DEPARTMENT OF POLITICAL SCIENCE GOVERNANCE BEYOND BORDERS The Extraterritorial Reach of OECD National Contact Points for Responsible Business Conduct Mirja Sund Master’s Thesis: 30 credits Master’s Programme in International Administration and Global Programme: Governance Date: 22 May 2023 Supervisor: Marina Nistotskaya & Magnus Åsblad Words: 13,555 Abstract As increasingly powerful actors in a globalised world, multinational corporations are often able to evade responsibility for human rights violations and other misconduct occurring in their own operations or supply chains. Since courts face limitations when attempting to exert jurisdiction outside of their state territories, state-based non-judicial mechanisms such as the OECD National Contact Points (NCPs) have been established to fill this gap by mediating between corporations and victims of corporate human rights abuses with the aim of providing remedy to the latter. The NCPs can accept complaints about corporate conduct outside of their own territories, which raises the question of whether these complaints are similarly effective as domestic cases. To investigate the impact of such extraterritorial powers on the outcome of the cases, this thesis employs a mixed-method approach. A quantitative part uses logistic regression analysis to test in a sample of 233 NCP cases between 2000 and 2022 whether extraterritoriality at least partially determines the success of complaints. The results show that extraterritorial cases are less likely to end in an agreement, although this relationship can only be detected in the data after 2011. Following this section, a qualitative part analyses two cases in a comparative case study to identify potential mechanisms that explain these findings. It finds that trust between the complainants and defendants is a key requirement for the success of the mediation, and that cultural differences and an imbalance in power can make it more difficult for NCPs to build trust between the parties. Keywords: Business and human rights, Non-judicial mechanisms, Extraterritoriality, OECD Guidelines for Multinational Enterprises, OECD National Contact Points 2 List of Abbreviations ING Internationale Nederlanden Groep NCP National Contact Point OECD Organisation for Economic Co-operation and Development OLS Ordinary Least Squares RSPO Roundtable on Sustainable Palmoil SDI Sustainable Development Institute TUAC Trade Union Advisory Committee UNGPs United Nations Guiding Principles on Business and Human Rights VIF Variance Inflation Factor WALHI Wahana Lingkungan Hidup Indonesia 3 Table of Contents 1 Introduction ....................................................................................................................... 5 2 Literature review and theoretical framework ................................................................ 7 2.1 The history of National Contact Points ....................................................................... 7 2.2 The strengths and weaknesses of NCPs ...................................................................... 8 2.3 Extraterritoriality in business and human rights .......................................................... 9 2.4 Hypotheses ................................................................................................................. 11 2.5 Causal mechanisms .................................................................................................... 12 3 Quantitative analysis ...................................................................................................... 14 3.1 Data and methodology ............................................................................................... 14 3.2 Operationalisation ...................................................................................................... 16 3.2.1 Dependent variable: Case outcome .................................................................... 16 3.2.2 Independent variable: Extraterritoriality ............................................................ 16 3.2.3 Control variable: NCP quality ............................................................................ 17 3.2.4 Control variable: Host OECD membership ....................................................... 18 3.2.5 Control variable: Year dummies, and pre-2011 and 2011-2022 periods ........... 19 3.3 Results ....................................................................................................................... 20 3.4 Regression diagnostics and robustness check ........................................................... 25 4 Qualitative analysis ......................................................................................................... 27 4.1 Methodology and data ............................................................................................... 27 4.2 Case 1: Dutch NGOs vs. ING Bank .......................................................................... 29 4.3 Case 2: Milieudefensie et al. vs. ING ........................................................................ 31 4.4 Analysis ..................................................................................................................... 32 4.4.1 Trust ....................................................................................................................... 32 4.4.2 Power ..................................................................................................................... 34 4.4.3 Culture and language ............................................................................................. 35 5 Conclusion ....................................................................................................................... 36 6 References ........................................................................................................................ 38 7 Appendix .......................................................................................................................... 44 4 1 Introduction In 2015, 168 former employees of the Congolese company Bralima, a subsidiary of the Dutch brewing company Heineken, accused Bralima of having violated their human rights by unlawfully dismissing them under the authorisation of the rebel movement RDC-Goma. One and a half years later, the former employees and Heineken reached an agreement that all parties deemed satisfactory and, according to reports, included monetary compensation to the complainants (Nieuwenkamp, 2018). This agreement was made possible by the Dutch National Contact Point for Responsible Business Conduct (NCP). NCPs are agencies established to promote the Guidelines for Multinational Enterprises set up by the Organisation for Economic Co-operation and Development (OECD), and to handle complaints about alleged breaches of the Guidelines by corporations. Currently, such NCPs exist in 50 countries. As a grievance mechanism, one of their main tasks is the handling of complaints, referred to as the specific instance procedure. Since they do not have judicial powers, they primarily offer their good offices to facilitate dialogue between the parties with the aim of reaching an agreement. As in the Heineken case, NCPs are not limited to allegations located within their own territories but also have the mandate to accept complaints about companies headquartered in their territories that operate abroad. Since Heineken is a Dutch company, the Dutch NCP was qualified to mediate, although the alleged violation of the Guidelines took place in the Democratic Republic of the Congo. This feature distinguishes NCPs from judicial courts that are traditionally limited to acts within their jurisdiction. It allows the countries that adhere to the OECD Guidelines to address human rights violations and other misconduct that occurred in the operations of multinational corporations or their supply chains abroad. These extraterritorial competences are especially relevant when the host country is not willing or able to provide remedy to the victims by other means. Scholars have highlighted the potential of this extraterritorial reach as a special feature filling a gap in the current international framework on corporate accountability (Achtouk- Spivak & Garden, 2022; Buhmann, 2019). Other research has analysed the current practices of the NCPs and addressed their strengths and weaknesses that determine the outcome of specific instances (Černič, 2008; Daniel et al., 2015; Macchi, 2017; Perillo, 2022). However, little attention has been paid to the relationship between extraterritoriality and whether a case ends in an agreement. Even beyond the focus on NCPs, the literature on extraterritoriality in the field of corporate accountability rarely pays attention to the potential challenges coming with a dialogue that does not only involve actors from the private sector, the public sector and civil 5 society but also individuals from different countries and cultural backgrounds. As one of only few mechanisms with a comparable power, NCPs provide a fitting case to study the impact of extraterritoriality on the ability of grievance mechanisms to deliver justice to victims of corporate misconduct. Therefore, this thesis addresses the following research question: How does extraterritoriality affect the outcome of specific instances handled by the OECD National Contact Points? While the NCPs still have a special status in their field, it is likely that similar mechanisms will gain importance in the near future. The United Nations Guiding Principles on Business and Human Rights (UNGPs), which make explicit mention of the NCPs as an example of state-based non-judicial grievance mechanisms, are the first global instrument designed to address corporate human rights responsibilities. In 2014, the UN Human Rights Council tasked an intergovernmental working group with the development of a binding international treaty on business and human rights. The draft for this treaty contains the requirement for states to establish non-judicial mechanisms to provide access to remedy for victims of corporate human rights violations (OEIGWG, 2021). Furthermore, it includes the expectation that companies are not only responsible for their own operations but also for operations linked to them through their business relationships. In light of this development, the findings of this thesis may contribute to an upcoming debate beyond the NCP grievance system. If institutions with similar - and perhaps stronger - powers emerge, it is likely that they will face similar challenges in the field of extraterritoriality. The thesis starts by reviewing the key literature on NCPs and extraterritoriality. It introduces the theoretical framework and the resulting hypotheses. To answer the research question, the thesis employs a mixed-method approach, consisting of a quantitative and a qualitative part. The combination of multiple methods allows researchers to approach a problem from different perspectives and to balance each method’s strengths and weaknesses (Gerring & Christenson, 2017). The quantitative part employs logistic regression to test whether a relationship between extraterritoriality and the outcome of specific instances can be detected in the data on 233 NCP cases. It finds that there is a negative relationship between the variables, meaning that extraterritorial cases are less likely to end in an agreement. Since this result does not show why such a relationship exists, it is followed by a qualitative comparative case study. The two cases are largely similar but differ in two aspects: The first case was not extraterritorial and ended in an agreement, whereas the second case was extraterritorial and did not end in an agreement. An analysis of these cases does not show a direct impact of factors such as cultural 6 and linguistic differences but instead illustrates the role of trust in the dialogue and, to a limited extent, the perception of equal power between the participants. Accordingly, the thesis finds that extraterritoriality does have an impact, although it is less straightforward than originally assumed. 