1 | P a g e
The effect of ownership structure
on conflict resolution: the
moderating effect of MNE’s
experience.
Msc Business Administration - International Management Track
Name Cedric Corstiaan Willems
UvA 11399724
Date 18 August 2017
Supervisor Dr. Ilir Haxhi Second reader Dr. Johan Lindeque
2 | P a g e
Statement of originality
The document is written by Cedric Corstiaan Willems who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business of the University of Amsterdam is responsible solely for the supervision of completion of the work, not for the content.
3 | P a g e
Abstract
Multinational Enterprises (MNEs) and indigenous communities are often involved in lengthy, extensive, and violent conflicts. These conflicts are the result of the contradicting interests that both of the opposing sides have: MNEs exploit the land’s resources for economic purposes, whereas the survival of the indigenous communities depends on the environment. By focusing mainly on isolated cases, the existing literature offers a limited view on the dynamics and global drivers of conflict. This research isolation has impaired our full understanding of the complexity of conflicts and MNE governance characteristics as potential drivers of these conflicts. We argue that since the MNE ownership structure impacts the company performance, in turn, it would potentially impact the conflict resolution as well. Therefore, in this study we explore the dynamics of conflict resolution and examine what the global drivers of conflicts are. More specifically, we first investigate whether ownership structure affects the conflict resolution, and second, to what extent the MNE’s experience moderates this relationship. Rather than taking a case-by-case approach, this study takes a global perspective. Using a sample of 640 cases from 57 different
countries, we argue that ownership structure positively influences conflict resolution and that this relationship is positively moderated by MNE’s experience. Against our
expectations, ownership structure has a negative effect on conflict resolution. State-owned and publicly traded MNEs were found to be involved in longer and more violent conflicts. However, only the negative effect on conflict severity was significant compared to the effect on conflict duration, which was also negative but not significant. Furthermore, no
supporting evidence was found that MNE’s experience with indigenous communities positively moderates the relation between ownership structure and conflict resolution. Even though our claims were not supported, our contribution to existing literature is threefold. First, this study enhances the understanding of global drivers of conflict by taking a large sample quantitative approach, rather than an isolated case qualitative approach. Second, by investigating the unexplored effect of ownership structure and MNE experience on conflict resolution, this study opens new avenues for future research to investigate the drivers of conflict. Finally, this study is relevant because it provides insights for both managers and governments to better understand the dynamics of conflict
resolution.
Keyword: indigenous communities; MNEs; conflicts; MNE’s experience; ownership structure; conflict resolution
4 | P a g e
Table of content
1. Introduction 7 2. Literature review 12 2.1 Indigenous communities 12 2.2 Conflict 132.3 Corporate governance: Ownership structure 14
2.4 MNE experience 16
2.5 Stakeholder theory 17
3. Theoretical framework 19
3.1 Ownership structure and conflict severity and duration 19 3.2 MNE’s experience and conflict severity and duration 21
3.3 Conceptual framework 22
4. Data & methods 24
4.1 Sample & data collection 24
4.2 Variables 24 4.2.1 Dependent variables 25 4.2.2 Independent variable 26 4.2.3 Moderating variable 26 4.2.4 Control variables 27 4.3 Method 28
5. Results and analysis 30
5.1 Descriptive statistics 30
5.2 Correlations and multicollinearity 33
5.3 Regression analysis 35 6. Discussion 39 6.1 Findings 39 6.2 Theoretical implications 40 6.3 Practical implications 40 6.4 Limitations 41 6.5 Future research 42 7. Conclusion 43 8. References 44 Appendix A 48 Appendix B 49
5 | P a g e
List of figures & tables
Figure 1 – Conceptual model of hypothesized relations Table 1 – Summary of regression models
Table 2 – Descriptive statistics and correlations
Table 3 – Tolerance and Variance Indicator Factor values
Table 4 – Regression results for dependent variable “Degree of violence”
Table 5 – Regression results for dependent variable “Conflict duration (months)” Table 6 – Regression results for dependent variable “Conflict duration (short vs. long)” Table 7 – Sample 1: home countries MNEs involved in conflicts
6 | P a g e
Abbreviations
CEO – Chief Executive Officer
CSR – Corporate Social Responsibility MNE – Multinational Enterprise
7 | P a g e
1. Introduction
Tensions between Multinational Enterprises (MNEs) and indigenous communities are growing (Calvano, 2008). These tensions are leading to conflict(s) between MNEs and indigenous communities. As such, indigenous communities want to preserve the land they inhabit as it is extremely essential to their existence, while MNEs are trying to exploit the land’s natural resources for business purposes (Fontana et al, 2015). MNEs overlook the environmental and socio-cultural consequences of their actions. In addition, the indigenous communities often lack power, legitimacy, and/or urgency in comparison with other
stakeholders of the MNE, which puts them at a disadvantage (Calvano, 2008; Mitchell et al., 1997).
The conflicts between MNEs and indigenous communities received a tremendous amount of attention by media and politicians in recent years (Castro & Nielsen, 2001). The media’s attention on the MNE’s role in this conflict can hurt the MNE’s reputation, which would make one think that MNEs intend to resolve the conflicts in a timely manner. However, conflict mitigation strategies focus on reducing the negative effects rather than resolving the conflict altogether (Getz & Oetzel, 2009).
The conflicts between communities and MNEs have been on the research agenda for a long time. The relationship between multinationals and local communities is a complex one, as interests of both parties are understandable yet contradictory. As resources are becoming scarce, MNEs are finding ways to extract resources from all over the world to maintain a profitable business. Moreover, at the same time, exploiting land worldwide has become easier due to the new opportunities available, which is the result of globalization (Garrett, 2000). On the other hand, there are the indigenous communities who have
inhabited their land for extended periods of time. These communities are highly dependent on their local environment (Calvano, 2008; Fontana et al, 2015; Lertzman & Vredenburg, 2005). The fact that both interests and aspirations cannot be met at the same time means conflict can potentially arise (Franks et al, 2014).
Many resource extraction firms are involved with indigenous communities worldwide. The extraction industry is under pressure for more sustainable resource development as they are causing the most obvious environmental impacts (Lertzman & Vredenburg, 2005). Qualitative research on conflict dynamics is available yet little
8 | P a g e quantitative research on the dynamics of conflict is accessible (Calvano, 2008; Kolk & Lenfant, 2010). Many single-case quantitative studies have been conducted throughout the years rather than testing hypotheses quantitatively using a complete dataset (Barber & Jackson, 2012; Bebbington & Bury, 2009). Qualitative research gives valuable insights on specific cases but they are unable to clarify relationships we can transfer across different settings. We therefore aim to quantitatively explore the relationships in this study in a larger set of cases, rather than looking at it on a case-by-case basis.
