MNEs transfer their irresponsible activities to weak institutional environments located in so-called pollution havens. This is done in response to mounting stakeholder pressures to act responsible. Hence, this strategy enables firm to be responsible in their home country, while simultaneously being irresponsible in the host country. Other scholars believe CSR and CSiR are dependent and not mutual exclusive, both as an extreme on the same continuum. This continuum is not viewed as static, instead corporations can move between the two extremes according to internal and external environmental factors (Jones, Bowd and Trench, 2009; Windsor, 2013; Perks et al., 2013). Appendix A gives a complete overview on the articles covering CSiR and the associated debates.

After discussing the literature and debates on CSR and more importantly on CSiR to comprehend the definitions and views on these phenomena more thoroughly, the subsequent step is to delve deeper into the drivers of acting (ir)responsible through an institutional lens.

2.3 National institutions and CSR

(13)

13

operating in well-organized and effective systems are more likely to behave in a socially responsible way, than corporations operating in less organized institutional environments with weak or even absent institutions (Campbell, 2007). However, the quality of institutional environments varies across borders, causing variation in macro-institutional pressures that determine CSR engagement (Aguilera and Jackson, 2003). In general, researchers distinguish between weak and strong institutional environments. The former is characterized by high regulatory uncertainty due to perpetual unexpected changes in governmental policies, high level of corruption, equivocal and complicated governance and insufficient legalistic enforcement and agents to coerce such regulation. The latter is characterized by the contrary aspects (Slangen and Van Tulder, 2009).

Literature on CSR in the institutional context is primarily conducted around explaining CSR variation between nations. This is done studying the macro-institutional pressures that determine the variation in CSR engagement (Mattan and Moon, 2008; Aguilera and Jackson,

2003).2_{In general, corporations will have to conform to macro-institutional pressures, to}

maintain legitimacy in to survive the competitive global market (Pfeffer and Salancik, 2003). Predominately, they will conform the regulative institutions (i.e. political and legal systems), since these institutions have been proven to have the most power to enforce CSR engagement (Bondy, Moon and Mattan, 2012; Ioannou and Serafeim, 2012). However, since institutions are to a great extend country specific, each institutional environment will have its unique level of stringency regarding CSR. Accordingly, this can explain CSR variation across-borders (Campbell, 2007). MNEs are challenged since they are embedded in multiple institutional environments. Thus, they have to conform to different, sometimes even contradicting, coercive pressures for CSR engagement (Marano and Kostova, 2016).

In a nutshell, CSR varies across borders because of distinct macro-institutional pressures. As MNEs are embedded in multiple institutional environments, they will have to simultaneously abide to the institutional orders of the home country as well as the institutional orders of all host countries. This increases the complexity for MNEs, especially when the institutional environments between home and host countries differ to a great extend from each other (Shirodkar and Konara, 2017). The subsequent sub-chapter discusses institutional distance in more depth, and the effect on CSR.

2_{The stream of literature within CSR in the institutional context is providing evidence for institutionalized CSR.}

(14)

14

2.4 Institutional distance and CSR

In general, institutional theory studies the embeddedness of organizations in the institutional environments, where institutions are seen as “the rules of the game” (North, 1990: 3). Formal institutions “determine the rules that govern economic activity and thus reduce uncertainty, risks and transaction costs” (Kostova et al., 2019: 4). Transaction costs are costs that arise from monitoring, negotiating and enforcement throughout the process of establishing an exchange between two parties (Jones and Hill, 1988; Ghoul, Guedhami and Kim, 2017).

However, each country has its own unique combination of institutional elements

(Kogut, 1991). This increases the complexity for MNEs that are embedded in multiple institutional environments. Thus, they will also need to conform to multiple institutional environments (Meyer, Mudambi and Narula, 1991). The differences in the formal institutions

between a given set of two countries is called formal institutional distance3. This refers to the

differences in laws and rules that influence business strategies. That is, the greater the formal institutional distance, the more differences there are between the formal institutional environments of the home and host country (Kostova et al., 2019). In this study institutional distances refers to formal institutional distance only.

Principally literature on institutional distances stresses that at greater institutional distance, MNEs’ will suffer from a higher level of lability of foreignness (LOF). Liability of foreignness describes the additional costs that foreign corporations experience that domestic corporations do not encounter (Zaheer, 1995). Moreover, greater institutional distance between home and host country is more challenging for MNEs since they will have to adapt to the unfamiliar local market. Besides costs, greater institutional distance creates increased risk and uncertainty for MNEs. Since they are not familiar with how business should be done in order to establish and maintain a legitimate business (Kostova and Zaheer, 1999; Xu and Shenkar, 2002; Eden and Miller, 2004). In addition, there is also a risk of internal tension among organizational units that are located in different countries, for they will try to work with the external institutional arrangements in their respective country (Kostova and Roth, 2002). In fact, Strike et al. (2006) demonstrate that an increase of internal diversification can also cause tension among organizational units, since all affiliates are embedded in distinct institutional environment. Accordingly, increased complexity can lead to CSiR (Strike et al., 2006). 3_{Informal institutional distance refers to the difference in rules embedded in values, norms and beliefs (Estrin,}

(15)