2 Literature review and theoretical framework 2.1 The history of National Contact Points The OECD Guidelines for Multinational Corporations are a set of recommendations adopted by governments from 50 countries to promote responsible business conduct by multinational corporations operating or headquartered in their territories. As the Guidelines are not legally binding, each adhering country is required to establish a National Contact Point (NCP) to oversee their implementation. Next to other responsibilities, NCPs investigate complaints about failures of multinational corporations to comply with the Guidelines, referred to as specific instances. After receiving complaints, the NCPs make a preliminary assessment and, if they decide to accept the complaint, attempt to mediate between the parties to find a common solution to the issue. In the past, successful cases have resulted in a statement acknowledging misconduct by the company, a change to the company’s policy or procedures regarding due diligence, and measures directly improving the situation for the victims (Daniel et al., 2015). Due to the non-binding nature of the Guidelines, the outcome largely depends on the willingness of corporations to cooperate and take part in the mediation process. The complaints can be submitted by different parties such as affected communities, employees, civil society organisations and trade unions. The Guidelines are the only international instrument addressing responsible business conduct that contain such a grievance mechanism (OECD, 2011). Non-OECD countries that adhere to the Guidelines are Argentina, Brazil, Bulgaria, Croatia, Egypt, Jordan, Kazakhstan, Morocco, Peru, Romania, Tunisia, Ukraine, and Uruguay. In practice, these NCPs have received very few complaints, with the exception of Brazil and Argentina. The Guidelines were first introduced in 1976 and revised five times, most recently in 2011. National Contact Points were officially part of the OECD Guidelines since 1984 but were only given their current role in the 2000 review of the Guidelines which established procedural standards for NCPs (OECD, 2016). The 2000 review also contained chapters on disclosure, environmental and labour standards, bribery, consumer interests, science and technology, competition and taxation. The most common issues raised in NCP cases under the 2000 Guidelines were labour rights and, in later years, the environment (Ruggie & Nelson, 2015). 7 An update to the Guidelines in 2011 added a chapter on human rights in line with the UN Guiding Principles on Business and Human Rights (UNGPs). It extended the responsibilities of corporations from their own operations to their supply chains and other business relationships (OECD, 2016). The 2011 Guidelines also introduced the concept of due diligence, here defined as “the process through which enterprises can identify, prevent, mitigate and account for how they address their actual and potential adverse impacts as an integral part of business decision-making and risk management systems” (p. 23). After 2011, the number of complaints concerning human rights as well as cases involving the manufacturing and financial sector increased, presumably due to the extended scope of the Guidelines (Ruggie & Nelson, 2015). After introducing the main structure and functions of the NCPs, the next section reviews the current state of the literature on the topic. 2.2 The strengths and weaknesses of NCPs The impact of the NCP system on the conduct of multinational corporations is disputed. An optimistic view holds that the existence of the specific instance procedure is enough to encourage compliance with the OECD Guidelines among some companies (Perillo, 2022). The procedure has several strengths that do not require the competence to impose sanctions: NCP cases have been used as a basis for legal proceedings, to damage a company’s reputation and to deter governmental or financial institutions from supporting the defendant in further operations (Achtouk-Spivak & Garden, 2022). In contrast to litigation, the specific instance procedure has the advantage of being comparatively cheap and accessible (Macchi, 2017). Despite acknowledging this potential of specific instances, Cîrlig finds that “when a conflict arises between pursuing profits, and observing human rights, soft law might be just a bit too soft to lean the balance in favor of human rights” (2016). Among the most criticised aspects is the inability of NCPs to impose sanctions for non-compliance and the lack of remedy provided to the victims (Černič, 2008; Daniel et al., 2015). Khoury and Whyte argue that the specific instance procedure lacks effectiveness because of the refusal of companies to participate in the process, and that the update to the Guidelines in 2011 has not improved this deficiency (2019). However, even beyond the direct impact of the NCPs on business conduct, the wide government support of the Guidelines and subsequent attention brought to the issue is often considered an important step towards the international regulation of corporate responsibility (Černič, 2008). Finally, Buhmann states that, as “non-judicial institutions, NCPs should be assessed in regard to what they can do, not what they cannot” (2019, p. 19). 8 General assessments of the system’s effectiveness are complicated by the large differences between individual NCPs. The principle of functional equivalence set by the Guidelines requires adhering states to establish NCPs that fulfil certain standards but may differ in their institutional setup. Structures range from monopartite NCPs consisting of members of a single ministry to independent agencies that are supported by civil servants from one or multiple ministries. The number of staff members and their access to economic resources vary, with some NCPs only having limited capacity to address complaints (Macchi, 2017). Consequently, the workload of the institutions differs substantially: Until 2022, several NCPs have handled more than fifty cases, while others have received no complaints at all (OECD, 2023). This diversity is viewed critically by some scholars. In contrast to the OECD’s argument that increased flexibility enhances the effectiveness of NCPs, Sanchez (2015) finds that “the fundamental differences that currently exist among NCPs […] are not well justified in normative terms” (p. 21). Moreover, the structure of an NCP can affect its perceived independence, with monopartite compositions being more likely to be suspected of bias towards the government’s economic interests (Buhmann, 2019; Daniel et al., 2015). This aspect is important when victims of corporate abuse decide to file a complaint to a specific NCP because of its reputation. Next to the differences in NCP structures and practices, scholars have pointed out regional differences in the work of NCPs. Ryngaert and Wouters (2009) note the relatively low number of complaints filed to the NCP in the United States given the large number of multinational corporations headquartered in the country and attribute this observation to legal and cultural differences: Victims of corporate abuse in the US may prefer legal action to maximise media attention while Europeans are more likely to seek dialogue through informal conflict resolution mechanisms (p. 974). As measured by concrete improvements for victims, EU-based NCPs, in particular in Belgium, France, Germany, the Netherlands and the United Kingdom, are among the most successful (Macchi, 2017). Moreover, the prevalence of some issues differs substantially. For instance, nearly all complaints concerning human rights or due diligence until 2018 were filed to NCPs in North-Western Europe (Buhmann, 2019). Such individual and regional differences indicate that any generalising statements about the work of NCPs must be made with caution. 2.3 Extraterritoriality in business and human rights After reviewing the academic debate on NCPs, this section continues with a discussion of the concept of extraterritoriality. The term extraterritoriality can take on a variety of meanings. 9 Traditionally, international law has been primarily governed by the territorial principle: States only have jurisdiction over alleged offences committed on their territory (Akehurst, 1973). Other principles, such as the nationality principle in which a state exerts jurisdiction over its nationals outside of its own territory, are therefore known as extraterritorial jurisdiction (Akehurst, 1973). In the history of state-building, the notion of extraterritoriality constituted an exception to the fundamental principal of territorial sovereignty and has thus often been viewed critically and associated with interventionism and legal hegemony (Handl, 2012). To clarify the concept, Addis (2012) distinguishes between ordinary and extraordinary extraterritorial jurisdiction. In ordinary cases, states assert jurisdiction outside of their territories while still having a connection to the activities in question, for instance through the citizenship of the perpetrators or victims. Cases of extraordinary extraterritorial jurisdiction do not assume such a connection, and commonly refer to customary international law instead (2012, p. 18). Another meaning of the term extraterritoriality refers to the exemption of individuals, mostly diplomats, from the jurisdiction of local law (Pal, 2019). In this thesis, the term is used only in the context of corporate accountability. In the field of business and human rights, extraterritorial jurisdiction is often seen as a promising solution to the corporate accountability gap arising from the operations of transnational corporations and their supply chains: When the host states of business operations are unable to adequately respond to corporate human rights abuses, the home states of the respective companies may step in (De Schutter, 2006; Bernaz, 2013). Correspondingly, the definition by Bernaz (2013) encompasses “any measure taken by the state aiming at enhancing corporate accountability for acts committed abroad” (p. 496). This broader definition supports the call by Zerk (2010) to view extraterritoriality not only as the direct assertion of jurisdiction over foreign territory but also as the use of domestically imposed measures that influence the conduct of actors abroad. Examples of such measures with extraterritorial implications are requirements to parent companies to supervise and report on the conduct of their subsidiaries. The UNGPs reflect a similar position. The second principle of the state duty to protect human rights permits, but does not require, states to take measures against human rights violations by corporations operating from their territories, even if these operations are located abroad (UN, 2011). To achieve this objective, the author Ruggie distinguishes between direct legislation over conduct abroad and domestic measures with extraterritorial implications such as international soft-law instruments, under which he lists the OECD Guidelines for Multinational Enterprises. Critics of the UNGPs argue that this principle fails to make the 10 necessary distinction between obligations imposed on states by international human rights law and own policy initiatives, which they doubt have comparable incentives to protect human rights (Augenstein & Kinley, 2013). Moreover, they propose that such international extraterritorial obligations are already implied by the state duty to regulate the conduct of non- state actors, which includes any actors that are primarily located within the state’s jurisdiction, even when they are operating abroad (Augenstein & Kinley, 2013; De Schutter, 2015). In contrast, Methven O’Brien (2016) supports the position of the UNGPs that states are currently not obliged to take extraterritorial measures but are permitted to do so. The ability of the NCPs to address corporate misconduct outside of their own territories is frequently highlighted as one of their strengths that distinguishes them from most courts (Achtouk-Spivak & Garden, 2022; Buhmann, 2019). International courts with extraterritorial powers rarely provide remedy for the victims and often do not have the mandate to address corporations (Buhmann, 2019). Moreover, NCPs from different countries can cooperate with each other, and some complaints are directly filed to multiple NCPs (Achtouk-Spivak & Garden, 2022). Buhmann states that, with regard to their extraterritorial competences, “NCPs currently fill a gap that neither national law, nor international law, nor private corporate social responsibility (CSR) schemes cover” (2019, p. 10). Catá Backer (2012) goes as far as arguing that the OECD Guidelines as a framework for corporate governance show the future possibility of a system governed by multinational corporations instead of states, making the notion of territoriality irrelevant. Despite the praise for the potential of the Guidelines in this field, little to no research has been done to consider how such extraterritorial cases work out in practice. Criticism of the NCPs does not generally make a distinction between domestic and foreign complaints, and it is not known whether a noticeable difference exists in the outcome of such cases. Yet, taking a closer look at the challenges and benefits of non-judicial mechanisms located outside of the country of harm contributes to the scholarly debate even outside of the OECD framework. A comparison of extraterritorial and domestic cases can be helpful to identify possible weaknesses in the current system and to make suggestions, both for the improvement of the NCPs and for future mechanisms with similar mandates. 2.4 Hypotheses Based on the theoretical background established in the previous sections, the quantitative analysis aims to test the following hypotheses: 𝐻0: Extraterritoriality has no effect on the outcome of a case. 11 𝐻1: Extraterritorial cases are less likely to result in an agreement. 