According to Fontana et al (2015), three drivers promote the likelihood of social conflicts. The first driver points out that the likelihood of conflict depends on the industry in which the MNE operates (Fontana et al, 2015). Specifically, mining industries are more often involved in social conflicts. The second element considers the violation of human rights and the mistreatment of communities as the main events that contribute to the likelihood of being involved in social conflicts (Fontana et al, 2015). The third element is corruption that exists in countries that lack stable political systems (Fontana et al, 2015). We strongly believe that an important driver is absent and this driver is corporate
governance. More specifically, we believe that ownership structure of MNEs should be considered as a fourth driver for social conflict. The ownership structure of MNEs will impact the performance of a company and could potentially have an impact on conflict dynamics (Brunninge et al, 2007). Studies showed that ownership structure strongly impacts performance and strategy of companies (Anderson & Reeb, 2003; Daily & Dollinger, 1992; Mehran, 1995). We expect that ownership structure will also have an impact on conflict resolution. We therefore want to investigate this relationship by testing our hypothesis using our dataset. To our knowledge, no present studies consider the MNE’s ownership structure as a driver of social conflict. For this reason we are pursuing to
empirically test the effects of a fourth driver, the MNE’s ownership structure, on conflict resolution.
Most corporate governance research is inspired by the agency theory and financial systems including banking systems (Aguilera & Jackson, 2003; Aguilera & Jackson, 2010; Dalton et al, 2007). However, ownership structure is a relatively unexplored area within this research field. Our aim is to investigate the direct effect ownership structure has on the conflict severity, captured by the degree of violence. In addition, we also look at the direct
9 | P a g e effect of ownership structure on conflict duration exemplified by the length of the conflict. We hypothesize that certain ownership structures are either related to more violent and longer conflict, while other structures may experience less violent and shorter conflicts. For example, we expect privately owned MNEs to experience more violent and longer conflicts compared to state-owned and publicly traded MNEs. This brings us to the first research question:
RQ1: To what extent does ownership structure of MNEs affect the conflict resolution (i.e., conflict severity and conflict duration)?
Some MNEs have experienced conflicts with indigenous communities in the past. When encountering a conflict with indigenous communities, it is important to handle the conflict in a professional and effective manner to achieve solutions for the conflict. For instance, those MNEs that endured some form of interaction with indigenous communities in the past might develop a relationship and build trust with the indigenous community member. Previous encounters such as these can help the MNE in effectively handling conflict and potentially preventing conflicts from happening in the future.
Managers have encounters with the members of the indigenous communities and these managers’ specific style may have an influence on the direction of the conservation. Different management styles exist when it comes to conflict management. The traditional view by Deutsch (1973) includes two single dimensions. The first dimension is selfishness (i.e. concern about own interests) and the second dimension is cooperativeness (i.e.
concern with other party’s interests). However, neither two of these approaches seem to be effective, rather we need to meet somewhere in between. Balancing the two conflict
management dimensions is what MNEs and indigenous communities need to consider to avoid conflicts in the future.
management can also be considered to mediate between the two parties. Co-management can be described as “the situation in which two or more social actors negotiate,
define, and guarantee amongst themselves an equitable sharing of the management
functions, entitlements, and responsibilities for a given territory or set of natural resources”
(Borrini-Feyerabend et al, 2000). Reducing the negative consequences of conflict is important but resolving conflicts altogether is important for lasting peace (Getz & Oetzel, 2009). Experience in resolving conflicts and reducing negative consequences might
10 | P a g e increase the capability of an MNE to communicate with local communities. Hence to further develop our understanding of conflicts between indigenous communities and MNEs, this study will look at the moderating effect of MNE’s experience with indigenous communities on the effect described in our first research question. More specifically, we want to
investigate whether MNEs that have been involved with local communities in the past can moderate the direct effect between ownership structure and conflict resolution.
Understanding the moderating effect of MNE experience will add to the literature to better understand how MNE’s experience affects the relationship between ownership structure and community-MNE conflict. In addition, it will give managers an important insight on how MNE experience can help them either deal effectively with conflicts or prevent conflicts altogether. We therefore address our second research question:
RQ2: To what extent does MNE’s experience moderate the relationship between ownership structure and conflict resolution (i.e., conflict severity and conflict duration)?
Conflicts between MNEs and indigenous communities are escalating and many different groups are calling for action to become more accountable for the impact on their stakeholders (Calvano, 2008). Indigenous communities are important stakeholders for MNEs active in the resource extraction industry (Lertzman & Vredenburg, 2005) because they are the ones who inhabit the land. Conflict has negative consequences for both the MNE and the local community. The MNE suffers from negative financial and reputational effects (Calvano, 2008). The conflicts can cause injuries, and sometimes even death, to the people in the local communities. For those two reasons, it is important for both sides to resolve conflicts effectively.
To empirically test our hypotheses and answer our research questions, we use 640 cases that include MNEs from 57 different countries all over the world. Using cases from a large variety of countries gives us a more global perspective opposed to limiting ourselves to an isolated analysis of single cases. We analyze the effect of ownership structure and MNE’s experience on conflict resolution, which is measured by conflict duration and
conflict severity. We propose that ownership structure will have a positive effect on conflict resolution and that this relation will be positively moderated by MNE’s experience.
11 | P a g e The contribution of this study to the previous literature is threefold: First, previous studies attempted to explore the dynamics of conflict by qualitatively analyzing isolated cases but the studies failed to provide findings that are transferable across different
settings. Using a quantitative approach and a large set of cases will give us insights that are transferable across different geographical settings. Second, this study aims to extend our understanding of the corporate governance driver ownership structure and its effect on conflict resolution, i.e. conflict duration and conflict severity. By building on previous research, we target to bridge the gap in current literature to provide a clearer
understanding of the drivers of conflict. In addition, we investigate the moderating effect of MNE’s experience on this relation. To our knowledge, both ownership structure and MNE’s experience are unexplored areas of research. Finally, this study will be relevant for both governments and managers to understand the factors influencing conflicts and how to contribute to resolve conflicts. Having this understanding will help in resolving conflicts in a more effectively and quicker as well as preventing potential conflicts in the future. The odds of being involved in long and violent conflicts will be improved if this understanding is gained.
To answer our research questions, we start with a review of the existing literature on the areas of indigenous communities, conflict, corporate governance (ownership
structure), MNE’s experience, and stakeholder theory. We will then present our theoretical framework, the hypotheses, and the conceptual model. Furthermore, we will discuss the research design by presenting the sample, variables, and the methods of analyses that are used in this study. Hereafter, we will present our analyses and results from the multiple tests, such as descriptive statistics, correlation matrix, and the regression analyses. This will be followed by the discussion section where we will discuss our findings, the
theoretical and practical implications, the limitations, potential directions of future research, and the conclusion.
12 | P a g e
2. Literature review
In the following section we will review the literature on several different topics that are relevant to our study. When conflicts arise multiple stakeholders will be involved. It is important to understand the different stakeholders and other relevant related topics. We will review the literature on indigenous communities, conflicts, corporate governance, and MNE’s experience. This will help us gain a better understanding of the different topics and will help answer our two research questions.