15

In sum, greater institutional distance leads to significantly higher costs and risks for the MNE, due to inadequate understanding of the institutional environment, inability to simultaneously adapt to the requirements set by the institutional orders, difficulties in establishing external legitimacy from stakeholders and increased internal as well as external complexity (Kostova et al., 2019). To reduce the aforementioned risks and uncertainties managers could be more causation with their business strategies in greater institutional distant environments (Olcott, 2009). In addition, MNEs could use CSR as a method to reduce the risks of greater institutional distance, by making contributions to the host country in order to enhance their reputation (Liu, Marshall and McColgan, 2017). Yang and Rivers (2009) advocate that in a case of greater institutional distance, MNEs will foster local CSR, in which a corporation adapts to the needs and standards of the host country. That is, the greater the institutional distance, the less does global CSR hold. Husted and Allen (2006: 840) describe local CSR as dealing “with the firm’s obligations based on the standards of the local community”, such as fighting against unemployment and HIV. Moreover, they identify global CSR as dealing “with the firm’s obligations on those standards to which all societies can be held” such as environmental protection and human rights (Husted and Allen, 2006: 840).

Nonetheless, the greater institutional distance between home and host country can also affect the willingness of MNEs to engage in CSR in the foreign market. Campbell et al. (2012) found that corporations operating in greater institutional distant environments are less likely to engage in CSR activities, that are activities “above and beyond” the government-mandated level of CSR. This is because these corporations already face enough challenges with complying to the host country’s unfamiliar regulation (Campbell et al., 2012). For a corporation to incorporate ‘local CSR’ they will have to acquire novel skills to conform to the higher CSR standards of the host market. Consequently, this will require additional resources and thus effect the willingness of MNEs in engage in CSR (Keig, Brouthers and Marshall, 2019; Campbell et al., 2012).

(16)

16

greater institutional distances are less likely to invest in local CSR, which will result into a lower CSP (Keig et al., 2019). However, not confirming to stakeholders’ pressures in the host countries, can lead to a lower level of competitive advantage at the subsidiary level for MNEs (Shirodkar and Konara, 2017).

So far literature on CSR in the institutional context has been discussed. Based on this literature, it appears MNEs face several challenges when operating in greater institutional distant environments, such as increased uncertainty and risk. Some authors argue MNEs will use CSR as a strategy to reduce these challenges (Liu, Marshall and McColgen, 2017). In contradiction other scholars argue MNEs will be less willing to engage in CSR because of the distance and its challenges (Campbell et al., 2012). In order to explore what effect institutional distance has on CSiR, the following sub-chapter will first review literature on national institutions and CSiR. Based on this literature the gap in existent literature with regards to institutional distances and CSiR will be critically assessed.

2.5 National institutions and CSiR

As previously mentioned, little is known about the antecedents of CSiR. Indeed, even less is known about the institutional level antecedents of corporate misbehavior, as the majority of literature on explaining drivers of CSiR has relied on the agency theory and on other internal factors (Amaeshi et al., 2016). Nevertheless, what is known about CSiR is that firms will engage in irresponsible activities when they think they will get away with such behavior (Campbell, 2007) or when they will gain more from the benefits than from the costs (Toner, 1999).

As mentioned above, CSiR has merely been linked to internal factors, such as underinvestment of corporate R&D investment (Wu, 2014), weak internal CSR policy (Campbell, 2007; Hil, 2011; Wu, 2014) and CEO characteristics (Tang, Qain, Chen and Shen, 2015). In fact, Child (1972) even argues that CSiR is strictly a matter of corporate strategic choice from the upper echelons and cannot be blamed on the institutional environment. Yet, there is some evidence that shows CSiR is more likely to arise in some institutional environments than others.

(17)

17

are more based on profits in LMEs, hence executives are more tempted to increase profits for their own benefits at the expense of other stakeholders (Hall and Soskice, 2001). Evidently, since LMEs offer more CSiR incentives, more CSiR engagement was found in LMEs (Walker et al., 2019).

Moreover, CSiR behavior was found to be more prevalent in countries with a lower level of freedom in speech and press. The predominant explanation for this is that CSiR behavior is more openly discussed and scrutinized in countries with high level of freedom of speech than in countries with a low level of freedom of speech and press. In contradiction, CSiR behavior is easier hidden for the public b in countries with a low level of freedom in press and speech. Since there is a lower level of availability of means to communicate CSiR behavior (Fiaschi, Giuliani and Nieri, 2017). Accordingly, countries with a lower level of freedom in press and speech have more incentives for CSiR, as MNEs could easier get away with CSiR. Another posited institution-based antecedent of CSiR is market competition. Campbell (2007) argues that the imperative of acting irresponsible will be higher in countries with an extreme level of competition. This pressures firms to cut corners and save money, to survive the intense competitive environment (Schleifer, 2004). Notably, the financial crisis of 2008 shows numerous examples of corporations showing social irresponsible actions to endure intense competition (Ioannou and Serafeim, 2012). In a similar vein, Keig et al. (2015) link formal corruption, market competitiveness and CSiR together. They explain that an MNE is more likely engage in CSiR to maintain its industry competitiveness when their market competitors do the same. On the other extreme, in institutional environments where competition is nought, corporations are empowered because of their monopolistic position. This can lead to severe irresponsible activities such as price gauging (Campbell, 2007). In the abovementioned external pressures, the decision to engage in CSiR is still a deliberate decision from the upper echelons, as proposed by Child (1972). Nevertheless, there is also evidence for institutionalized CSiR.