2.5 Causal mechanisms Due to the lack of research on the extraterritorial powers of non-judicial mechanisms and the connected challenges, this thesis does not draw on any established theories in this field. Instead, it identifies potential factors based on the literature that could explain the relationship between extraterritoriality and the outcome of cases. The first causal mechanism is trust, both towards the NCP and towards the opposing party. In the context of this research, Pruitt’s definition is used: Trust is defined, in this context, as a belief that the other party is also ready for coordination. Trust is necessary because coordinative behavior always lays one open (to a greater or lesser extent) to the possibility of being exploited. Such exploitation seems less likely to materialize if the other party also seems motivated to achieve coordination (1982, p. 16). Both sides face risks when entering an NCP-mediated dialogue. Civil society organisations have commonly invested time and resources, and, in some countries, face risks to their security when filing complaints (Daniel et al., 2015). On the other side, corporations fear reputational damage and economic consequences if the government of the country acts upon recommendations by the NCP (Achtouk-Spivak & Garden, 2022). The incentives of each party to cooperate also depend on the predicted alternative to a negotiated agreement: If the NCP determines whether a violation of the Guidelines had occurred in cases where the mediation is not successful, the complainant could hope to achieve their goals through public attention instead of an agreement, while corporations might be more inclined to cooperate (Sanchez, 2015). Consequently, trust in the good faith of the other parties is seen as essential to overcome the fear of exploitation (Davarnejad, 2011). The second potential mechanism introduced in the literature is power. Power in this context is defined with regard to the relationship between two actors and their ability to potentially overcome resistance by the other party (Emerson, 1962). Power asymmetry limits the foundations for reciprocity between actors, as one actor is less dependent on the cooperation of the other side, and is therefore more likely to create distrust in their relationship (Lee & Lee, 2022). This power imbalance is common in NCP-mediated dialogues, where communities affected by corporate misconduct often have little access to resources, especially in comparison to large multinational corporations (Bhatt & Erdem Türkelli, 2021). Accordingly, NCPs are considered to have twofold responsibilities where, depending on the use of the mechanisms at 12 their disposal, they can reinforce existing power structures or shift power to the victims of corporate abuse (2021, p. 433). In a case study of Lok Shakti Abhiyan et. al. vs ABP, a complaint about the construction of a Korean steel plant in India, Balaton-Chrimes and Haines (2017) criticise that the Dutch NCP did not take sufficient measures to address the imbalance of power between the parties. The perceived inequality finally led the complainants to withdraw from the process (Balaton-Chrimes & Haines, 2017). While Bhatt and Erdem Türkelli advocate for an international fund for victims to equalise the standing of the parties, Khoury and Whyte claim that a process with such extreme power imbalances between the parties “can never resemble a consensual process” (2019, p. 376). In this sense, a perceived power balance in the process is a key contributing factor to the establishment of trust. The third potential mechanism is culture. The role of cultural differences in the NCP process is severely under-researched and rarely addressed in the literature (e.g. Hackett et al., 2019). Nonetheless, the importance of culture in negotiations has been studied in behavioural research for a long time. Culture has been found to have an impact on a person’s strategies with regard to communication, social status and identity (Adair et al., 2004; Brett et al., 2000). In discussions on matters such as human rights violations, sensitivity to the cultural background of the participants can have a fundamental impact on their trust in the outcome and their willingness to participate. A closely related aspect is the role of language in the process. Bülow-Møller (1993) shows that non-native speakers of a language tend to be at a disadvantage when negotiating with native speakers, although they can occasionally use this imbalance to their advantage by requiring measures of accommodation from the other party. Since extraterritorial cases are commonly located in the home country of the company, complainants from other countries need to adapt to the language spoken in the country or to English as a lingua franca. Access to interpreters is primarily a matter of resources, which tend to be scarce for NCPs and civil society organisations (Macchi, 2017; Piewitt, 2010). In addition to linguistic differences, geographic distance between the actors involved is increasingly bridged by the use of digital tools. Research comparing videoconferencing to in-person meetings has found participants often find it harder to build trust in online settings (Ebner, 2017). Naturally, the different mechanisms are closely intertwined. Linguistic differences can create perceptions of power imbalance and, consequently, lower the trust between the parties. Conversely, efforts made to address these problems can help to restore the trust and achieve a mutually beneficial outcome. After establishing factors that could link the key independent variable to the dependent variable, the 13 following section aims to test whether such a relationship can be detected based on the available data. Then, the qualitative part of the analysis compares two similar cases to assess whether these mechanisms are able to explain their different results. 3 Quantitative analysis 3.1 Data and methodology The first analytical part of the thesis conducts a quantitative analysis of the link between extraterritoriality and the outcome of cases, utilising data of all completed cases in the NCP complaint system. This section elaborates on the selection of the data and the use of logistic regression. The data used for the analysis comes from OECD Watch’s database of specific instances. Three main databases capture the complaints filed to NCPs: The OECD database of specific instances, the OECD Watch complaints database and the database of the Trade Union Advisory Committee (TUAC). However, none of these bodies capture all complaints filed to NCPs from the beginning (Ruggie & Nelson, 2015). The OECD did not include any complaints that were rejected before 2011, and OECD Watch and the TUAC only include complaints filed by civil society and trade unions, respectively. The OECD Watch database was chosen here because it covers a wider range of issues than the TUAC database, and because it is more detailed and precise in determining the outcome of specific instances than the OECD. While the OECD database only classifies cases as “concluded”, “not accepted” and “in progress”, OECD Watch includes the options of “agreement”, “blocked”, “filed”, “no resolution”, “rejected”, “under review” and “withdrawn”. Between 2000 and 2020, NGOs submitted 40% of NCP cases compared to 26% by trade unions and 25% by individuals (OECD, 2020). Nevertheless, the exclusion of complaints by actors other than civil society imposes a limitation on the applicability of the results to other sectors. The cases were only selected according to their status at the time of writing. In instances where one complaint is filed against multiple companies, OECD Watch lists a case for every company involved. For example, if a single complaint against six companies is rejected by the NCP because it does not concern issues covered by the Guidelines, the database will list six rejected specific instances. As the unit of analysis is NCP cases, not companies, specific instances that concern different companies but are otherwise identical to already listed cases are excluded from the dataset used for the analysis. This step concerns 76 cases from the OECD Watch database, reducing the number of eligible cases from 309 to 233 in the final dataset. The choice of the variables was partially informed by interviews with two members of the secretariat 14 of the Dutch NCP and with a policy advisor and an intern at OECD Watch. OECD Watch is a network of more than 130 civil society organisations that aims to improve the implementation of the OECD Guidelines through advocacy and advice offered to NGOs on how to use the Guidelines. The interview with OECD Watch took place at their headquarters on 9th March 2023, and the interview with the Dutch NCP took place online on 20th March 2023. The interviewees preferred not to be named and are therefore referred to under their positions. While ordinary least-squares (OLS) regression is the most frequently used method in regression analysis, it requires a continuous dependent variable (Kellstedt & Whitten, 2018). Since the dependent variable is dichotomous, the use of OLS regression in this case would be problematic because of the lack of linear association between the key variables, estimated values outside of the 0-1 range and problems with heteroscedasticity (Mehmetoglu & Jakobsen, 2022). Hence, a more appropriate method to test the hypotheses is logistic regression. Instead of least squares, the logistic model uses maximum likelihood estimation. Maximum likelihood estimation determines the parameter values with the highest likelihood of producing the observed values. Therefore, the interpretation of logistic regression differs from OLS regression: The coefficient indicates the change in the log of the odds for agree = 1 if the independent variable changes by one unit (Mehmetoglu & Jakobsen, 2022). Since this interpretation is less intuitive, the predicted probability for each case to end in an agreement is then calculated. Next to logistic regression, another option for the analysis would be probit regression which employs standard normal distribution instead of logistic distribution to determine a binary outcome (Mehmetoglu & Jakobsen, 2022). The differences between logistic and probit regression are widely considered negligible when it comes to substantive results (Long, 1997). Since logistic regression offers a simpler approach to interpretation, it is the preferred method in the context of this thesis. 15 3.2 Operationalisation 3.2.1 Dependent variable: Case outcome The dependent variable is the outcome of cases, specifically whether there was an agreement or not.1 Since the direct effect of the process is difficult to quantify, the analysis uses a binary variable measuring whether there is an agreement. The OECD Watch complaints database distinguishes between the following outcomes: agreement, blocked, filed, no resolution, under review and withdrawn. Cases that are filed or under review are not completed yet and are therefore not considered in this analysis. The other options indicate the stages at which a specific instance can succeed or fail: The NCP can reject the complaint, the company might refuse to participate in the process, or there is no agreement after a lengthy mediation process. In addition, there have been several cases where the complainants withdrew from the process. Since a withdrawal can happen for multiple reasons, including success through another mechanism, these cases are not included in the analysis. 35 cases of the total 233 cases in the dataset ended in an agreement and are therefore coded as “1”, and the remaining 198 cases were rejected, blocked or ended in no resolution (coded as “0”). Descriptive statistics on this and all other variables are reported in the appendix (Table 1). The variable agree does not indicate whether a complaint failed because it was rejected by the NCP or because the subsequent dialogue was not successful. Therefore, a second variable, accept, was created to distinguish between cases that were accepted by NCPs and either ended in an agreement or no resolution (coded as “1”), and cases that were rejected or blocked (coded as “0”). As Table 2 reports, 105 specific instances were accepted by the NCP, whereas 128 were not. 3.2.2 Independent variable: Extraterritoriality The key independent variable is extraterritoriality and captures whether a complaint is received by the same country in which the alleged violation took place. It is a binary variable using data from the OECD Watch complaints database. The cases in the dataset are assigned a value of 0 if the host country and the NCP country are identical, and a value of 1 if they are different. 1 It is important to note that an agreement alone does not always determine the success of a case. Even after all parties agree that a breach of the Guidelines had occurred, an actual improvement for the victims is far from guaranteed. Daniel et al. (2015) find that only one percent of complaints between 2000 and 2015 led to concrete change for the affected communities and individuals, and financial compensation was never given. Conversely, cases that do not end in an agreement can still have a positive impact on the affected communities. This is especially the case with NCPs that determine whether there had been compliance or non-compliance with the Guidelines after the specific instance is concluded (OECD Watch policy advisor). Such a determination can damage the reputation of a company and even lead to consequences from the government such as limited access to economic benefits (Achtouk-Spivak & Garden, 2022). 