2.1 Indigenous communities
Indigenous communities are one of the primary stakeholders of MNEs when looking at the resource extraction industry (Lertzman & Vredenburg, 2005). It is thus important to get a clear understanding of what an indigenous community is. Although there is no clear definition on communities, Calvano (2008) noted that the definition of communities includes dimensions of geography, social interaction, and identity. This is a good starting point, but we need a more precise definition on indigenous communities for our research. Communities are described more thoroughly as “aggregates of people who share common
activities and/or beliefs and who are bound together principally by relations of affect, loyalty, common values, and/or personal concern (i.e., interest in the personalities and life events of one another)” (Brint, 2001, p. 8). The definition ‘indigenous’ is described in many different
ways and again no exact definition exists throughout the literature. A simple definition is ‘to be born in a specific place’ (Cunningham & Stanley, 2003). In 2007, the rights of indigenous people were positively affected and were recognized in the United Nations Declaration on the Rights of Indigenous Peoples. The UNDRIP does not attempt to give one universal definition. Instead the declaration provides common characteristics of
indigenous people. It is not a binding declaration, but it does empower indigenous
communities with new arguments against corporate policies for example (Bellier & Préaud, 2012; United Nations, 2007).
According to the UNDRIP, indigenous peoples should be affirmed as equal to all other peoples, while recognizing the right of all peoples to be different. Additionally, indigenous peoples are described as those who suffered historic injustices because of the
13 | P a g e colonization or dispossession of their land and resources (United Nations, 2007). As well, the UNDRIP recognizes that respect for indigenous communities will contribute to
sustainable, rightful development, and suitable management of the environment. MNEs are involved with indigenous communities especially in the resource extraction industry (Lertzman & Vredenburg, 2005). It is therefore especially important for firms in this
industry to recognize the indigenous communities. However, the communities are not often given priority over other stakeholders (Calvano, 2008).
2.2 Conflict
Conflict can be described as “any relationship between opposing forces whether
marked with violence or not” (Deloges & Gauthier, 1997, p.4). This is a very broad definition
of conflict, and a more exact definition is given by Ochieng Odhiambo: “each party wants to
pursue its own interest to the full, and in doing so ends up contradicting, compromising, or even defeating the interest of the other” (2000, p. 8). The onset of a conflict between MNEs
and indigenous communities remains a much-debated matter and many perspectives exist throughout the literature (Calvano, 2008; Getz & Oetzel, 2009). According to Nillesen and Bulte, three main perspectives can be identified. The first perspective associates resource scarcity with more incentives for conflict, while the second, opposite perspective states that an abundance of resources will lead to more conflicts. A third perspective points out the complex relation between resources and conflict. According to this last perspective, many factors need to be considered such as institutional context and political regimes (Nillesen & Bulte, 2014).
It is evident that indigenous communities are dependent of the land’s natural resources the communities inhabit (Calvano, 2008), while globalization offers MNEs new opportunities to expand across borders for new business openings (Garret, 2000).
However, the friction between indigenous communities and MNEs can lead to social conflict (Fontana et al, 2015). The mining industry is often associated with conflict in comparison with other industries (Fontana et al, 2015). However, mining is not the only field that experiences conflict. Other fields like fisheries, agriculture, wildlife, and forestry among other related fields also have the chance of being involved in conflicts (Castro & Nielsen, 2001).
14 | P a g e In addition to the discussed points, indigenous communities are also skeptical when it comes to the intentions of MNEs with corporate social responsibility programs. The community feels that the MNE is only is involved in these programs to promote their image, as came forward from a survey among Argentine stakeholders (Fontana et al, 2015). MNEs only show acts of kindness, CSR, to cover up their own misdeeds or it is done as a symbolic attitude regarding the MNE’s responsibilities (Calvano, 2008; Fontana et al, 2015).
According to Calvano (2008), three factors are considered to influence MNC-community conflict. First, stakeholder power inequality where the lack of power of the community puts them at a disadvantage compared to the MNEs. Second, stakeholder perception gaps where both parties, the community and the MNE, have a certain perception of the other party, which is influenced by their ideologies. Lastly, cultural context needs to be considered as MNEs are most often not aware of the micro- or community-level cultural differences (Calvano, 2008) and context of region needs to be considered (Fontana et al, 2015). Fontana et al (2015) adds a fourth factor to this discussion which is the lack of stable political systems and the presence of corrupt governments. This makes it more likely for conflict to arise in certain countries underdeveloped countries (Fontana et al, 2015).
Conflicts can differ in length (Calvano, 2008; Getz & Oetzel, 2009) and in degree of violence (Calvano, 2008; Castro & Nielsen, 2001; Getz & Oetzel, 2009). Lengthy conflicts are not good for the MNE’s reputation, when covered in the media, and can be costly for the MNE as well. Violent conflicts are described as “organized physical force resulting from grievances between two or more parties and leading to injury or death to persons or damage or desctruction to propery” (Oetzel et al, 2007, p. 331). Violent conflict includes war, revolution, rebellion, insurgency, and sustained campaigns of violence or terrorism (Getz & Oetzel, 2009).
2.3 Corporate governance: ownership structure
Multiple definitions on corporate governance exist throughout the literature but Aguilera and Jackson gave a broad definition on corporate governance. “Corporate governance may be defined broadly as the study of power and influence over decision making within the corporation. Yet scholars have approached the subject of corporate
15 | P a g e governance from a variety of disciplinary perspectives, including economics, management, law, political science, culture, and sociology” (Aguilera & Jackson, 2010, p. 487). Aguilera herself uses a more in depth definition saying that corporate governance is the study of the spreading of rights and responsibilities amongst different participants in the business, including managers, shareholders, the board of directors, and other stakeholders (Aguilera, 2004).
Most corporate governance research is inspired by the agency theory and financial systems including banking systems (Aguilera & Jackson, 2003; Aguilera & Jackson, 2010; Dalton et al, 2007). According to Aguilera & Jackson (2010), there are two types of corporate governance structures. First, the Anglo-Saxon orientation, which is a more shareholder focused structure. Second, the Rhineland orientation, which is a more stakeholder oriented structure. The component of corporate governance studied
throughout this research is ownership structure. According to Brunninge et al (2007), the involvement in uncertain activities is expected to be influenced by the ownership structure and governance of the firm.