(18)

18

the local regulations and do not have local connections. Thus, they will most likely use corrupt connections and pay bribes to the host government to establish and maintain their business operations (Campos and Giovannoni, 2007). Moreover, corrupt environments also have more incentives for companies engage in CSiR, as in some environments irresponsible behavior of corporations is rarely criticized or punished. In fact, bad deeds can be bought off with gifts (Wu, 2014).

(19)

19

(Surroca et al., 2013; Mani and Wheeler, 1997; Dam and Scholtens, 2007). Although corporations do not violate the law by using “pollution havens” locations for their dirty business, this strategy still harms stakeholders and therefore is seen CSiR for the entire MNE, as irresponsible acts at the subsidiary level will reflect on the MNE’s irresponsibility (Strike et al., 2006)

From prior literature, as discussed in the previous paragraph, it appears that aspects of CSiR have been studied separately in the institutional context. Only limited studied have studied the concept as a whole using an institution-based view. Thus, there is a need for more research on CSiR as a whole. Nevertheless, there is enough evidence from previous studies (i.e. Walker et al., 2019; Wu, 2014; Fiaschi et al, 2017) to argue for a connection between the institutional environment and CSiR behavior. However, this line in theory is still in its elementary stage. This study takes this line in literature on CSiR in the institutional context a step further, by (1) acknowledging MNEs are embedded in multiple institutional environments and (2) recognizing the difference in institutions between home and host country. From previous literature it appears that these institutional distances create challenges and opportunities for the MNE, which affect CSR (Marano and Kostova, 2016). Building on this, expectations are made on the relationship between the institutional distances of an MNEs host and home country and its CSiR behavior. Thus, the institutional theory and its sub-theory on institutional distances will be central in to develop hypotheses in the subsequent chapter.

3. Theoretical development

3.1 Setting rules and policies

(20)

20

and policies, which can result into a higher level of liability of foreignness and its associated costs and challenges (Zaheer, 1995).

The capacity of the government to effectively formulate and implement regulation and policy to ensure correct behavior is measured by Government Effectiveness and Regulative

Quality (Kaufmann et al., 2008). Please refer to appendix B for a detailed overview of the

characteristics of high and low qualities of Governmental Effectiveness and Regulative Quality.

3.1.1 Government Effectiveness

Government effectiveness indicates “the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government’s commitment to such policies.” (Kaufmann et al., 2008: 223). Examples of public services are competition and consumer protection, the environmental protection and the justice system. These public services protect the stakeholders by promoting responsible behavior and punishing irresponsible behavior. Additionally, civil services formulate and develop regulations and policies to ensure responsible behavior. Furthermore, these services should be independent from political pressures, otherwise corporations could try to seek control or bend the regulations in their favor by capturing regulators (Bernstein, 1955; Vogel, 1989).

(21)

21

In addition, adapting to multiple institutional environments, results into an increase in internal complexity. This can lead an increase in likelihood harming stakeholders (Strike et al., 2006). Thus, operating in countries with greater institutional distance in terms of government effectiveness, increases the likelihood for MNEs to misinterpret policies. Moreover, MNEs are challenged with adapting to standards of these services and quality of policies. As a consequence, MNEs could harm their stakeholders (Orr and Scott, 2008). Moreover, operating in more distant institutional environments in terms of government effectiveness results into higher costs and more internal complexity, which can result into CSiR. Accordingly,

H1: Institutional distance in terms of government effectiveness has a positive effect on the CSiR performance of an MNE.

3.1.2. Regulative quality

The regulative quality captures “the ability of the government to formulate and implement sound policies and regulations that permit and promote sector development” (Kaufmann et al., 2008: 223). These policies and regulations aim to reduce the level of poverty and promote economic growth in a country. This is best achieved through private enterprises supported by policies and regulations made by the government to ensure well-functioning and competitive markets (Schulpen and Gibbon, 2002). Corporations operating in markets where governments promote extreme levels of competition, can be pressured by the host government to act irresponsibly by cutting costs to survive the competitive environment. In a similar vein, monopolistic markets also incentivize corporations to engage in opportunistic behavior such as price gauging. As a result of the empowered monopolistic position of corporations (Campbell, 2007). Thus, there are market conditions that pressure foreign and domestic corporations to engage in CSiR. Indeed, these pressures could be higher for foreign affiliates, since foreign affiliates face additional costs due to the liability of foreignness (Zaheer, 1995). Sunk costs make it difficult for MNEs to pull out of these challenging markets. Therefore, MNEs must adapt to these markets despite the additional costs (Vachani, 1995). Thus, to stay competitive relative to the domestic competitors that do not face liability of foreignness, MNEs could be even more pressured to act socially irresponsible.

(22)

22

host countries’ institutional orders. This can lead to violation of local laws and policies (Mezias, 2002). In fact, MNEs facing challenges in understanding the complex and unfamiliar market conditions due to the high distance, could mimic domestic competitors as an adaptation strategy (Pinto, Leana and Pil, 2008). However as mentioned, certain market conditions could also pressure domestic corporations to engage in CSiR. This can lead to MNEs mimicking strategies that could conflict with the MNEs home country regulations on matters such as work conditions (Pinto et al., 2008). Consequently, irresponsible acts at the subsidiary level will reflect on the MNE’s CSiR (Strike et al., 2006). Accordingly,

H2: Institutional distance in terms of regulative quality has a positive effect on the CSiR performance of an MNE.