16 Generally, the NCPs of the host country and of the company’s home country are eligible to accept a complaint, meaning that if the case is extraterritorial, the NCP of the home country will be responsible. The practice of filing complaints to the NCP of another country is very common: The dataset created for this analysis contains 178 extraterritorial cases compared to 55 cases in the same country (Table 3). It is noteworthy that most of the latter cases (31 out of 55) were filed to NCPs outside of Europe. Interestingly, none of the interviewees reported having observed an effect of extraterritoriality on the specific instances, partially because they had been involved in few cases that were not extraterritorial. Importantly, this variable does not include the nationality of the parties involved. A complaint may be filed by an affected group or community in the host country, or by other representatives such as an international NGO. While national and cultural differences are more likely to occur in extraterritorial cases, exceptions are common. In the case Milieudefensie et al vs. ING that will be analysed in a later section, a Dutch NGO took the lead in filing a complaint on behalf of local communities negatively affected by palm oil production, together with the Indonesian and Liberian chapters of Friends of the Earth. Therefore, representatives from the host country, the home country and a third country participated in the specific instance on the side of the complainants. Since there is no information available on the individuals that are present in the mediation, it is not possible to determine, for instance, which languages were spoken by the participants. 3.2.3 Control variable: NCP quality The first control variable is the quality of the NCP. Despite the principle of functional equivalence set in the Guidelines, NCPs differ substantially in both structure and practice (Buhmann, 2019; Sanchez, 2015). This diversity of NCPs has been shown to have implications for the outcome of specific instances. In the past, complaints that were rejected by one NCP have sometimes been successful in another country (Černič, 2021). In addition, even when a case does not lead to an agreement, some NCPs give recommendations to the parties or make a public determination of whether the corporation had breached the Guidelines, which can be “very important for the complainant, because it's really a big push for an agreement to be reached” (OECD Watch policy advisor). Buhmann warns that the institutional diversity needs to be considered when assessing the strengths and weaknesses of the NCP system as a whole (2019). The quality and reputation of an NCP may also affect extraterritoriality because civil society occasionally selects the NCP that is most expected to produce a beneficial outcome 17 (OECD Watch policy advisor). Although the Guidelines state that the parties should file their complaints to the host or home country, it is possible to circumvent this rule in practice, for example when the company has a subsidiary in another country (OECD Watch intern). Accordingly, NCPs that are considered stronger and more independent may receive a higher number of specific instances from other countries. The data for this variable is obtained from evaluations of each NCP by OECD Watch. The NCPs are evaluated on 40 indicators in the categories of procedures, organisation and communication. Theoretically, the NCP quality can assume values ranging from 0 to 40, with one point given for each criterion the NCP fulfils. Following this distribution, the NCPs in the dataset receive scores between 10 and 35 (Figure 1, Table 4). 3.2.4 Control variable: Host OECD membership While researchers have addressed the differences between the NCPs, little work has been done on the role of the host country. As the location of the allegations, the host country is the starting point for complaints and, thus, a relevant factor in the development of the specific instance. In most cases, nationals of the host country are involved in filing the complaint and in the mediation (OECD Watch, 2022). Therefore, what is the host country and whether it is an OECD member or not, provides some information relevant to the outcome of the case. First, it may affect the background of the complainants. The interview with an OECD Watch policy advisor revealed the importance of expertise for the success of a specific instance, and that “very, very rarely will a local community have the expertise to speak to the complexities of the OECD guidelines”, although it is possible to mitigate this problem by consulting a larger and more experienced NGO. In this regard, the OECD membership could have an impact on the expertise of the affected parties. While it can be assumed that the Guidelines are better known in the countries that adhere to them, research on the awareness of the Guidelines in different countries is lacking. Furthermore, there is a severe discrepancy between the resources distributed to civil society in OECD and non-OECD countries (Piewitt, 2010). Even though non-judicial mechanisms are often considered cheaper alternatives to legal action, sufficient resources are important for a party to be able to engage in the lengthy proceedings. Furthermore, larger NGOs with more resources are able to exert additional pressure, in particular through media reports, which often increases the incentives for a company to participate in the mediation (OECD Watch policy advisor). Second, the link to extraterritoriality is evident: Since only few non-OECD countries adhere to the Guidelines and have their own NCPs, parties from these countries are forced to 18 take their complaints abroad. Notable exceptions in the dataset are Argentina, Brazil and Peru, which are not in the OECD and have primarily accepted cases located within their own territories. The variable hostOECD is binary and captures whether the host country in a case is a member of the OECD. The data on the host country is obtained from the complaints database by OECD Watch and cross-checked with the list of OECD members. If multiple host countries were involved in the case, the value is chosen according to the primary host country, as identified by OECD Watch. Of the 233 cases in the dataset, 54 are located in OECD countries compared to 179 cases in non-OECD countries (Table 5). 3.2.5 Control variable: Year dummies, and pre-2011 and 2011-2022 periods In order to control whether the time of the specific instance has any effect on extraterritoriality and the outcome of the case, the analysis includes year dummies. The variable taken from the OECD Watch complaints database measures the year in which a complaint was filed. As Figure 2 shows, the number of cases per year differs substantially. It is important to note that the dataset only contains cases that have been concluded and that, given the long duration of the process, most cases from 2021 and 2022 are still ongoing at the time of writing, partially explaining the drop in recent years. Similarly, Figure 3 shows the share of all cases resulting in an agreement since 2000. Considering the wide range of these values, it seems sensible to control for effects related to the year in which the complaint was filed. Moreover, all interviewees noted a qualitative change in NCP cases throughout the years, specifically dealing with more issues relating to labour, human rights and due diligence. Both OECD Watch and the Dutch NCP criticised an increasingly legalistic approach which involves lawyers on both sides of the negotiations instead of regular NGO members and the company management or human rights department. The variable is similarly important to control for any changes related to the update of the OECD Guidelines in 2011. It is likely, for instance, that the new expectation towards companies to assume responsibility for their supply chains and other business relationships opened the door to more extraterritorial complaints. In addition to the year dummies, the impact of the 2011 update is estimated in a separate analysis, which compares the relationship between the key variables in the period until 2011 to the period after 2011. To control for the changes made in 2011, the variable reform was created (Table 6). It is coded as 0 for cases until 2011 (98 cases in total), and 1 for cases after 2011 (135 cases in total). 19 3.3 Results After introducing the dependent and independent variables, this section presents the results of the regression analysis. Table 1 displays the results of the logistic regression for four models. Model 1 reports a bivariate regression coefficient for the main independent variable of extraterritoriality (N=233). The coefficient is statistically significant at a 95% level (p<0.05). The sign of the coefficient indicates a negative relationship, suggesting that the probability of an agreement decreases when the case is extraterritorial. Model 1 provides initial support for 𝐻1 and for the rejection of 𝐻0. Model 2 introduces the first control variable, the quality of the NCP. When adding this variable to the logistic regression, both coefficients enter statistically significant at p<0.01. The coefficient for the quality of an NCP is positively signed, suggesting that an NCP of higher quality is more likely to produce an agreement between the parties. In addition to the previous two variables, Model 3 controls for the OECD membership of the host country. Since this variable enters not statistically significant, the outcome of a case does not seem to be affected by whether the host country is a member of the OECD. The final model adds the year in which the complaint was filed to the regression to control for potential changes over time. Importantly, the introduction of this variable reduces the size of the sample from 233 to 147. This decrease can be attributed to the lack of agreements in cases from ten years, mainly in the first decade of the NCP complaint system. Observations from these years were dropped by Stata because of the lack of variation in the dependent variable. In Model 4, the coefficients for extraterritoriality and the NCP quality remain statistically significant, while the coefficient for OECD membership remains statistically not significant. According to these results, specific instances that are located within the same country and received by a strong NCP can be expected to have a higher rate of success. 20 Table 1. Logistic regression. Model 1 Model 2 Model 3 Model 4 Agreement Agreement Agreement Agreement Extraterritoriality -0.786** -1.152*** -1.032* -1.309** (0.39) (0.42) (0.59) (0.66) NCP evaluation score 0.105*** 0.103*** 0.123*** (0.03) (0.03) (0.04) OECD member host 0.166 0.039 (0.57) (0.62) Year dummies no no no yes Constant -1.173*** -3.745*** -3.820*** -2.370 (0.32) (0.90) (0.94) (1.63) N 233 233 233 147 Note: Standard errors in parentheses ***p <0.01, **p<0.05, *p <0.10. The sign of the coefficients indicates the direction of the relationship between the variables but says little about the size of the effect of the independent variables on the dependent variable. The interpretation of the magnitude of the coefficients can be made more intuitive by transforming the values from logit to predicted probabilities and by looking at the average marginal effects. Average marginal effects indicate the average change in the predicted probability of the dependent variable associated with a one-unit change in an independent variable across all observations (Long & Freese, 2001). Accordingly, a one-unit change in extra – from no extraterritoriality to extraterritoriality – is associated with a 0.198 decrease in the predicted probability of an agreement (Table 2). Similarly, a change in NCPevl of one point increases the predicted probability by 0.019. All other variables being held equal, the predicted probability for an agreement in a specific instance of the lowest-scoring NCP, New Zealand, is 0.19, while the probability for the highest-scoring NCP, Norway, is 0.67. 21 Table 2. Average marginal effects. Average marginal effects Number of obs = 147 Model VCE: OIM Expression: Pr(agree), predict() Delta-method dy/dx std. err. z. P>z [95% conf. interval] extra -0.198 0.096 -2.060 0.040 -0.387 -0.009 NCPevl 0.019 0.005 3.620 0.000 0.009 0.029 After having found a relationship between extraterritoriality and the predicted probability of an agreement, it is still not clear whether there are fewer agreements because extraterritorial complaints are more likely to be rejected by NCPs, or whether the dialogue fails at a later point, for instance because the company is not willing to participate. Therefore, the analysis in Table 3 replaces the dependent variable agree with accept to identify whether extraterritoriality affects the case assessments by NCPs. As the coefficient of extra is not statistically significant, such a relationship cannot be detected. The coefficient of NCPevl remains significant, indicating that NCPs of a higher quality are less likely to reject or block specific instances. One case from the year 2022 was dropped by Stata, reducing the number of observations to 232. The results of this regression analysis indicate that extraterritoriality does not affect the decision of an NCP to accept or reject a case, but it does affect the mediation process once a complaint is accepted. Table 3. Logistic regression using accept. Model 1 Accept Extraterritoriality -0.646 (0.46) NCP evaluation score 0.048** (0.02) hostOECD -0.343 (0.46) Year dummies yes 22 Constant 0.245 (1.36) N 232 Note: Standard errors in parentheses ***p <0.01, **p<0.05, *p <0.10. In addition to controlling for the year of the complaint, the quantitative analysis aims to estimate the impact of the 2011 Update of the OECD Guidelines. First, another regression analysis is conducted with a model which replaces the year dummies with the binary reform variable. As Table 4 reports, the coefficient of this variable is not statistically significant, but it affects the coefficient of extraterritoriality which is also no longer significant. Table 4. Logistic regression using reform. Model 1 Agreement Extraterritoriality -0.971 (0.60) NCP evaluation score 0.105*** (0.03) hostOECD 0.164 (0.58) reform 0.528 (0.41) Constant -4.250*** (1.02) N 233 Note: Standard errors in parentheses ***p <0.01, **p<0.05, *p <0.10. 