The dimension of corporate governance we will examine more closely in this study is ownership structure. Ownership structure can be divided into three different groups including state-owned, public traded, and privately owned. Previous studies looked at how ownership structure influences firm behavior (Hart & Moore, 1990). Ownership structure in relation to firm performance is also studied and several findings are presented in studies throughout the years. For instance, Mehran (1995) argued that privately owned companies positively influences firm performance. In addition to this claim, Anderson & Reeb (2003) claimed that family-owned firms outperform non family-owned businesses, especially if the CEO is a family member. Ownership structure of the MNE can potentially get a different response of indigenous communities, which is why this will be relevant to research. Boot et al. (2006) discuss the difference between publicly traded and privately owned companies as follows:
”Public ownership involves publicly traded shares and public corporate governance, with diffused ownership and control. Private ownership operates without a market listing and involves private contracting, typically with concentrated ownership and control. That is, private and public ownership differ along two dimensions namely, investor liquidity and the
16 | P a g e
allocation of control between managers and investors, which is determined through corporate governance.” (Boot et al, 2006, p. 803)
2.4 MNE’s experience
Many studies describe the causes of social conflict, while little attention is devoted to prevent conflicts in the future (Calvano, 2008; Fontana et al, 2015). The experience MNEs have gained during a conflict, or potentially multiple conflicts, in the past might help them deal with conflicts more effectively in the future. To better understand MNE
experience we need to look at collaborations between MNEs and indigenous communities in the past. If both parties collaborate on projects, i.e. when local resource users have share or exclusive rights to make decisions, conflicts may be prevented in the future (Castro & Nielsen, 2001). We will discuss some of the studies that looked at MNE experience and co-management and what dimensions may help to deal with conflict in a more effective manner or ultimately prevent conflict in the future.
As discussed earlier, there is a power inequality between MNEs and indigenous communities with the communities at a disadvantage (Calvano, 2008). Co-management agreements can be an effective way of dealing with natural resource-based conflicts (Castro & Nielsen, 2001). Co-management for many analysts refers to decision making about
natural resource access between state and communities (Castro & Nielsen, 2001). However, a broader definition is described as “the situation in which two or more social actors negotiate, define, and guarantee amongst themselves an equitable sharing of the management functions, entitlements, and responsibilities for a given territory or set of natural resources” (Borrini-Feyerabend et al, 2000, p. 79). This last definition is more appropriate for our study, as it can be the co-management between MNEs and the
indigenous communities rather than the co-management between state and communities. Recently, the term co-management is used where any institution, local peoples or key stakeholders engage in a partnership (Castro & Nielsen, 2001). Conflict management is historically seen as the responsibility of states, while recently MNEs are expected to
participate in conflict management as well (Getz & Oetzel, 2009). Co-management is
applicable worldwide and in multiple fields of resources like forestry, fisheries, agriculture, wildlife, and other protected natural resource areas (Castro & Nielsen, 2001). The essential
17 | P a g e part of co-management is that sharing decision making will make the process of resource management easier (McCay & Jentoft, 1998).
MNE’s experience with indigenous communities results in two benefits for the MNE that will help in resolving conflicts more effectively. First, the MNE will gain experience and knowledge about the conflicts. This experience and knowledge can be used by the MNE’s employees to have an advantage going into future conflicts with local communities. Second, the MNEs will build relationships with indigenous communities. These relationships can potentially result in a more effective way of dealing with future conflicts and the employees can apply what they have learned in previous conflicts with local communities (Alcantara & Nelles, 2013). To better understand the differences between cultures Lertzman &
Vredenburg (2005) explain that it is a skill of adaptation. In addition, they explain how both sides need try and understand each other and learn from each other (Lertzman &
Vredenburg, 2005). Moreover, it is important to keep in mind on how certain groups like to make decisions. For example, some indigenous communities might prefer to do this out in the open land instead of a formal setting (Whiteman, 2009). For both of these reasons, MNEs should use the experience and knowledge with conflicts to their advantage.
2.5 Stakeholder theory
For our study it is important to know a little bit more about the different stakeholder theories. Stakeholder theory can be described as the attempt to articulate which stakeholder groups are deserving of management’s attention and which are not (Mitchell et al., 1997). Mitchell et al. (1997) contributed to this and said that stakeholder can have none, one, two, or three of the following dimensions. The three dimensions discussed by Mitchell et al. (1997) are power, legitimacy, and urgency. The more of these dimensions a stakeholder includes, the more influence and more important this
stakeholder is for the MNE (Mitchell et al., 1997). The degree to what extent managers give priority to different (groups of) stakeholders is described as “salience” (Mitchell et al., 1997). Calvano (2008) adds to this stakeholder perspective that the inequality between the different stakeholders is what leads to conflict. Recognizing the different stakeholder
18 | P a g e groups can thus be a useful way to solve conflicts and potentially avoid conflict in the future.
The three different dimensions of stakeholder theory; power, legitimacy, and
urgency will now be explained. First, power can be explained as the influence one actor can have on another one. More specifically, it means that one actor can influence someone else to do something they otherwise would not have done (Mitchell et al. 1997). Second,
legitimacy can be described as the perception that actions are desirable and proper within certain norms, values, and beliefs (Mitchell et al., 1997). Lastly, urgency can be explained as to what extent a certain stakeholder’s claim calls for immediate action (Mitchell et al., 1997).
19 | P a g e
3. Theoretical framework
This study argues that two factors are influencing conflict resolution namely, ownership structure and MNE’s experience. Conflict resolution is measured by the combination of both conflict duration and conflict severity. This is because conflicts can differ in length as well as the degree of violence (Oetzel et al., 2007). An important characteristic of conflict is that a stakeholder, in our study the indigenous communities, may not be able to influence the firm’s decisions. The stakeholder’s inability to influence the MNE can result from the lack of power, legitimacy, and/or urgency (Mitchell et al., 1997). This results in long and violent conflicts (Calvano, 2008).
The resource extraction industry has experienced many violent conflicts as MNEs and indigenous communities are struggling to divide land, resources, and rights. Despite the fact that there are potential benefits to resolving conflicts between MNEs and
indigenous communities for MNEs, communities, and governments, no substantial attention has been devoted to the research on conflict resolution. First, the dynamics of conflicts has been addressed in fragmented ways where studies focused either on conflict severity or conflict duration (Calvano, 2008; Fontana et al, 2015), but focusing on the two dimensions together has not been explored yet. Second, previous research emphasized insights on specific conflicts situations using single case studies of conflicts between MNEs and local communities. To create a generalizable methodology, it is important to look at a large sample of MNE-community conflicts. Using a large dataset will result in the ability to transfer our findings to different geographic settings. Moreover, not enough research is available on the drivers of conflicts. More specifically, the research on corporate
governance dimensions is quite underdeveloped throughout the existing literature as a driver of conflict. We want to examine if ownership structure, as a dimension of corporate governance, affects conflict resolution.