3.2 Respecting rules and policies

Simply setting laws and administrative rules to control corporate irresponsible behavior is not sufficient to reduce CSiR (Sarre, Fiedler and Doig, 2002). Regulation and laws are worthless, if they are not respected by the state, the corporations and its citizens. Thus, regulations and laws should be enforced by the theretofore appointed agents. Moreover, the agents employed by the institutions should be accountable for the society at large, and not influenced by the state or corporations (Bernstein, 1955). The respect of citizens and the state for the institutions that govern social and economic interactions is measured by Rule of Law and Control of Corruption (Kaufmann et al., 2008). Please refer to appendix B for a detailed overview of the characteristics of high and low qualities of Rule of Law and Control of Corruption.

3.2.1 Rule of law

(23)

23

society, if corporations lack confidence in the nation’s enforcement system and thus believe they can get away with harming their stakeholders. In the light of greater institutional distances, the confidence that agents have in the institutions and abide by the rules of society differs significantly. Accordingly, MNEs operating in institutional distant markets in terms of rule of law, MNEs face challenges with adapting to the high institutional rule of law. Since they are not familiar to the level of stringency of the rule of law, which leads to higher adaptation costs (Andreopoulos, Barberet and Nalla, 2018). Moreover, MNEs might try to work with their external institutional arrangement in the respective host country, which could conflict with the way of doing business in the respective host country (Kostova et al., 2019). Thus, this could reflect into internal complexity of an MNE, which could lead to conflicts of home and host countries’ institutional pressures (Strike et al., 2006). Moreover, this might lead to underestimating the quality of contract enforcement and property rights, and therefore not comply with the laws and regulations of the host country. Furthermore, if corporations are not familiar with the way the host countries’ enforcement act agents such as the police, courts and with the level of crime and violence, the likelihood of making mistakes during legal procedures increases. This can result into further lawsuits and unforeseen costs (Mezias,2002). Accordingly,

H3: Institutional distance in terms of rule of law has a positive effect on the CSiR performance of an MNE.

3.2.2 Control of corruption

(24)

24

paying bribes (Keig et al., 2015). In fact, in some institutional environments’ corruption is sometimes even institutionalized (Collins et al., 2008; Mombeuil, Fotiadis and Wang, 2019). Correspondingly, corporations are compelled to conform to these pressures in order to maintain a legitimate business (Imerman, 2018). On the other hand, corporations operating in a corrupt environment believe they can easier get away with ignoring regulations, as they can buy their sins off with gifts (Wu, 2014). Thus, there is enough evidence to believe corrupt institutional environments have more carrots than sticks to engage in CSiR.

In the light of institutional distances, the greater the distance in terms of corruption the more or less corrupt the host country’s institutional environment is in comparison to the home country. Thus, MNEs face a higher liability of foreignness and therefore more difficulties adapting to a more unfamiliar market. To reduce the liability of foreignness, MNEs could mimic business practices from local corporations. Moreover, they could also choose to cooperate with local corporations or governments, to maintain competitiveness and legitimacy in local the market (Pinto et al., 2008). However, these local governments and corporations can involve corrupt business practices. Moreover, adapting to a distant environment in terms of corruption could cause internal conflicts for MNEs as the host country’s practices could be in conflict with the home country’s way of doing business (Kostova and Roth, 2002). On the contrary, MNEs from corrupt environments could attempt to engage in corrupt business practices in a host environment where corruption is not tolerated. This would result into a scandal, which affect a MNEs CSiR performance score. Accordingly,

(25)

25

3.3 Conceptual model

Figure 1 summarizing the expected relationships between the institutional distances in terms of all indicators and the CSiR performance of MNEs.

Figure 1: conceptual framework

4. Methodology

4.1 Research Philosophy

(26)

26

CSiR. Lastly, the data was collecting from a cross-sectional perspective, given the stable nature of the four independent variables measuring the institutional environment (Kostova et al., 2019). Moreover, to eliminate the possibility of bias, a strict procedure was followed which will be described in section 4.2.

4.2 Data collection

The secondary data were retrieved from three different databases: Thomas Reuters ASSET 4 ESG, Bureau van Dijk ORBIS, and the World Bank Group. These databases were chosen for their high level of reliability and broad availability and accuracy of information. Firstly, the ASSET 4 database was used to obtain company information on CSiR. This databank offers environmental, social, and governance (ESG) information based on more than 250 key performance indicators (Refinitiv, 2019). More interesting for this study, ASSET 4 also provides ESG controversies scores that indicate CSiR behaviour. Additionally, Orbis was used to collect current subsidiary data well as the MNE-level control variables from the years 2017 and 2018, since ORBIS is well-known for its containing data of approximately 365 million companies’ around the globe (Bureau van Dijk, n.d.). Lastly, the World Bank Group database was used to collect the country-level control variables as well as data on the World Governance Indicators, which are six indicators that measure the quality of a country’s governance (Kaufmann et al., 2008). Hence, four of these indicators were used to calculate the institutional distance between MNE home and host countries.

4.3 Sample

(27)

27

at least one foreign affiliate within their level 1 subsidiaries. Moreover, corporations from which the ROA and/or the number of employees was unknown were dropped from the sample, since these two variables were going to be used as firm-level control variables. After executing all these steps, the final sample counted 272 MNEs from 40 home countries. Altogether these MNEs had 19929 subsidiaries distributed over 166 different host countries.