23 To examine the effect of the update in more detail, the regression analysis is separated into the periods of 2001 to 2011, and 2012 to 2022, as 2012 was the first year in which the new Guidelines were implemented. After dividing the observations according to the time periods, the sample consists of 98 observations from 2001 to 2011, and 135 observations from 2012 to 2022. To avoid a further decrease of the already small number of observations in both periods, the year dummies are excluded from the regression. The likelihood-ratio test in the following section shows that the omission of the year variable does not lead to an omitted-variable bias. Furthermore, the variable of OECD membership is excluded for reasons of efficiency. Table 5 reports the results for both periods. In Model 1, neither of the coefficients are statistically significant, but in Model 2, extra and NCPevl are significant at a 95% and 99% level, respectively. These results indicate that the relationship between the key variables of interest can only be observed for cases after 2011. This limitation needs to be considered for further analyses of causal mechanisms. Nevertheless, it is important to be cautious with claims of causality. Despite having found no significant impact of the year in which a case was filed, other factors besides the change in the Guidelines could have influenced the outcome of specific instances. Table 5. Logistic regression of the before the reform and after reform periods. Model 1 (2001-2011) Model 2 (2012-2022) Agreement Agreement Extraterritoriality -0.913 -1.150** (0.79) (0.51) NCP evaluation score 0.077 0.125*** (0.05) (0.04) Constant -3.438** -4.094*** (1.42) (1.19) N 98 135 Note: Standard errors in parentheses ***p <0.01, **p<0.05, *p <0.10. 24 After reviewing the results of the quantitative analysis, 𝐻1 finds support in the data. Extraterritorial cases are more likely to end in no agreement, but this effect seems to hold only for the 2012-2022 period after the update of the OECD Guidelines. Another relevant aspect is the quality of NCPs, as stronger NCPs are more likely to achieve an agreement between the parties. 3.4 Regression diagnostics and robustness check Similar to OLS regression, logistic regression requires certain assumptions to be met to ensure fitting and unbiased maximum likelihood estimates. First, the logit of the dependent variable must be a linear function of the independent variables if the chosen model is properly specified (Mehmetoglu & Jakobsen, 2022, p. 184). Although the model specification is primarily based on theory, potential issues can also be detected by conducting a link test. A correctly specified model is indicated by a significant _hat variable, while the _hatsquared variable should not be significant. As Table 7 shows, both conditions apply at p-values of 0.010 and 0.710, respectively. These results suggest that the analysis can be continued with the current model. It is also possible to test the goodness of fit by comparing the predicted values of a model to the observed values. For this purpose, a probability of more than 0.5 can be considered a prediction that a case will end in an agreement, whereas a lower probability predicts that it will not. The model predicted 11 cases to end in an agreement and 136 cases not to end in an agreement (Table 8). Of the 11 cases predicted to be successful, 7 were correctly classified and 4 were incorrectly classified. Of the 136 cases predicted to be unsuccessful, 108 cases were correctly classified, and 28 cases were incorrectly classified, resulting in a final percentage of 78.23% correct classifications. Despite this relatively high rate of correct predictions, the low value of sensitivity towards successful cases has to be taken into account when assessing the goodness of fit of the model. The second assumption of logistic regression addresses possible biases caused by the inclusion or omission of variables (Mehmetoglu & Jakobsen, 2022). Omitted-variable bias occurs when a relevant variable is missing from the regression, which can distort the effect another independent variable has on the dependent variable (Kellstedt & Whitten, 2018). The control variables above were selected to avoid this bias and potential limitations were discussed in the theoretical section on the variables. Instead of the chosen independent variables relating to the NCP and the host country, it was also considered to treat these country variables as dummy variables and include broader country-fixed effects in the analysis. This approach 25 would have the advantage of controlling for other factors relating to the countries that were not included in the regression above. In practice, however, the lack of variation in many countries would have caused most observations to be dropped and reduced the sample to n=41. As a consequence, more specific variables were chosen in relation to the NCP and the host country, at the risk of omitting other country-related factors. Additional variables that were considered but not included due to their complexity were the sectors and the chapters of the Guidelines referred to in a specific instance. Some sectors and types of corporate conduct are associated with a higher rate of extraterritoriality; for example, labour rights violations in the textiles industry occur most frequently in non-OECD countries without own NCPs. Since most complaints refer to multiple chapters, and many of the multinational corporations operate across different sectors, the complexity of identifying such overlaps and their relationships with the main variables exceeds the objectives of this thesis. Similarly, the analysis should not include irrelevant variables for reasons of efficiency (Mehmetoglu & Jakobsen, 2022). To test whether a variable is relevant to the analysis, a likelihood-ratio test is conducted. The test compares the full model to another model that excludes one of the variables to test whether this constraint reduces the log likelihood of the model (Long & Freese, 2001). If there is no significant change, the exclusion of the variable does not produce an omitted-variable bias. Table 9 shows the results for a likelihood-ratio test excluding each of the three control variables. The χ2 value is only statistically significant in the first test, meaning that the model controlling for NCP quality is better suited to explain the observed values than a model excluding this variable. Since the tests for the other models are not significant at p<0.05, the control variables of OECD membership and year do not improve the performance of the model and should therefore be excluded to maximise efficiency. The third assumption of logistic regression states that the observations need to be independent from each other (Mehmetoglu & Jakobsen, 2022). Although the specific instances took place at different points over more than two decades, they do not directly affect each other’s outcome and are therefore independent. The comparative case study in the qualitative part of this thesis further illustrates that even similar cases can lead to vastly different results. To avoid additional overlaps, complaints that are filed to multiple companies but are otherwise identical are counted as only one case. The fourth assumption of logistic regression concerns multicollinearity. To test whether the independent variables are correlated with each other, Table 10 shows the VIF statistics for 26 extra, NCPevl and hostOECD. Since the VIF for all variables is close to 1, multicollinearity is not an issue. Another potential problem in logistic regression concerns influential observations which can produce misleading results. Such observations can be identified in logistic regression by using Pregibon’s (1981) dbeta. A scatterplot of dbeta shows twenty clear outliers (Figure 5). When excluding these twenty cases from the regression, the coefficients increase slightly (from -1.152 to -1.014 for extraterritoriality and from 0.105 to 0.119 for NCP quality). Both coefficients remain statistically significant (Table 11). Accordingly, the influence of the outliers on the results seems to be small and it is possible to proceed without further consideration. 4 Qualitative analysis 4.1 Methodology and data After the quantitative analysis has supported the hypothesis that extraterritoriality has a negative impact on the outcome of specific instances, the following analysis aims to identify potential causal mechanisms to explain this relationship by conducting a comparative case study. The case study follows the most similar systems design. In this design, two cases share every factor, except for one variable, but produce different results. According to Mill (1843), the developer of this method, it is then possible to infer that the differing independent variable has an effect on the dependent variable. This design has the advantage of being applicable to small samples and limiting the number of possible confounders (Steinmetz, 2019). While no NCP cases are completely identical, some overlap considerably and are therefore suitable for the most similar systems design. The cases chosen for this analysis are Dutch NGOs vs. ING Bank from 2017 and Milieudefensie et al. vs. ING from 2019. The cases are similar in nature and identical on the side of the defendant and the NCP but differ in the aspect of extraterritoriality (Table 6). The Dutch NGO Milieudefensie participated in both specific instances together with other NGOs and alleged in their complaint that ING, a bank headquartered in the Netherlands, contributed to the environmental harm caused by its clients. Both complaints were received by the Dutch NCP without the involvement of other NCPs. The cases differ in the main explanatory variable – extraterritoriality – and in their outcome: The allegations in Dutch NGOs vs. ING Bank concerned operations within the Netherlands, and in Milieudefensie et al. vs. ING, the alleged harm took place in several African and Asian countries. The case Dutch NGOs vs. ING Bank resulted in an agreement, whereas the other case did not. In this sense, the cases reflect the main hypothesis that extraterritorial cases are less likely to result in an agreement. Therefore, the 27 comparative case study can contribute to exploring the role of extraterritoriality in the success of a specific instance and help identify possible causal mechanisms to substantiate this relationship. Table 6. Overview of cases. Case Dutch NGOs vs. ING Bank Milieudefensie et al. vs. ING Date 8 May 2017 5 July 2019 NCP The Netherlands The Netherlands Complainants BankTrack, Greenpeace Netherlands, Milieudefensie, SDI, WALHI Milieudefensie, Oxfam Novib Defendant ING ING Extraterritoriality No Yes Outcome Agreement No agreement Before proceeding with a more detailed analysis, it is necessary to consider other potential differences between the cases. Apart from Milieudefensie, the side of the complainants consisted of different NGOs: The Dutch NGOs BankTrack, Greenpeace Netherlands and Oxfam Novib participated in 2017, whereas WALHI from Indonesia and SDI from Liberia joined in 2019. The extent of each NGO’s individual contribution is not publicly traceable, and the repeated participation of an organisation or corporation does not necessarily mean that the same individuals are part of the process (Dutch NCP secretariat member #1). Nonetheless, the central role of Milieudefensie in at least the second specific instance indicates that there were enough similarities on the side of the complainants to warrant a comparison. It is also important to consider qualitative differences between the cases, in particular the burden of proof. Although NCPs do not officially demand evidence in a legal sense, their disproportionately high standards for accepting a specific instance have been criticised in the past (Daniel et al., 2015). In contrast to the second case, which dealt with human rights and labour rights violations at palm oil plantations next to environmental issues, the first case does not concern a specific group of victims but addresses ING’s impact on the global climate, in particular its reporting practices. In such a case, it can be argued that it is “easier to provide the evidence, because it's literally in their financial statements” (OECD Watch policy advisor). A member of the Dutch NCP secretariat responded that this specific instance was challenging for other reasons, in particular disagreements about technical issues relating to different methodologies and approaches to the problem. Considering that the Dutch NCP accepted both 28 complaints without demanding further substantiation, the