3.1 Ownership structure and conflict severity and duration
Ownership structure is used as a common corporate governance dimension, along with other dimensions such as board composition and top management (Brunninge et al, 2007). In addition, ownership structure is related to the performance and strategies of
20 | P a g e MNEs (Anderson & Reeb, 2003; Daily & Dollinger, 1992; Mehran, 1995). MNE’s ownership structure can take three different forms namely: privately owned, publicly traded, or state-owned. Privately owned companies are sort of self-sufficient and publicly traded are more concerned about public reputation as they depend on their stocks and shareholders (Boot et al, 2006). In addition, the publicly traded companies that have more dispersed type of control will increase agency problems (Boot et al., 2006; Brunninge et al., 2007). On the other side there are state-owned companies and these strive directly for more
socioeconomic and political objectives. Thus, instead of creating maximum profits, the wealth distribution seems to be the main objective for state-owned companies (Dewenter & Malatesta, 2001). The effect of corporate governance on conflict resolution has not been researched thoroughly yet, more specifically the effect of MNE’s ownership structure on conflict resolution has not been studied yet. Therefore, the way MNEs are organized could affect conflict resolution.
As discussed above, the key objective of privately owned MNEs is to increase profits opposed to publicly traded MNEs that care mainly about their reputation. The way MNEs are organized can result in different responses from indigenous communities. If the MNE’s sole objective is about wealth distribution instead of creating maximum profits, which tends to be the case for state-owned or publicly traded MNEs rather than privately owned MNES, the local communities will be likely to be less violent throughout the conflict.
Following this reasoning, we expect to see more violent conflicts between privately owned MNEs and indigenous communities. This is because the indigenous communities disagree with the privately owned MNEs in regards the maximization of profits pursued, while the privately owned MNE does not take into consideration the desires of the local communities. On the contrary, we expect to see less violent conflicts for publicly traded and state-owned MNEs due to the fact that publicly traded companies invest more in their reputation and state-owned MNEs invest more in wealth distribution. We therefore predict that:
H1: Ownership structure positively affects conflict severity, where privately owned MNEs show higher degrees of violence.
In line with the previous hypothesis, we expect the influence of ownership structure on conflict duration to be similar as the conflict severity. Different studies show the
21 | P a g e (Anderson & Reeb, 2003; Daily & Dollinger, 1992; Mehran, 1995). Similarly to the
previously presented arguments where privately owned MNEs are mainly focused on maximizing profits. This could potentially result in longer conflicts for privately owned MNEs. The reason behind this is that privately owned MNEs may receive more resistance from the indigenous communities when they want to extract resources from the
communities’ land. On the other hand with publicly traded and state-owned MNEs, they may have more trust with the indigenous communities due to their business objectives as mentioned above. We therefore propose:
H2: Ownership structure positively affects conflict duration, where privately owned MNEs tend to be involved in longer conflicts.
3.2 MNE’s experience and conflict severity and duration
In general, if MNEs want to be able to resolve conflicts effectively, their focus should be on political, strategic, social, and ethical dimensions. They should also attempt to gain the trust of the indigenous communities (Berkes, 2009 & Fontana et al., 2015). No matter how the MNE is organized (i.e. ownership structure), they need to have experience with the indigenous communities to build a foundation of trust. Thus, the fact that MNEs have prior experience with indigenous communities will result in more knowledgeable employees of the MNEs who will assist in better conflict management. In addition to the development of knowledge through their experiences, the MNEs will also form relationships with
indigenous communities. As discussed, a stakeholder may lack power, legitimacy, and urgency to influence MNE’s decisions or strategy (Mitchell et al., 1997). So, these relationships have a greater importance if the government institutions are weak in the countries where the indigenous communities are located.
MNE’s experience can exist in many dimensions, one of these being co-management. Co-management can be an effective way for MNEs to prevent conflicts from happening (Castro & Nielsen, 2001). The application of co-management is valid worldwide, which is relevant for our study as we take a global perspective. Moreover, co-management will be effective as it involves the community in the decision-making process (Castro & Nielsen, 2001). McCay and Jentoft (1998) add on co-management that sharing authority and
22 | P a g e decision-making will make the process of dividing resources easier. Moreover, MNEs can gain experience in negotiation strategies and gain trust with the communities.
Furthermore, the experience could also exist of improving the knowledge of the culture, institutions, and customs. If MNEs fail to recognize the differences, it can create tensions between the local community and the MNEs. These tensions between MNEs and indigenous communities can lead to conflict (Calvano, 2008).
For our moderating variable, MNE’s experience with indigenous communities, we expect the MNE’s experience to positively moderate the direct effect between ownership structure and conflict resolution. For our research, MNE’s experience is expressed in years, starting from the first time they worked with an indigenous community. MNEs that are superior at building relationships throughout the years with indigenous communities will have an advantage over the MNEs that do not have an established relationship. MNE’s experience can be seen as a mediator between the MNE and the indigenous community due to the foundation of trust among both parties involved. We can expect that MNEs who have been working together with indigenous communities for a longer period of time to be involved in less severe conflicts due to the experience gained in the past. We therefore predict that:
H3: MNE experience positively moderates the relationship between ownership structure and conflict severity.
Following the same reasoning as the moderating effect of MNE’s experience on conflict severity, we also expect to see a positive moderating effect on conflict duration. If MNEs manage to have a trusted relationship with the local communities, we expect them to not be involved in long conflicts. The reason for this is that communities will be more likely to cooperate with the MNE, as they trust the MNE. We therefore predict that:
H4: MNE experience positively moderates the relationship between ownership structure and conflict duration.
3.3 Conceptual framework
This study will empirically test four hypotheses where each hypothesis will discuss one of the hypothesized relations. Ownership structure is the independent variable and is
23 | P a g e one of the dimensions of corporate governance. Ownership structure is researched as a driver for conflict in this study. The dependent variables conflict severity and conflict
duration form one overarching dependent variable called conflict resolution. To empirically
analyze our hypotheses, we will use degree of violence for conflict severity and we will use conflict length for conflict duration.
Figure 1 shows the conceptual framework we use for this study. It is expected that ownership structure positively affects conflict severity and positively affects conflict duration. In addition, it is expected that MNE’s experience will positively moderate the relationship between ownership structure and conflict severity. Similarly, the relationship between ownership structure and conflict duration is expected to be positively moderated by MNE’s experience.
Figure 1 – Conceptual model of hypothesized relations
CG: ownership structure MNE experience MNE experience Conflict severity Conflict duration H1 H2 H4 H3
24 | P a g e
4. Data and methods
In this section we will explain how the data was gathered, what sample we used, and what variables we include to test our data. In addition, we will show what method is used to test our data.