4.4 Measurement of variables 4.4.1 Dependent variable

The dependent variable in this research is CSiR. Until now, most academics have used the KLD measurements for measuring corporate irresponsible behavior (Strike et al., 2006; Tang et al, 2015; Kang et al., 2016). In contradiction to these studies, this research will measure CSiR using the controversies scores provided by the ASSET 4 database since the controversies scores provided by ASSET 4 are the closest to measuring CSiR behavior (Eccles, Ioannous and Serafeim, 2014; Drempetic, Zwergel and Klein, 2019). The controversy score is based on 23 ESG controversies topics and are calculated per corporation, using formula 1 and 2. This score reflects whether a corporation or one of its foreign affiliates has been involved in a scandal in one of the controversies ESG categories (appendix C) within one fiscal year. If a company has been involved in a scandal the corporation will be given the score 1. Similarly, corporation where no scandals have occurred in that particular fiscal year will be given a score of 0. Thereafter the corporations are benchmarked against other corporations that operate in the same industry. For example, a sample of 6 corporations, from which four corporations score 0 and 2 corporations score 1. The two corporations with one controversy will score: (0+ 2/2)/6 (*100%) = 17% while the other four corporations with no controversies score (2+4/2) /6 (*100%) = 67%. The polarity was originally negative, meaning the higher the score the lower the firm scores on controversies. However, this score was reversed for interpretation, using formula 3. Thus, after reversing the score, the score 0 indicates low on controversies and the score 100 indicates high on controversies. Although the controversies scores are split up into three different categories (environment, social, and governance) this study uses the overall controversies score, since this research focusses on the overall controversies behaviour of the firm.

(28)

28

(1) Formula controversies firm scores with few or no controversies = (amount of

controversies other companies + companies without controversies)/ the total amount of companies

(2) Formula controversies firm scores with controversies = (amount of controversies other

companies + companies with controversies)/ the total amount of companies

(") %&'&()& *+,% -&(./(0123& )3/(& ./( &13ℎ 3/05126 )3/(& = (100 − 3/2;(/'&(),&) )3/(&)

4.4.2 independent variables

Most studies measure institutional distance by taking the average of the home-host distance of all six World Governance Indicators (WGI) constructed by Kaufmann et al., (2008). However, this study will examine four of the six WGI indicators Government Effectiveness, Regulative

Quality, Rule of Law, and Control of Corruption separately to reveal which indicators have the

most effect on CSiR. Besides, this study focusses strictly on the year 2018 rather than focussing on multiple years given the stable nature of the WGI indicators (Kostova et al., 2019). Please refer to appendix D for a complete overview of the indicators and how they are constructed. After extracting the data of the WGI indicators, I calculated for each of the four WGI indicators the distance between the home and host country, by subtracting all the host countries’ institutional profile scores from the home countries’ institutional profile scores, as proposed by Keig et al. (2019). Moreover, all these scores were standardized by dividing all distances by the standard deviation of all home-host distances. Lastly, I multiplied the number of subsidiaries the firm had in each home country to give weight to each distance score. I repeated this process for each WGI, which resulted in 4 different MNE-level institutional distance scores for each indicator. The formulas 4,5,6, and 7 were used for the calculations of these scores and are stated below.

(4) =2);,;>;,/21? @,);123& AB = CAB! − AB"#

+;@&'(AB)

$

(29)

29 (5) =2);,;>;,/21? @,);123& %E = C%E_+;@&'(%E)! − %E"# $ !%& (6) =2);,;>;,/21? @,);123& %G = C%G_+;@&'(%G)!− %G"# $ !%& (7) =2);,;>;,/21? @,);123& ** = C**_+;@&'(**)! − **"# $ !%&

Where , denotes the ,'( subsidiary of the MNE; 2 indicates the number of subsidiary locations

for the MNE; Q* denotes the home country of the MNE; AB is used as an abbreviation for

government effectiveness and denotes the respective value; %E is used as an abbreviation for

regulative quality and denotes the respective value; %G is used as an abbreviation for regulative

quality and denotes the respective value; ** used as an abbreviation for control of corruption

and denotes the respective value. Lastly, +;R@&' denotes the standard deviation of each measurement across the MNE’s subsidiary locations. Moreover, all independent variables were transformed by taking the natural logarithm of all variables in order for the data to be as "normal distributed" as possible and, thus, increase the validity of the associated statistical analyses (Feng, Wang, Lu, Chen, He, Lu and Tu, 2014).

4.4.3 control variables

To capture the effect of institutional distance on CSiR performance this study included several control variables from different levels: firm-level, country-level, and industry level.

Firm-level controls

(30)

30

this study controls for firm size by taking the natural logarithm of number of the employees. This measurement is widely used as a demographic firm characteristic (Klarner and Raisch, 2013). Indeed, firm size has also been found to positively affect CSiR performance, since larger corporations have more difficulties managing their social relationships due to the high complexity that comes with their size (Strike et al., 2006). Moreover, financial performance should also be taken into consideration, since corporations are more likely to engage in CSiR when they experience financial strains (Campbell, 2007; McWilliams and Siegel, 2001). Accordingly, the return on assets was used as a proxy to measure firm performance.