burden of proof does not seem to be a central aspect in the two cases. It is possible, however, that such a difference between extraterritorial and non-extraterritorial complaints exists in other cases. The time span of two years between the two complaints is relatively short given that the work of the NCPs has covered more than two decades. Nevertheless, there is a noticeable difference in the rate of success between the years: Four out of eight NCP complaints filed in 2017 ended in an agreement, which is by far the highest rate of agreements in a year since the establishment of the NCP system. In 2019, the number of complaints doubled to 16, but not a single case ended in an agreement. The reason for this high fluctuation is not evident and requires further research. A possible explanation could be the onset of the COVID-19 pandemic. The dialogue meetings in the first case were held between February 2018 and January 2019, whereas the meetings in the second case took place between February 2020 and July 2021 (Dutch NCP, 2019; Dutch NCP, 2022). Starting in March 2020, the Dutch Government imposed contact restrictions in response to the spread of SARS-CoV-2. The restrictions also affected the work of the NCP which reported having to hold meetings online that would normally have been conducted in person (Dutch NCP secretariat member #2). The implications of this shift for the outcome of the dialogue is not clear, although it could be that the circumstances made it more difficult for the parties to build mutual trust or that companies were less interested in participating in the dialogue while facing economic struggles caused by the pandemic. After considering the potential limitations to this research design, the following section will describe the timeline of each specific instance in more detail before proceeding with the analysis. The primary sources are official statements and press releases from the Dutch NCP, the participating NGOs and ING. Further insights were taken from the interviews with representatives of the Dutch NCP and OECD Watch. Unfortunately, the representatives of Milieudefensie and WALHI were not available for an interview. The following sections describe the timelines of each case before proceeding with the analysis. 4.2 Case 1: Dutch NGOs vs. ING Bank On 8 May 2017, the Dutch NGOs BankTrack, Greenpeace Netherlands, Milieudefensie and Oxfam Novib filed a complaint against ING to the Dutch NCP under the specific instance procedure. The complaint alleged that ING had violated the OECD Guidelines, specifically relating to Chapter III (Disclosure), Chapter VI (Environment) and Chapter VIII (Consumer Interests), by failing to disclose its direct and indirect greenhouse gas emissions, including 29 emissions resulting from loans and investments (Oxfam et al., 2017). In addition, the NGOs requested the bank to publish emission reduction targets for its loans that align with the Paris Agreement. ING responded to the complaint that it has been publishing and setting targets for its own emissions but was unable to collect data on the emissions of clients and, therefore, enforce such targets on clients (2017). On 14 November 2017, the Dutch NCP accepted the case for further examination and invited the NGOs and ING to participate in a dialogue, which all parties agreed to. The parties came together in four dialogue sessions and two expert meetings between February 2018 and January 2019 (Dutch NCP, 2019). A central part of the dialogue revolved around the question of which methodology would be most appropriate to measure indirect emissions caused by financing activities. After considering different technologies, ING developed the Terra approach which measures and sets targets for carbon emissions, using the Paris Agreement Capital Transition Assessment (PACTA) methodology for oil and gas sectors, the Platform Carbon Accounting Financials (PCAF) methodology for real estate, the Poseidon Principles for shipping and the Sustainable STEEL Principles for steel (ING, 2022). The bank also updated its previous targets of limiting global warming to well below 2°C to net-zero targets in most sectors and announced intermediary targets (ING, 2022). In a final statement published in May 2019, the NCP summarised the agreement between the parties on the adoption of the Terra approach and the setting of intermediary targets. The bank and the NGOs jointly call on the Dutch Government to request the development of two scenarios by the International Energy Agency to reduce global warming to 1.5°C at a 66% chance. In light of the productive dialogue, the NCP does not make a determination regarding ING’s compliance or non-compliance with the Guidelines. Nevertheless, the final statement reiterates the requirements of the Guidelines regarding the measurement of greenhouse gas emissions and target setting while acknowledging that the appropriate methodology is still evolving. The NGOs welcome this development in a press release, but call upon ING to phase out all investments in the oil and gas sector following the bank’s announcement to reduce the financing of thermal coal to close to zero by 2025 (Oxfam Novib, 2019). One and a half years after the official conclusion of the case, the Dutch NCP published an evaluation after holding another meeting with the parties. The NCP positively highlighted the efforts made by ING to measure and decrease their climate impact, and encouraged the bank to take further steps to reduce emissions caused by fossil fuels. 30 4.3 Case 2: Milieudefensie et al. vs. ING On 5 July 2019, Milieudefensie filed a new complaint against ING with the Dutch NCP, joined by two other NGOs. All three parties, Milieudefensie from the Netherlands, Wahana Lingkungan Hidup Indonesia (WALHI) from Indonesia and Sustainable Development Institute (SDI) from Liberia, belong to the Friends of the Earth International network. The complainants argued that ING had breached the Guidelines by contributing to harm caused by its clients involved in palm oil production in multiple countries, including Indonesia, Cameroon, Sierra Leone and Liberia (Milieudefensie et al., 2019). The renewal of loans to corporations causing human rights, environmental and labour rights violations was considered a failure of ING to fulfil its due diligence obligations under the Guidelines. The clients in question were Noble Group Ltd., Bolloré Group/Socfin Group S.A., and Wilmar International Ltd. and their subsidiaries. The NGOs argued that the bank had been made aware of these practices by, among others, a previous complaint filed against Socfin and Bolloré to the French and Belgian NCPs (Milieudefensie et al., 2019). In response to the complaint, ING (2019) acknowledged these issues but explained its approach of collaborating with the businesses to improve their operations instead of disengaging. The bank also pointed out the commitment of all of their clients to be certified as sustainable by the Roundtable on Sustainable Palm Oil (RSPO), an initiative promoting sustainable palm oil production. On 20 January 2020, the Dutch NCP published an initial assessment of the complaint, stating that it met the necessary criteria and offering its good offices for a dialogue between the parties. The dialogue sessions focused on ING’s due diligence policies, its involvement with the conduct of the companies and its responsibilities with regard to remedy. A central question was whether ING was “linked to” or “contributed to” the adverse impacts caused by its clients, according to the Guidelines (Dutch NCP, 2022). After multiple session and plans to proceed further, ING withdrew from the mediation. The reason given for this decision was a lack of trust in the willingness of Milieudefensie to reach a joint agreement after the NGO had published a report criticising the certification process of RSPO (Dutch NCP, 2022). While this report did not mention ING, Dutch newspaper de Volkskrant released an article shortly after in which Milieudefensie was quoted criticising the reliance of financial institutions, including ING, on RSPO’s certifications (Vos, 2021). According to ING, these publications showed that Milieudefensie “had already drawn its conclusions” (Dutch NCP, 2022, p. 7) about the issues in question and was no longer interested in continuing the open dialogue. Milieudefensie apologised for not informing ING and the Dutch NCP about the publications in advance but 31 stated that it had not breached confidentiality and was allowed by the OECD Guidelines to continue campaigning during the dialogues (Dutch NCP, 2022). Despite Milieudefensie’s assurance of participating in good faith, further communication between the parties was not successful in resolving these disagreements and the dialogue ended in August 2021. In a final statement published in April 2022, the Dutch NCP expressed regret about the premature ending of the dialogue sessions and encouraged the parties to continue discussing the questions raised in the dialogue. The conclusion of the statement clarified that banks are more often linked to adverse impacts but may in some instances contribute to them if the financed misconduct was foreseeable and the bank failed to carry out proper due diligence. Since the dialogue ended before ING’s role in the instances submitted by the complainants was discussed, the NCP did not continue examining these questions in detail. Nonetheless, the NCP stated that banks and other enterprises relying on initiatives such as RSPO should conduct periodic reviews to assess whether they met their due diligence obligations when cooperating with these initiatives. 4.4 Analysis After outlining the process of the two cases, this section analyses what factors have played a role in their respective outcomes and whether the variable of extraterritoriality is able to explain the different results. The qualitative analysis draws on statements made by the NGOs, bank and NCP involved, together with information obtained from the interviews. It examines the primary sources in relation to the causal mechanisms introduced in the theoretical section, namely trust, power and culture. When appropriate, additional information beyond the case study is considered in order to include factors that could explain the relationship detected in the quantitative part, even if they were not present in the specific cases selected. 4.4.1 Trust A comparison of the cases illustrates the importance of trust between the parties. After the case Dutch NGOs vs. ING Bank was concluded, all parties emphasised the constructive nature of the dialogue and the fundamental need to act against climate change. Despite further criticisms of ING’s financing of fossil fuels, the NGOs appeared pleased with the NCP’s work and the progress that had been made by ING (Oxfam Novib, 2019a). The head of business ethics at ING, Arnaud Cohen Stuart, expressed similar appreciation for the NCP’s role in the dialogue and the cooperation of the NGOs (ING, 2019). Both the final statement and the subsequent evaluation by the NCP reflect the mutual respect between the parties. With regard to the future, 32 the NCP expressed “confidence in the continuation of the constructive dialogue between the parties, after the closing of the procedure” (2019, p. 6). This positive assessment raises further questions about the failure of the second mediation. Apart from the Dutch NCP, neither of the parties publicly addressed the conflict ending the mediation. On the day of the NCP’s final statement, Milieudefensie published a statement criticising ING’s withdrawal, supported by statements from WALHI and SDI. The NGOs accuse the bank of running “away from its responsibilities” (Milieudefensie, 2022, p. 1) and continuously profiting from the adverse impacts caused by its clients in the palm oil industry. Further insights gained from the interviews added context to the publications by Milieudefensie and ING’s subsequent decision to end the dialogue. The Dutch NCP clarified that the event did not constitute a breach of confidentiality in a strict sense, but rather a breach of trust between the parties (secretariat member #2). While OECD Watch encourages complainants to continue campaigning before and during the dialogue, if permitted by the NCP’s rules of procedure, the policy advisor acknowledged that caution is advised with regard to the media, and that press statements are ideally agreed upon with the NCP and the other parties before publication. After ING and Milieudefensie had worked together successfully in the specific instance of 2017, their cooperation ended with ING’s lack of trust in Milieudefensie’s intentions. During both interviews conducted for this thesis, the importance of trust was repeatedly highlighted, both with regard to the NCP, and between the complainants and defendants. The members of the Dutch NCP explained that “we really take the time and put lots of efforts in building that trust so that the parties trust the NCP, that they trust the process, and we also try to build trust between the parties” (secretariat member #1). Similarly, the policy advisor at OECD Watch argued that trust in the independence of NCPs and their ability to produce a beneficial outcome for the complainants was essential for a more effective NCP system. This statement is in line with the results of the quantitative analysis, which found that the quality of the NCP was an influential variable. Considering that ING and the NGOs expressed appreciation for the NCP in the first case and did not openly voice criticism in the second case, trust in the NCP might help to explain the success of the first case but not the failure of the second. As the selected cases show, the challenges of an NCP lie not only in building trust between the parties but also maintaining it throughout the negotiation. While extraterritoriality may not be inherently linked to a lower level in trust, it can have a negative effect when interacting with the other mechanisms explored below. 