4.1 Sample & data collection
Using purposive sampling, from the 709 available cases we decided to exclude 69 cases due to missing values of one or more of the variables. The sample therefore consists of 640 cases of conflicts between MNEs and indigenous communities in Argentina,
Australia, Bolivia, Brazil, Cambodia, Canada, Cayman Islands, Chile, China, Colombia, Denmark, Dominican Republic, Ecuador, Ethiopia, Finland, France, Germany, Hong Kong, India, Ireland, Italy, Japan, Jersey (Channel Islands), Kenya, Kuwait, Libya, Luxembourg, Malaysia, Mauritius, Mexico, Namibia, Netherlands, Nicaragua, Nigeria, Norway, Pakistan, Panama, Peru, Philippines, Poland, Republic of Korea, Russia, Saudi Arabia, Singapore, South Africa, Spain, Sweden, Switzerland, Taiwan, Tanzania, Thailand, United Arab
Emirates, United Kingdom, United States of America, Venezuela, Vietnam, and Zimbabwe. The data used for this study is secondary data meaning that the data is retrieved from third party databases and reports on conflicts between indigenous communities and MNEs. This data was gathered by students who were writing their thesis under the
supervision of Dr. Haxhi. The data is recorded using a set coding scheme, so that all data would be coded the same. The collected dataset consists of cross-sectional data, which means that there will only be one observation per variable of the conflicts. Including cases from all over the world in our sample is appropriate for our analysis to get a better
understanding if the hypotheses would be true in globally and if so could receive more attention by future researchers.
4.2 Variables
We will now outline all of the variables included in our study, including the dependent, independent, moderating, and control variables. We will outline what this variable means and how it is coded in the dataset used for our study.
25 | P a g e
4.2.1 Dependent variables
This study uses two dependent variables to explain conflict resolution, including: degree of violence and duration of conflict.
First, we will use conflict length as a dependent variable, as conflicts can differ in length (Calvano, 2008; Getz & Oetzel, 2009). The duration of conflict will be measured in two different ways. In the dataset we have the duration conflict recorded as total months, which will be one of the variables we will use. We will record conflict duration also as short and long. After gathering our sample, we calculated the mean in months. The mean of conflict duration for this sample is 119.06 and conflicts shorter than the mean will be coded as short (0) and conflicts longer than the mean will be coded long (1). By coding this way, we create a dummy variable for conflict length. In this study a conflict will be
considered ended in case of a settlement either with or without the involvement of the legal system. An example of a short conflict is the Reid Gartner Coal plant project between the Moapa Band of Paiute Indians and NV Energy in the United States. This conflict lasted for 23 months until a settlement was reached. An example of a long conflict is the pollution turtle cove between the Mohawks and ALCOA in the United States. The conflict lasted for 363 months until a settlement was reached.
Second, we will use conflict severity as a dependent variable and for this variable we used degree of violence from our dataset. Conflicts can vary in severity of violence, which can be measured as degree of violence (Calvano, 2008; Castro & Nielsen, 2001; Getz & Oetzel, 2009), which will be measured on a 5-point scale from 1 (low violence) to 5 (max level of violence) where:
(1) Low level of violence: peaceful protests, peaceful negotiations (2) Low level of violence: court actions
(3) Mid-level of violence: mid-level of violence (from either side) and intimidation tactics
(4) High level of violence: high level of violence (from either side) including physical damage, but no kidnapping; including kidnapping, but no death.
(5) Max level of violence: high level of violence (from either side) including death The two variables together form an overarching dependent variable namely conflict resolution. An example of a conflict with the maximum level of violence is in South Africa
26 | P a g e between Lonmin and the Marikana community. The Lonmin mine resulted in a maximum level of violence as 34 miners were shot dead and 78 wounded (McClenaghan & Smith, 2013). An example of a conflict where only low level of violence has been recorded is in Peru concerning the Lagunas Norte mine where Barrick Gold Corporation and Andean communities were involved in the conflict.
4.2.2 Independent variables
For the independent variable we will use a corporate governance dimensions of MNEs, which is the ownership structure of the MNE. For ownership structure there are three possibilities: namely publicly traded, state-owned, and privately owned. We will make two groups out of the three possibilities, which will structure the data on the ownership structure as a dummy variable where:
(0) Publicly traded and state-owned (1) Privately owned
An example of a publicly traded company involved in a conflict with an indigenous community is Canadian Gold Canyon Resources. The MNE is involved with the Mang’anja (Bunta) community in Malawi over the Mulanje Massif Rare Earth Mineral Exploration. An example of a state-owned MNE is Hydro Quebec, where the company is involved in a conflict over James Bay Cree hydroelectric in Canada with the Cree community. Lastly, an example of a privately owned MNE is Green Resources AS. The company is involved in a conflict with the Yao community in Mozambique over the AS Niassa Project.
4.2.3 Moderating variables
MNE’s experience with indigenous communities is used for the moderating variable in this study. MNE’s experience can have affect the relationship between the ownership structure and conflict resolution. The MNEs can start building relationships with
communities and prevent future clashes from happening, this is why MNE’s experience will act as the moderating variable in this study. MNE’s experience is recorded as the first year in which the MNE worked together with an indigenous community. We have created a new variable in which we calculated the amount of years it has been since the first project with an indigenous community. By recording MNE’s experience this way, it gave us the MNE
27 | P a g e experience, starting from the first encounter until 2017, in total years as our moderating variable. By coding MNE’s experience this way, we created a continuous variable.
4.2.4 Control variables
This study used control variables on three levels of analysis to control for the effect of country, community, and firm characteristics. At the country level, institutional strength is used as a control variable. This, newly calculated variable, includes six indicators from the Worldwide Governance Indicators (WGI) project by the World Bank: voice and
accountability, political stability, government effectiveness, regulatory quality, rule of law, and control of corruption. All of these indicators may contribute to the resolution of conflicts and therefore needs to be controlled for. Each one of these six indicators is coded into a six-category scale, where:
(1) 0 to 9th percentile (2) 10 to 24th percentile (3) 25 to 49th percentile (4) 50 to 74yh percentile (5) 75 to 89th percentile (6) 90 to 100th percentile
Institutional strength is then computed by adding all six indicators and taking the average of these six.
At the community level, geographic isolation is used as a control variable. When a community is more isolated, there will be less contact between the indigenous community and the rest of the world. More geographically isolated communities will most likely have less media coverage, NGO attention, and political attention. The opposite is considered to be true for less isolated communities, they will receive more attention. Besides the
geographic isolation, there might also be cultural and language barriers for communities to communicate. This variable is coded as follows:
(1) Members live within a community with little to no contact with outside world (2) Members live within broader municipalities with mixed populations
At the firm level, number of projects that affect indigenous communities is used as a control variable. The number of projects where MNEs are involved with indigenous
28 | P a g e communities can affect the conflict resolution. This variable can be closely related to MNE’s experience with indigenous communities and needs to be controlled for this particular reason. The number of projects that affect indigenous communities will be coded as the count number of projects around the world in community territory or affecting
communities since the start of operations.