Country-level controls

Furthermore, two country-level controls were used to control for the socio-economic conditions of the host countries. Baughn, Bodie and McIntosh (2006) found a direct link between economic and social conditions and CSR. Correspondingly, this research controls for these conditions. Firstly, GDP per capita reflects the level of the standard of living and economic growth in a country. This variable was transformed into the natural logarithm of

GDP per capita. Moreover, this study controls for human development, using the HDI index

for each host country (Halkos and Skouloudi,2016). Both control variables were extracted from the World Bank database for the years 2018 and 2017. Data from 2018 is used for the main model, while one of the robustness tests will be performed using the control variables of 2017.

Industry level controls

(31)

31

TABLE 1:

Variables description, measurement, and sources

Variable Description or measurement Data sources

Independent variables

ID GE Natural logarithm value of the calculated distance in terms of government effectivness

World Governance Indicators

ID RQ Natural logarithm value of the calculated distance in terms of regulative quality

ID RL Natural logarithm value of the calculated distance in terms of rule of law

ID CC Natural logarithm value of the calculated distance in terms of control of corruption

Depedent variable

CSiR Performance Controversy score of MNE, measured by the number of scandels involved in 2018, scale 0-100 Thomas Reuters ASSET4

Control variables

nr. of employees Natural logarithm value of the number of people that are directly employed by the MNE

ROA Natural logarithm value of Return on Assets, measured for 2018 and 2017

Bureau van Dijk's ORBIS database

nr. of host countries Internationalisation in terms of scope. This number indicates the numer of host countries that the MNE is present.

Bureau van Dijk's ORBIS database

HDI

The human Development index score, indicates the actual level of human development in a given country.

Measured for 2018 and 2017 World Bank database

GDP p.c.

Natural logarithm value of GDP per capita, measured for 2018 and 2017

Work Bank database

(32)

32

4.5 Data analysis

To test the relationship between institutional distance and CSiR performance the Ordinary Least Squared (OLS) regression was performed using STATA 16.1 as a statistical software program. Besides, a logit regression was performed as an additional test, to analyse all four hypotheses. Moreover, two additional OLS regressions were performed to test the effect of the directions of distance on CSiR performance. The first step before conducting regressions is determining whether the assumptions for several principals are met (Verma and Abdel-Salam, 2019). Based on the results from preliminary tests, all assumptions were met except the assumption for multicollinearity. Despite all independent variables were standardized in order to reduce multicollinearity, the high VIF values indicated the presence of multicollinearity. This was not surprising and can be explained by the way the World Governance Indicators are constructed. All four measurements are closely related since they all measure distinct segments of the institutional environment. Nonetheless, to solve the issue of multicollinearity all independent variables were tested in separate regressions. For an extensive analysis of the assumptions for parametric tests, please refer to appendix E.

4.6 Descriptive statistics

(33)

33

TABLE 2: Descriptive Statistics

N Mean Median min max skewness kurtosis Independent variables ID GE 272 2.034 1.955 -.789 6.415 .377 2.585 ID RQ 272 2.088 1.947 -3.314 6.46 .309 3.214 ID RL 272 2.208 2.265 -1.557 6.53 .091 2.896 ID CC 272 2.304 2.166 -2.927 6.776 .183 3.348 Dependent variable CSiR performance 272 47.054 46.26 4.570 89.06 .031 1.651 Firm-level controls nr. of employees 272 7.423 7.765 0 12.573 -.590 2.847 ROA 2018 272 1.384 3.845 -97.260 89.78 -1.165 14.834 ROA 2017 271 .532 2.66 -83.610 62.07 -2.031 13.905 nr. of host countries 272 10.029 5 2 104 3.516 18.399 Country-level controls Gdp per capita 2017 272 10.421 10.716 7.591 11.584 -1.758 5.243 Gdp per capita 2018 272 10.465 10.741 7.606 11.667 -1.765 5.262 HDI 2017 272 .888 .919 .543 .953 -1.997 5.912 HDI 2018 272 .890 .920 .543 .981 -1.996 5.998

Note: all independent variables are standardized and from 2018

TABLE 3:

Descriptive statistics industry dummy

Frequency Percentages Cumulative

Oil and Gas (0) 75 27.57 27.57

Mining (1) 192 72.43 100.00

(34)