33 4.4.2 Power After comparing the levels of trust in the two cases, this section discusses the role of actual or perceived power dynamics. According to the literature, distrust between the participants is more likely to arise when one party has considerably more power than the other. A similar point was made in the interview when the members of the NCP secretariat pointed out the need to consider an imbalance in power, in particular between large corporations, on one side, and individuals or communities, on the other. Power can take several forms which differ in their expression and measurement. Political power can affect the proceedings, especially when the NCPs are closely linked to the governments. Pressure can be exerted from companies and governments, and in some countries, attempts to bribe or threaten the complainants were reported (OECD Watch policy advisor). However, no indications were found in the research that political pressure played a role in either of the ING cases. A second form of power is linked to financial resources, not just for the participation in the NCP process but also for separate media strategies, meaning that a comparison can give an indication of the power dynamics in the process. For an estimation of each side’s resources, ING’s provisions for litigation are compared to the budget of the NGOs. The information is obtained from financial reports for the year 2019, since the timeline of both cases overlapped in this year. The total expenditures of the NGOs are chosen instead of a more precise measurement such as the expenditure on climate-related campaigns to allow for a comparison because the financial statements differ in their structures, and it is not always clear how costs resulting from NCP cases are categorised. In 2019, the total budget of the NGOs involved in the first case amounted to approximately €230,000,000 (BankTrack: €478,623; Greenpeace Nederland: €22,859,000; Oxfam Novib: €191,975,000; Milieudefensie: €18,021,999). In the same year, ING’s provisions for litigation amounted to €102,000,000. The magnitude of these numbers suggests that both parties had sufficient resources to participate in the dialogue over a longer period of time, and to acquire additional expertise if necessary. The total NGO budget in the second case is more difficult to determine, as SDI and WALHI did not make their financial reports publicly available. Although both are established NGOs, their financial resources are unlikely to be of the size of Greenpeace or Oxfam – the complaints of the first case. Although the budget of Milieudefensie is relatively high, the combined budget of the NGOs in the second case is likely to be smaller than the budget of the NGOs in the first case, which could have resulted in the perception of a being at a disadvantage compared to ING. It is possible that Milieudefensie’s decision to follow a parallel strategy in the media resulted 34 partially from a lack of confidence that the mediation would result in an agreement. Moreover, ING was able to withdraw from the dialogue without fearing the amount of negative publicity that could have been produced by the coalition from the first case. Nevertheless, it is important to note that it is difficult to evaluate how the parties perceived their relationship and their own power without speaking to them, hence the results of the analysis of the power mechanism have to be taken with a “pinch of salt”. 4.4.3 Culture and language The final aspect considered in the analysis is the impact of language and cultural differences on the mediation. The main assumption here is that it is more difficult to build trust between parties from different cultural backgrounds and languages. Considering that the main conflict in the second case arose between the Dutch bank ING and the Dutch NGO Milieudefensie, instead of the other complainants from Indonesia and Liberia, and partially concerned a report by a Dutch newspaper, language and cultural differences do not seem to play a substantial role in the outcome of the second case. On the contrary, it is questionable whether the involvement of an Indonesian or Liberian newspaper would have had a similar effect considering ING’s primarily European customer base. While such assumptions require further substantiation, it is conceivable that problems can arise when complainants and defendants operate in the same territory. However, the interviewees at the Dutch NCP confirmed that they take the cultural background of the participants into account when offering their good offices, and they have noticed different approaches to the negotiations in the past, in “the way people react and the information they give and what is important to them” (secretariat member #1). They added that external circumstances, such as an armed conflict in the host country, further complicate the mediation. The assumption that the use of communications technology required by geographic distance explains the lower rate of success in extraterritorial cases seems to be unsubstantiated, as the members of the NCP secretariat stated that all mediation sessions before the COVID-19 pandemic had been conducted in person. Furthermore, they confirmed that language differences can be problematic, as the NCP does not have the resources to provide translations, and civil society organisations are not always able to do so. While factors relating to language and culture are less likely to have been influential in the cases selected above, they are nevertheless important to gain a broader understanding of the issue beyond the comparative case study, and to help explain the relationships detected in the quantitative part of the analysis. Given the limited explanatory power of individual cases, it is 35 not surprising that not all mechanisms were present in the case study. The additional information from the interviews suggests that other cases might have been more fitting to illustrate the role of language and culture, potentially at the cost of other aspects. Nevertheless, the data chosen for the case study reflects the importance of trust and, to a lesser degree, of a balance in power. To summarise, the conducted case study suggests that trust between the parties is an important factor leading to a successful outcome of the mediation process, although it may not be directly linked to extraterritoriality in the selected cases. While the budgetary differences between the parties hint at a change in the power dynamics that could potentially affect the level of trust, more evidence is needed to substantiate this claim. Finally, the case study did not find any indications of language and cultural differences playing an essential role in the process. 5 Conclusion The thesis set out to investigate the impact of extraterritoriality on the outcome of specific instances submitted to OECD National Contact Points by employing a mixed-method approach. In the quantitative analysis, it found that extraterritorial cases are less likely to result in an agreement than domestic cases, although this relationship was only found for the complaints filed after the 2011 update of the OECD Guidelines. No impact was found on the decision of NCPs to accept or reject a complaint, suggesting that extraterritoriality instead plays a role in the mediation process. The analysis also found that the quality of the NCP plays a substantial role in determining the outcome. Therefore, the problem of the low success rate is unlikely to be solved simply by setting up NCPs in more countries so that issues can be addressed domestically instead of abroad. Instead, NCPs require more resources and support by independent experts so that countries can fulfil the evaluation criteria set by OECD Watch. These findings were consolidated by the qualitative analysis: If not even the Dutch NCP, one of the strongest agencies in the field, has sufficient resources to provide translation services to overcome language barriers, it is likely that other NCPs are even less equipped to address such difficulties arising from cultural and linguistic diversity among the parties. The comparative case study illustrated the importance of mutual trust among the participants for the success of the mediation by comparing a case in which all parties appeared to be satisfied with the dialogue to a case where a lack of trust led to the withdrawal of the defendant. However, the case study also showed that the mechanisms connecting the main variables are less clear than initially expected, as not even the experts consulted for the thesis had considered extraterritoriality as an influential factor prior to the interviews. The analysis 36 indicated that an imbalance in financial resources complicates the building of trust and is thus connected to a lower chance of arriving at an agreement. Language and cultural differences did not seem to play an important role in the process but were confirmed to be considered by the Dutch NCP in other cases. More research is necessary to detect such patterns considering that “every case is so different” (Dutch NCP secretariat member #1). Given the confines of the thesis, there is a number of limitations to the findings. The first limitation is the exclusion of specific instances submitted by actors other than NGOs. In particular the complaints filed by trade unions could have given more insights into cases concerning labour rights, similar to the Heineken case that was outlined in the beginning. Secondly, it is important to consider that the Dutch NCP has a strong reputation and is not necessarily representative of all NCPs, some of which have yet to accept complaints at all. Since the analysis did not distinguish between different sectors and issues according to the OECD Guidelines, it is possible that not all factors linking extraterritoriality and the case outcomes were included in the analysis. Further research should cover a larger variety of cases, in particular with regard to the side of the complainants: Complaints filed by individuals or communities without the support of an internationally active NGO may face additional challenges that have not been addressed in this thesis. The thesis aims to contribute to present and future debates on the regulation of multinational corporations. Considering the legal difficulties of extraterritorial jurisdiction, non-judicial mechanisms such as the OECD National Contact Points provide an alternative to courts and will likely continue to do so in the future. However, the thesis has shown that the challenges of mediation involving parties from multiple countries have been underestimated in the past. To ensure the effectiveness of the NCPs, their weaknesses need to be identified and more work needs to be done to compensate for their lack of “bite” (Achtouk-Spivak & Garden, 2022, p. 608), so that complainants will be heard no matter their country of origin, and so that the success of the Heineken case does not remain the exception. The OECD and its member states can support this process by providing more resources to NCPs and an international fund for complainants. External actors like OECD Watch can help in establishing a more equal standing for complainants by offering assistance to affected communities and individuals. If NCPs are able to address these challenges and provide justice for domestic and foreign victims of corporate human rights abuse, they have the potential to become pioneers for future extraterritorial mechanisms in the field of business and human rights. 37 6 References Achtouk-Spivak, L., & Garden, R. (2022). OECD National Contact Point Specific Instances: When ‘Soft Law’ Bites? In Journal of International Dispute Settlement (Vol. 13, Issue 4, pp. 608–642). Oxford University Press (OUP). https://doi.org/10.1093/jnlids/idac017 Adair, W., Brett, J., Lempereur, A., Okumura, T., & al, e. (2004). Culture and negotiation strategy. Negotiation Journal, 20(1), 87-111. Retrieved from https://www.proquest.com/scholarly-journals/culture-negotiation- strategy/docview/205191318/se-2 Addis, A. (2012). Community and Jurisdictional Authority. In G. Handl, J. Zekoll, & P. Zumbansen (Eds.), Beyond Territoriality (pp. 11-33). Leiden, The Netherlands: Brill. Akehurst, M. (1972-1973). Jurisdiction in International Law. British Year Book of International Law, 46, 145-258. Augenstein, D., & Kinley, D. (2013). When human rights ‘responsibilities’ become ‘duties’: the extra-territorial obligations of states that bind corporations. In Human Rights Obligations of Business (pp. 271–294). Cambridge University Press. https://doi.org/10.1017/cbo9781139568333.017 Balaton-Chrimes, S., & Haines, F. (2016). Redress and Corporate Human Rights Harms: An Analysis of New Governance and the POSCO Odisha Project, Globalizations, 14(4), 596–610. Informa UK Limited. https://doi.org/10.1080/14747731.2016.1223958 BankTrack. (2019). Financial Report 2019. BankTrack. https://www.banktrack.org/download/banktrack_financial_report_2019/200414_financ ial_report_2019_2.pdf Bernaz, N. (2012). Enhancing Corporate Accountability for Human Rights Violations: Is Extraterritoriality the Magic Potion? Journal of Business Ethics, 117(3), 493–511. Springer Science and Business Media LLC. https://doi.org/10.1007/s10551-012-1531- z Bhatt, K., & Erdem Türkelli, G. (2021). OECD National Contact Points as Sites of Effective Remedy: New Expressions of the Role and Rule of Law within Market Globalization? Business and Human Rights Journal, 6(3), 423–448. Cambridge University Press (CUP). https://doi.org/10.1017/bhj.2021.30 Brett, J. M. (2000). Culture and negotiation. International journal of psychology, 35(2), 97- 104. 