4.3 Methods
Since this study includes two dependent variables, two separate analyses are
performed. The relationship between ownership structure and conflict severity and conflict severity (degree of violence) are tested using a linear regression analysis and are described by the following equation:
𝛾 = 𝛽0 + 𝛽1𝑋1 + 𝛽1 ∗ 𝛽2𝑋2 + 𝜀
In this case Y represents the dependent variable conflict severity. The regression coefficients are represented by β0, which is the intercept, and β1, which represents the MNE’s ownership structure as an independent variable, and β1*β2, which represents the interaction between MNE’s experience and ownership structure. Lastly, ε stands for the difference between Xi and the actual Xi (Field, 2009). The dependent variable, degree of violence, is a categorical variable but still a linear regression is used. According to
Rhemtulla et al. (2012), linear methods yield better results than other alternatives when the dependent variable has five or more categories. Since degree of violence has five categories, it is justifiable to use a linear regression according to Rhemtulla et al. (2012).
The relationship between ownership structure and conflict duration is tested using two separate analyses. First, for conflict duration in months another linear regression is used and this is tested using a stepwise approach. Second, for the dependent variable conflict duration, which is recorded as a binary variable, a logistic regression will be used. For the binary variable two outcomes are possible, namely short or long. The logistic regression is done using a stepwise approach starting with the control variables, then adding the predictor variables later to see their effects. According to Field (2009), a stepwise approach is a good option when hypotheses that you will be testing are not supported by a lot of evidence. In Table 1, the stepwise approach is shown as different models where each model includes one of the dependent variables.
29 | P a g e
Control variables Dependent variables Independent
variable Moderating variable Numbers of projects affect communities Geographic isolation Institutional strength Degree of violence Conflict duration (months) Conflict duration (short/long) Ownership structure MNE experience Model 1 X X X X Model 2 X X X X X Model 3 X X X X X X Model 4 X X X X Model 5 X X X X X Model 6 X X X X X X Model 7 X X X X Model 8 X X X X X Model 9 X X X X X X
30 | P a g e
5. Results and analysis
In this chapter the statistical analyses are presented. Before conducting the
regression analysis, it is important to analyze the correlations between all of the variables. The descriptive statistics, correlations, and tests for multicollinearity will be discussed first. After this the findings of the regression analysis will be presented. The results will be used to draw conclusions regarding the hypotheses, which were proposed earlier.
5.1 Descriptive statistics
Table 2 shows all of the descriptive statistics and the correlations coefficients of the dependent, independent, moderating, and control variables. The average ownership
structure is 0.23 with a standard deviation of 0.42. Since this variable is binary and it only takes the values 0 and 1, this means that 23% of the companies are privately owned and 77% is either publicly traded or state-owned. The average degree of violence is 2.76 with a standard deviation of 1.47. The average level of violence in conflicts therefore lies at level 3, which is considered a medium level of violence including moderate violence from either side and intimidation tactics. The average for the duration of conflicts is 119.06 months with a standard deviation of 114.31. This means that the average conflict of this specific sample goes on for 9 years and 11 months. The minimum duration of conflict for this sample is 1 month and the maximum duration of conflict for the sample is 1098, which is 91 years and 6 months. Conflict duration is also coded as a binary variable where we look at short and long conflicts, where a conflict is coded as short (0) when it is below the mean of 119 months and coded as long (1) when the conflict duration was longer than the mean. The average for the length of conflict (short vs. long) is 0.36 with a standard deviation of 0.48. This means that 36% of the conflicts were long and 64% of the conflicts short. The average MNE experience with indigenous communities is 26.96 with a standard deviation of 25.64, which means that the average amount of years the MNEs have experience since they first experienced an encounter with indigenous communities is almost 27 years. The average of geographic isolation is 1.46 with a standard deviation of 0.55. The average of the number of projects that affected indigenous communities is 31.34 with a standard
31 | P a g e deviation of 39.47. On average 31 projects were affecting indigenous communities in the cases used in the sample. Lastly, the average for institutional strength is 3.62 with a standard deviation of 1.42.
32 | P a g e Mean SD 1 2 3 4 5 6 7 8 1. Degree of violence 2.76 1.468 1 2. Conflict length (months) 119.06 114.31 .159** 1 3. Conflict length (short vs. long) .36 .480 .150** .680** 1 4. Ownership structure .23 .420 -.066 .013 -.004 1 5. MNE experience (years) 26.96 25.643 -.066* .064* .112** -.088** 1 6. Number of projects affected 31.63 39.686 -.018 -.022 .021 -.114** .328** 1 7. Institutional strength 3.620 1.142 -.174** -.204** -.155** -.175** .003 .078** 1 8. Geographic isolation 1.64 .553 -.069 -.038 -.034 -.029 .066* .055 .021 1
*. Correlation is significant at the 0.05 level (2-tailed). **. Correlation is significant at the 0.01 level (2-tailed). +. Correlation is significant at the 0.1 level (2-tailed).
33 | P a g e 5.2 Correlations and multicollinearity
Table 2 displays the correlation between the independent, dependent, moderating, and control variables of this study. For the correlation test the Kendall’s tau coefficient is used, as this is the most suitable for datasets where many scores have the same rank. In addition, it is also argued that more accurate generalizations can be drawn from Kendall’s statistic (Field, 2009). The first significant correlation is found between the dependent variables conflict length, both short vs. long and in total months, and degree of violence. This relationship indicates that when the length of conflicts increase, the level of violence increases as well. In addition, conflict length in months and conflict length (short vs. long) strongly and significantly correlate with each other.
The correlation between ownership structure and degree of violence is negative at -.066, yet insignificant. This is opposite of what we would expect based on hypothesis 1, as we expected this relationship to be positive. An interesting difference is found between ownership structure and conflict duration, this difference lies in the way conflict length is recorded. Even though, the correlations are insignificant it is an interesting finding to mention. The correlation between ownership structure and conflict length in months is positive at .013. However, the relationship between conflict lengths as a binary variable (short vs. long) is a slightly negative relationship at -.004. The relationship between ownership and conflict duration was expected to be positive based on hypothesis 2. As discussed, none of the correlations between the independent variable (ownership structure) and the dependent variables (conflict duration and conflict severity) were significant.
MNE’s experience displays significant correlations with both the independent and dependent variables. MNE’s experience negatively correlates with conflict severity at -.066 (p < .05). In addition, MNE’s experience also correlates negatively with ownership
structure at -.088 (p < .01). MNE’s experience positively correlates with both conflict length recorded as a total months, at .064 (p < .05), and with conflict length recorded as a binary variable at .112 (p < .01). Among the control variables some correlations are worth discussing in greater detail. First, the variable ‘number of projects that affect indigenous communities’ negatively correlates with ownership structure -.114 (p < .01) and positively
34 | P a g e with MNE’s experience .328 (p < .01). Institutional strength shows many significant
correlations, as it correlates negatively with the independent and dependent variables. Institutional strength correlates -.174 (p < .01) with degree of violence. It also negatively correlates with conflict duration in months at -.204 (p < .01) and with conflict duration (short vs. long) at -.155 (p < .01). Moreover, institutional correlates with ownership structure at the -.175 level (p < .01). Geographic isolation correlates at a significant level with MNE’s experience at .066 (p < .05).