34 TABLE 4: Correlation matrix Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) Independent variables (1) ID GE 1.00 (2) ID RQ 0.77*** (0.0090) 1.00 (3) ID RL 0.86*** (0.0000) 0.94*** (0.0000) 1.00 (4) ID CC Dependent variable 0.81*** (0.0000) 0.96*** (0.0000) 0.98*** (0.0000) 1.00 (5) CSiR Performance Control variables 0.13** (0.0369) 0.14** (0.0253) 0.19*** (0.0020) 0.17*** (0.0046) 1.00 Firm-level (6) nr. of employees 0.13** (0.0320) 0.14** (0.0217) 0.19*** (0.0022) 0.17*** (0.0032) 0.50*** (0.0000) 1.00 (7) ROA 2018 0.01 (0.9095) 0.02 (0.7060) 0.02 (0.7059) 0.02 (0.7399) 0.22*** (0.0002) 0.22*** (0.0003) 1.00 (8) ROA 2017 0.02 (0.7635) 0.02 (0.7060) 0.03 (0.5796) 0.04 (0.5489) 0.22*** (0.0003) 0.27*** (0.0000) 0.24*** (0.0001) 1.00 (9) nr. of host countries 0.43*** (0.0000) 0.47*** (0.0000) 0.60*** (0.0000) 0.58*** (0.0000) 0.47*** (0.0000) 0.44*** (0.0000) 0.11* (0.0812) 0.07 (0.2262) 1.00 Country-level (10) GDP p.c. 2018 0.36*** (0.0000) 0.29*** (0.0000) 0.28*** (0.0000) 0.27*** (0.0000) -0.18*** (0.0034) 0.27*** (0.0000) -0.04 (0.5036) -0.12** (0.0486) -0.01 (0.9346) 1.00 (11) GDP p.c. 2017 0.35*** (0.0000) 0.28*** (0.0000) 0.27*** (0.0000) 0.26*** (0.0000) -0.18*** (0.0027) (0.0000) -0.27*** -0.04 (0.4851) -0.12* (0.0526) -0.02 (0.7905) 1.00*** (0.0000) 1.00 (12) HDI 2018 0.28*** (0.0000) 0.29*** (0.0000) 0.26*** (0.0000) 0.24*** (0.0000) -0.14** (0.0216) -0.24*** (0.0001) -0.06 (0.2896) -0.11* (0.0733) -0.03 (0.6585) 0.83*** (0.0000) 0.83*** (0.0000) 1.00 (13) HDI 2017 Industry-level (dummy) 0.28*** (0.0000) 0.30*** (0.0000) 0.26*** (0.0000) 0.25*** (0.0000) -0.13** (0.0296) -0.24*** (0.0001) (0.2529) -0.07 -0.09 (0.1200) -0.02 (0.7037) 0.82*** (0.0000) 0.82*** (0.0000) 0.99*** (0.0000) 1.00 (14) Mining 0.05 0.01 0.04 0.03 0.03 0.07 -0.02 0.06 0.02 -0.15** -0.15** -0.15** -0.14** 1.00 (0.4248) (0.9041) (0.5588) (0.5588) (0.6103) (0.2255) (0.7935) (0.3305) (0.7485) (0.0154) (0.0116) (0.0152) (0.0250)

(35)

35

4.7 Correlation matrix

Table 3 illustrates the correlation matrix of all variables. This table shows that all independent variables are correlated to the dependent variable. The distance in government effectiveness is correlated to CSiR r (0.13) at p=0.0369, while the distance in regulative quality is correlated with CSiR performance r (0.14) at p=0.0253. Furthermore, the distance in rule of law significantly correlates with CSiR performance r (0.19) at p=0.0020. Lastly, control of

corruption is correlated to CSiR performance r (0.17) at p=0.0046.In addition, table 3 shows

high multicollinearity between all independent variables (r > 0.7), despite the transformation of the independent variables to reduce multicollinearity. The VIF mean showed a value of 31.34, which is above the tolerated value of VIF 10. Thus, there is an issue of multicollinearity and therefore all independent variables will be tested separately. Additionally, the control variables HDI 2017 and HDI 2018 as well as the GDP per capita 2017 and GDP per capita 2018 are highly correlated. However, since these variables will not be used in the same models, as GDP per capita 2017 and HDI 2017 will be used for the robustness test, this is not seen as a problem.

4.8 Robustness tests

Lu and White (2014) argue “robustness checks” should be performed to research how certain “core” regression coefficient act when modifying regression specifications. This type of fragility regression coefficient estimator can indicate specification errors and helps with revealing misspecifications (Leamer, 1983). Accordingly, two robustness tests were performed. In the first robustness tests the control variables of 2018 were replaced with data of 2017, which is a typical way of checking the preformed tests’ robustness (Lu and White, 2014). The second robustness test re-runs the same tests while excluding the observations from the most common home country within the sample, as proposed by Ioannou and Serafeim (2012). Correspondingly, the regressions were re-run excluding MNEs from Australia. This was done to mitigate the level of bias of the entire sample. The results of the robustness tests are described in the next chapter.

5. Results

(36)

36

model to test the proposed hypotheses. Firstly, all control variables were tested in model 1, to identify whether the control variables have significant explanatory power. Thereafter, each independent variable was tested one by one in separate models (model 2-5). Table 4 presents the results. Interestingly, all firm control variables are significantly correlated to the dependent variable, while none of the country control variables are significantly correlated to CSiR performance. Moreover, the R-squared remains stable around 0.35-0.36, which indicates that in all models the independent variable explains around the same the amount of variance for CSiR performance.

5.1 Results main model

(37)

37

TABLE 5:

Results main model OLS Regression

Model Model Model Model Model

Variables (1) (2) (3) (4) (5) Indepedent variables ID GE -3.38*** (1,26) ID RQ -1.93 (1.18) ID RL -2.96** (1.14) ID CC -2.57** (1.18) Firm control variables

nr. of employees -3.25*** -2.89*** -3.10*** -2.92*** -3.06*** (0.59) (0.60) (0.60) (0.60) (0.59) ROA 2018 -0.18* -0.20** -0.19** -0.19** -0.20* (0.08) (0.08) (0.08) (0.08) (0.08) nr. of host countries -0.63*** -0.43*** -0.50*** -0.42*** -0.45*** (0.11) (0.13) (0.13) (0.13) (0.14) Country control variables

HDI 2018 -42.03 -46.91 -37.56 -45.81 -32.78

(63.37) (62.67) (63.24) (62.72) (63.07)