38 Buhmann, K. (2019). National Contact Points under OECD’s Guidelines for Multinational Enterprises: Institutional Diversity Affecting Assessments of the Delivery of Access to Remedy. In L. Enneking, I. Giesen, A-J. Schaap, C. Ryngaert, F. Kristen, & L. Roorda (Eds.), Accountability, International Business Operations and the Law: Providing Justice for Corporate Human Rights Violations in Global Value Chains (pp. 38-59). Routledge. Globalization: Law And Policy https://doi.org/10.4324/9781351127165-3 Bülow-Møller, A. M. (1993). Negotiating in a foreign language. HERMES-Journal of Language and Communication in Business, (11), 11-25. Catá Backer, L. (2012). Governance without Government: An Overview. In G. Handl, J. Zekoll, & P. Zumbansen (Eds.), Beyond Territoriality (pp. 87-123). Leiden, The Netherlands: Brill. Černič, J. L. (2008). Corporate responsibility for human rights: A critical analysis of the OECD guidelines for multinational enterprises. Hanse L. Rev., 4, 71. Cîrlig, R.E. (2016). Business and human rights: from soft law to hard law? Juridical Tribune, Bucharest Academy of Economic Studies, Law Department, 6(2), 228-246. Daniel, C., Wilde-Ramsing, J., Genovese, K., & Sandjojo, W. (2015). Remedy Remains Rare: An analysis of 15 years of NCP cases and their contribution to improve access to remedy for victims of corporate misconduct. OECD Watch. Davarnejad, L. (2011). In the shadow of soft law: the handling of corporate social responsibility disputes under the OECD guidelines for multinational enterprises. J. Disp. Resol., 351. De Schutter, O. (2015). Towards a New Treaty on Business and Human Rights. Business and Human Rights Journal, 1(1), 41–67. Cambridge University Press (CUP). https://doi.org/10.1017/bhj.2015.5 Dutch NCP. (2019, April 19). Final Statement Oxfam Novib, Greenpeace Netherlands, BankTrack and Friends of the Earth Netherlands (Milieudefensie) versus ING. Dutch National Contact Point. Dutch NCP. (2020, January 20). Initial Assessment Milieudefensie/Friends of the Earth Netherlands, WALHI/Friends of the Earth Indonesia and SDI/Friends of the Earth Liberia versus ING. Dutch National Contact Point. Dutch NCP. (2020, November 25). Evaluation of the NCP Final Statement dated 19 April 2019 regarding the notification Oxfam Novib, Greenpeace, BankTrack, Milieudefensie vs. ING. Dutch National Contact Point. 39 Dutch NCP. (2022, April 7). Final Statement Milieudefensie/Friends of the Earth Netherlands, WALHI/Friends of the Earth Indonesia and SDI/Friends of the Earth Liberia versus ING. Dutch National Contact Point. Ebner, N. (2017). Negotiation via videoconferencing. The Negotiator's Desk Reference. St Paul: DRI Press (2017). Emerson, R. (1962). Power-Dependence Relations. American Sociological Review, 27(1), 31- 41. Greenpeace Nederland. (2019). Jaarverslaag 2019. Greenpeace. https://www.greenpeace.org/static/planet4-netherlands-stateless/2020/06/dc552124- greenpeace_jaarverslag_2019.pdf Gerring, J., & Christenson, D. (2017). Applied Social Science Methodology. Cambridge University Press. Hackett, C., O'Kelly, C., & Patton, C. (2019). The Case of the Irish National Contact Point for the OECD Guidelines for Multinational Enterprises: Challenges and Opportunities for the Business and Human Rights Landscape in Ireland. Irish Jurist, 61, 99-121. Handl, G. (2012). Extra-territoriality and Transnational Legal Authority. In G. Handl, J. Zekoll, & P. Zumbansen (Eds.), Beyond Territoriality (pp. 1-9). Leiden, The Netherlands: Brill. ING. (2017, May 8). Reaction to National Contact Point complaint about ING by Greenpeace, BankTrack, Milieudefensie and Oxfam. Dutch National Contact Point. ING. (2019, July 5). ING reaction to NCP notification on palm oil. ING Bank. ING. (2022). 2022 Climate Report. ING Bank. https://www.ing.com/MediaEditPage/2022- Climate-Report.htm Kellstedt, P. M., & Whitten, G. D. (2018). The fundamentals of political science research (Third edition). Cambridge: Cambridge University Press. Khoury, S., & Whyte, D. (2019). Sidelining corporate human rights violations: The failure of the OECD's regulatory consensus. Journal of Human Rights, 18(4), 363-381. Lee, J., & Lee, J. (2022). Compatibility of the Incompatible: How Does Asymmetric Power Lead to Coexistence of Trust and Distrust in Adversarial Policy Networks? International Journal of Public Administration, pp. 1-16. Long, J. S., & Freese, J. (2006). Regression models for categorical dependent variables using Stata (Vol. 7). Stata press. 40 Macchi, C. (2017). The Role of the OECD National Contact Points in Improving Access to Justice for Victims of Human Rights Violations in the EU Member States. SSRN Electronic Journal. Elsevier BV. https://doi.org/10.2139/ssrn.3592947 Mehmetoglu, M., & Jakobsen, T. G. (2022). Applied statistics using Stata: a guide for the social sciences. Sage. Methven O’Brien, C. (2016). The Home State Duty to Regulate the Human Rights Impacts of TNCs Abroad: A Case of Extraterritorial Overreach? SSRN Electronic Journal. Elsevier BV. https://doi.org/10.2139/ssrn.2854275 Milieudefensie. (2019). Financieel verslag en jaarrekening. Milieudefensie. https://milieudefensie.nl/jaarverslag2019/ Milieudefensie, WALHI, & SDI. (2019, July 5). Complaint Milieudefensie v. ING Bank. Dutch National Contact Point. Milieudefensie. (2021, March 19). Press release: Palm Oil Certification: Not ‘Out of the Woods’. Milieudefensie. https://en.milieudefensie.nl/news/press-release-palm-oil- certification-not-out-of-the-woods Mill, J. S. (1843). A System of Logic, Ratiocinative and Inductive: 1 (Vol. 1). Parker. Namegabe Bugabo, J., Matabaro Rubanza, G., & Bayongwa Mirimba, G. (2015). Plainte contre la Société Bralima (Filiale de Heineken N.V en RDCongo) et Heineken N.V (basée à Amsterdam aux Pays-Bas) adressé au PCN Néerlandais. Dutch NCP. Nieuwenkamp, R. (2018). Beer, conflict and compensation: Heineken-Congo agreement. OECD Observer. OECD Watch. (2022). Complaints database. OECD Watch. https://www.oecdwatch.org/complaints-database/ OECD. (2011). OECD Guidelines for Multinational Enterprises, OECD Publishing. http://dx.doi.org/10.1787/9789264115415-en OECD. (2016). Implementing the OECD Guidelines for Multinational Enterprises: The National Contact Points from 2000 to 2015. OECD Guidelines for Multinational Enterprises. https://mneguidelines.oecd.org/oecd-report-15-years-national-contact- points.pdf OECD. (2020). What are National Contact Points for Responsible Business Conduct? OECD Guidelines for Multinational Enterprises. https://mneguidelines.oecd.org/ncps/what- are-ncps-for-rbc.htm 41 OEIGWG (2021, August 17). Legally Binding Instrument to Regulate, in International Human Rights Law, the Activities of Transnational Corporations and Other Business Enterprises. OEIGWG Chairmanship, third revised draft. https://www.ohchr.org/sites/default/files/Documents/HRBodies/HRCouncil/WGTrans Corp/Session6/LBI3rdDRAFT.pdf Oxfam Novib, Stichting Greenpeace Nederland, Stichting BankTrack, & Vereniging Milieudefensie. (2017, May 8). Klacht tegen ING vanwege schending van de OESO- richtlijnen op het gebied van klimaat. Dutch National Contact Point. Oxfam Novib. (2019a, April 19). OECD Guidelines: Banks must publish climate targets in line with Paris Climate Agreement. Dutch National Contact Point. Oxfam Novib. (2019b). Annual Report 2019/2020. Oxfam Novib. https://www.oxfamnovib.nl/Files/rapporten/2020/Jaarverslag/Oxfam-Novib-Annual- Report-2019-2020.pdf Pal, M. (2019). Early modern extraterritoriality, diplomacy, and the transition to capitalism. The Extraterritoriality of Law (pp. 69-86). Routledge. Perillo, P. S. (2022). The Role of the OECD Guidelines for Multinational Enterprises and the National Contact Points in Shaping the Future of Corporate Accountability, International Community Law Review, 24(1-2), 36-56. doi: https://doi.org/10.1163/18719732-bja10071 Piewitt, M. (2010). Participatory Governance in the WTO: How Inclusive Is Global Civil Society? Journal of World Trade, 44(2), 467-488. Pruitt, D. G. (2013). Negotiation behavior. Academic Press. Ruggie, J. G., & Nelson, T. (2015). Human Rights and the OECD Guidelines for Multinational Enterprises: Normative Innovations and Implementation Challenges. SSRN Electronic Journal. Elsevier BV. https://doi.org/10.2139/ssrn.2601922 Sanchez, J. C. O. (2015). The Roles and Powers of the oecd National Contact Points Regarding Complaints on an Alleged Breach of the oecd Guidelines for Multinational Enterprises by a Transnational Corporation. In Nordic Journal of International Law 84(1), 89–126. Brill. https://doi.org/10.1163/15718107-08401006 Steinmetz, J. (2019). Comparative Politics. Politics, Power, and Purpose: An Orientation to Political Science. 42 UN. (2011). Guiding Principles on Business and Human Rights. United Nations. https://www.ohchr.org/sites/default/files/documents/publications/guidingprinciplesbus inesshr_en Vos, C. (2021, May 29). ING en pensioenfondsen investeren in omstreden palmoliebedrijf, landroof in Kameroen gaat gewoon door. De Volkskrant. https://www.volkskrant.nl/cs-b6ed6922 Wouters, J. and Ryngaert, C., ‘Litigation for overseas corporate human rights abuses in the European Union: the challenge of jurisdiction’, The George Washington International Law Review, Vol. 40(4) (2009), pp. 939-975, pp. 943- 944. Zerk, J. A. (2010). Extraterritorial jurisdiction: lessons for the business and human rights sphere from six regulatory areas. Corporate Social Responsibility Initiative Working Paper No. 59. Cambridge, MA: John F. Kennedy School of Government, Harvard University. 43 7 Appendix Table 1. Frequency table of agreement. Tabulation of agree Agreement Freq. Percent Cum. (binary) 0 198 84.98 84.98 1 35 15.02 100.00 Total 233 100.00 Table 2. Frequency table of acceptance by NCP. Tabulation of accept Freq. Percent Cum. 0 128 54.94 54.94 1 105 45.06 100.00 Total 233 100.00 Table 3. Frequency table of extraterritoriality. Tabulation of extra Extraterritori Freq. Percent Cum. ality 0 55 23.61 23.61 1 178 76.39 100.00 Total 233 100.00 Table 4. Descriptive statistics of NCP quality. Descriptive Statistics Variable Obs Mean Std. Dev. Min Max NCPevl 233 25.502 6.983 10 35 44 Figure 1. Evaluation scores of NCPs. Table 5. Frequency table of host OECD membership. Tabulation of hostOECD Freq. Percent Cum. 0 179 76.82 76.82 1 54 23.18 100.00 Total 233 100.00 Table 6. Frequency table of reform. Tabulation of reform Freq. Percent Cum. 0 98 42.06 42.06 1 135 57.94 100.00 Total 233 100.00 45 Figure 2. Number of completed cases per year. Figure 3. Share of agreements per year. 46 Table 7. Link test of full model including extraterritoriality, NCP quality, OECD membership and year dummies. Logistic regression Number of obs = 147 LR chi2(2) = 24.82 Prob > chi2 = 0.0000 Log likelihood = -68.272782 Pseudo R2 = 0.1538 agree Coefficient Std. err. z P>z [95% conf. interval] _hat 1.132 0.441 2.560 0.010 0.267 1.997 _hatsq 0.059 0.159 0.370 0.710 -0.253 0.371 _cons 0.025 0.325 0.080 0.939 -0.613 0.662 Table 8. Classification statistics, full model. Results Table D ~D Total + 7 4 11 - 28 108 136 Total 35 112 147 Classified + if predicted Pr(D) >= .5 True D defined as agree != 0 Sensitivity Pr(+|D) 20.00% Specificity Pr(-|~D) 96.43% Positive predictive value Pr(D|+) 63.64% Negative predictive value Pr(~D|-) 79.41% False + rate for true D Pr(+|~D) 3.57% False – rate for true D Pr(-|D) 80.00% False + rate for classified + Pr(~D|+) 36.36% False – rate for classified – Pr(D|-) 20.59% Correctly classified 78.23% Table 9. Likelihood-ratio test for NCP quality (1), OECD membership of host country (2) and year (3). 1) Likelihood-ratio test Assumption: Mod a nested within Mod d LR chi2(1) = 12.76 Prob > chi2 = 0.0004 2) Likelihood-ratio test Assumption: Mod b nested within Mod d LR chi2(1) = 0.00 Prob > chi2 = 0.9499 3) Likelihood-ratio test Assumption: Mod c nested within Mod d LR chi2(11) = 9.65 Prob > chi2 = 0.5624 47 Figure 4. Influential observations. Note: The red line marks the mean of dbeta at 0.191166. Table 10. VIF statistics. Variable VIF SQRT VIF Tolerance R-Squared extra 1.77 1.33 0.5665 0.4335 NCPevl 1.07 1.03 0.9370 0.0630 hostOECD 1.70 1.30 0.5890 0.4110 Table 11. Logistic regression excluding outliers. Model 1 Model 2 Agreement Agreement Extraterritoriality -0.703* -1.014** (0.39) (0.42) NCP evaluation score 0.119*** (0.03) Constant -1.173*** -4.099*** (0.32) (0.93) N 213 213 48