GDP per capita 2018 5.47 5,72 4.84 5.72 4.59

(5.81) (5.74) (5.80) (5.74) (5.78) Industry control variables

oilgas_mining 0.56 0.56 0.65 0.28 0.72 (2.99) (2.96) (2.98) (2.96) (2.97) Constant -42.54** -38.57** -38.33** -39.46** -38.96** (18.70) (18.55) (18.83) (18.54) (18.64) Observations 272 272 272 272 272 R-squared 0.35 0.36 0.35 0.36 0.36 Adjusted R-squared 0.33 0.35 0.34 0.35 0.34

(38)

38

5.2 Additional test 5.2.1 Logit regression

Although the assumptions for a parametric test were met, the normality tests revealed that the CSiR performance scores are slightly divided into two groups: “low on CSiR score” and “high on CSiR score”. Consequently, an additional test was done using the Logit regression. To do so, CSiR score was recomputed into a binary variable, by dividing the sample into two groups and splitting them at the median (46) of the sample. As a result, all corporations scoring low on controversies (0-46) were labelled as 0, and all corporations that scored high on controversies (47-100) were labelled as 1. Table F1 in appendix F illustrates the outcomes of the logit regression. The results of the logit regression were similar to the OLS main model in terms of the negative coefficient, except for model 4 that showed a positive (insignificant) coefficient for the relationship between rule of law and CSiR performance. Moreover, all the coefficients of the logit regressions were insignificant. In addition, a multi-level logit regression was performed, since the raw data for the independent variables was at start multi-level but converted into MNE-multi-level. The results of the multi-multi-level logit regression gave the same values as the logit regression. Thus, according to the (multi-) logit regression there is not enough evidence to support hypotheses 1, 2, 3 and 4. Comparing these results to the main model, the logit regression also resulted in negative coefficients, but insignificant at all significant levels. Interestingly, model 3 testing the relationship between institutional distance in regulative quality and CSiR performance did show a positive (insignificant) relationship, while the main OLS model showed a negative insignificant coefficient. Thus, both the Logit regression and the OLS regression find no significance for all hypotheses. Additionally, the pseudo-R-squared is 0.25 in all models. In comparison to the main models R-square of approximately 0.35, the OLS model appears to be of better fit.

5.2.2 Directional Institutional Distance test

(39)

39

distance can lead to reaping benefits from greater opportunities of institutional arbitrage, such as the so-called escape-based arbitrage advantage, in which MNEs can escape from the constraints posed by the formal institutions of their home country (Shirodkar and Konara, 2017). For example, MNEs choose to invest in locations with lower CSR standards (Dam and Scholtens, 2008). These locations are called “pollutions havens” and are characterized by high corruption, poverty, and weak institutions (Mani and Wheeler, 1997). On the contrary, MNEs from weaker institutional environments that operate in higher institutional environments are less likely engage in CSiR, since the high-quality institutions would not let corporations get away with such behaviour (Campbell, 2007).

Thus, the direction in institutional distances offers insights on the quality of the institutional environment that companies internationalize to, which can be interesting with regard to CSiR performance. Correspondingly, two additional tests were done in the same set-up as the main model. Firstly, an OLS regression was performed on the sample of companies internationalizing to weak institutional environments in terms of all indicators. Table F2 in appendix F illustrates the outcomes of the tests. Out of all five models, only model 4 (rule of law) showed a negative significant coefficient (β= - 81.32, p=0.080). This indicates MNEs operating in weaker institutional environments in terms of rule of law, will have a negative effect on CSiR performance. This is in line with the results from the main model, however the main model does not consider directions of institutional distance. Secondly, another OLS regression was performed with the sample of MNEs internationalizing to institutional stronger countries, in terms of all indicators. Table F3 in appendix F illustrates the outcomes of the tests. However, for testing the negative signs of the four independent variables, none of the models appeared to be positively significant correlated to CSiR performance. Instead, all values were negative and insignificant.

5.3 Robustness tests

To test the strengths of the statistical models two robustness tests were performed, which will be described and compared to the main model in this section.

(40)

40

In addition, a second robustness test was performed using all variables of 2018, but excluding the Australian home-based companies. Consequently, 58 MNEs were omitted from the sample and an another OLS regression was performed with a sample of N=214, where N denotes the number of observations. Table G2 in appendix G illustrates the results of OLS regression. Once again, no support was found for hypotheses 1, 2, 3 and 4, since all coefficients showed negative coefficients. This is in line with the main OLS model and the first robustness check. In addition, model 2 showed a weaker level of significance (decrease from p<0.01 to p<0.5) in comparison to the main OLS regression. In the same light, the significance level of model 4 testing the distance in terms of rule of law, has also declined (from p<0.01 to p<0.5). Yet all coefficients remained significant, which were also significant the main regression. Moreover, the signs of the coefficients remained the same as in the main OLS.

In sum, both robustness test shows similar results as the main model. This implies the results are robust and the hypotheses are not supported. In the subsequent chapter, these results will be reviewed and discussed.

6. Discussion

This study was conducted to open the black box on the effect of institutional distance on CSiR by studying 272 MNEs operating in the extractive industry, more specifically the oil and gas sector and the mining sector. To measure this mechanism, insights from two bodies of literature have been brought together, namely the socially responsibility theory and theory on institutional distances. To the best of my knowledge, this mechanism has not been studied before. Accordingly, this study contributes significantly to the literature in theoretical and in practical ways. In this chapter, the implications of the results chapter will be analyzed and discussed, followed by a section on the limitations and suggestions for further research.

6.1 Implications

Opening the Black Box: The Effect on Institutional Distance of Corporate Social Irresponsibility