To test for multicollinearity between variables two different methods were used. First, Table 2 shows that none of the correlations are above 0.80, which would be
considered high (Field, 2009). This is the first indicator that no multicollinearity exists. In addition, the Variance Indicator Factor (VIF) is used as a second measure to test if
multicollinearity between variables exists. Table 3 displays the VIF values and tolerance for the independent, moderating, and control variables. The VIF indicates if any of the
predictors has a strong linear relationship with any of the other predictors. A strong linear relationship exists if the VIF value exceeds 10. As shown in Table 3, none of the variables VIF scores is higher than 10. This means there is no need for concern about these values (Field, 2009). Based on the correlations and VIF tests, we can confidently say that no multicollinearity exists.
Conflict duration (months) and degree of
violence
Conflict length (short vs. long)
N = 640 N = 435
Tolerance VIF Tolerance VIF
Ownership structure .953 1.049 .965 1.037
MNE experience .891 1.122 .927 1.078
No. projects affect communities .881 1.136 .904 1.106
Geographic isolation .983 1.017 .987 1.013
Institutional strength .950 1.053 .957 1.045
35 | P a g e 5.3 Regression analysis
To be able to evaluate the hypotheses of this study, regression analyses are
performed using IBM SPSS Statistics version 24. The regression analysis were performed using a stepwise approach where we started with only the control variables and later adding the independent variable followed by the moderating variable. We did this for each of the dependent variable, as such we have 9 models to systematically show our findings. The β-values for the variables are used to evaluate the effect of each variable on the
dependent variable (Field, 2009). R² is used to determine the overall fit of the model and a results in the analyses are considered significant if the significance (p-value) is lower than .01. For the moderating variable, MNE experience, an interaction variable between
ownership structure and MNE experience is created by multiplying both centralized variables. For model 1 through 6, the sample consists of 640 cases. Cases were excluded if data was missing for one or more of the variables included in the analyses. For model 7 through 9 a total of 435 cases was used, as we excluded cases where data was missing for one or more of the variables used for the analyses. In addition, cases were excluded where the conflict was still ongoing and the total months did not yet exceed the mean of 119 months. Since we cannot be sure whether or not this conflict will exceed the mean of 119 months, it was better to remove these specific cases.
Table 4 shows the results of the regression analyses that are performed to analyze the effect of the explanatory variables on conflict severity, in terms of degree of violence. Ownership structure has a significant negative effect (β -.420, significance .002) on the dependent variable degree of violence. This suggests that ownership structure and degree of violence move in opposite directions. When ownership structure increases towards privately owned MNEs, the degree of violence will decrease. However, this is opposite of hypothesis 1. Hypothesis 1 is therefore rejected. MNE experience has no significant effect on the relationship and hypothesis 3 is therefore rejected. An interesting finding in models 1 through 3 is that institutional strength shows a significant negative coefficient (β -.312, significance .000) with degree of violence. This suggests that conflicts in stronger
institutional environments tend to be less violent. The adjusted R squared for model 1 through 3 increases slightly when we added ownership structure. However, when we added the moderating variable in the model it decreases slightly. The adjusted R squared
36 | P a g e suggests that 8.7 percent of the values are a good fit in this model. More specifically, it suggests that 8.7 percent of the variance of conflict severity can be explained by ownership structure, MNE experience, number of projects affected local communities, geographic isolation, and institutional strength.
Table 5 and 6 show the results of the regression analyses that are performed to analyze the effect of the explanatory variables on conflict duration (in months). The following models are related to hypotheses 2 and 4. The way of coding conflict duration makes a different in the observed effects. Ownership structure shows a negative effect (β -11.455, significance .291) on conflict duration recorded as a continuous variable, this is when conflict duration is recorded in months. However, a positive effect (β .067,
significance .798) is observed when conflict duration is coded as a binary variable, this is when conflict duration is either short or long. The findings are contradictory, yet
insignificant. Therefore, hypothesis 2 is rejected. The adjusted R squared in model 4
through 6 is around .050, which is low. The goodness-of-fit suggests that 4.8 percent of the variance of conflict duration in months can be explained by ownership structure, MNE experience, number of projects affected local communities, geographic isolation, and institutional strength.
Model 1 Model 2 Model 3
β β β
Constant 3.876* 4.079* 4.079*
Control
variables Number of projects affected communities .001 .001 .001
Geographic isolation -.059 -.073 -.073 Institutional strength -.292* -.312* -.312* Independent variable Ownership structure -.420* -.420* Moderating variable MNE experience .001 Model strength Adjusted R Square .076 .088 .087
*. Correlation is significant at the 0.01 level (2-tailed). **. Correlation is significant at the 0.05 level (2-tailed). +. Correlation is significant at the 0.1 level (2-tailed).
37 | P a g e Lastly, the moderating effect of MNE’s experience is tested. Similar to the effect of the independent variable, the effect of the moderating variable (MNE’s experience) is contradictory. In Model 6 this effect is positive (β .119, significance .781) and in Model 9 this effect is negative (β -.010, significance .341). Since the findings are not significant, hypothesis 4 is rejected. Interesting is that institutional strength shows significant negative coefficients. Institutional strength in relation to conflict duration in months shows β -19.831 with a significance of .000 and in relation to conflict duration short vs. long shows β -.516 with a significance of .000. This suggests that conflicts in stronger institutional
environments tend to be shorter in length compared to conflicts in weaker institutional environments. The adjusted R squared for model 7 through 9 is around .120, this is a low score for the adjusted R squared. This goodness-of-fit measure suggests that 12.1 percent of the variance of conflict length (short vs. long) can be attributed to ownership structure, MNE experience, number of projects affected local communities, geographic isolation, and institutional strength.
Model 4 Model 5 Model 6
β β β
Constant 189.069* 194.799* 194.877*
Control
variables Number of projects affected communities .088 .075 .079 Geographic isolation -3.116 -3.493 -3.301 Institutional strength -18.699* -19.246* -19.381* Independent
variable Ownership structure -11.856 -11.455
Moderating
variable MNE experience .119
Model
strength Adjusted R Square .049 .050 .048
*. Correlation is significant at the 0.01 level (2-tailed). **. Correlation is significant at the 0.05 level (2-tailed). +. Correlation is significant at the 0.1 level (2-tailed).
38 | P a g e
Model 7 Model 8 Model 9
β β β
Constant 2.382* 2.358* 2.363*
Control
variables Number of projects affected communities .001 .001 .001
Geographic isolation -.211 -.212 -.234 Institutional strength -.528* -.526* -.516* Independent variable Ownership structure .072 .067 Moderating variable MNE experience -.010 Model strength Adjusted R Square .119 .119 .121
*. Correlation is significant at the 0.01 level (2-tailed). **. Correlation is significant at the 0.05 level (2-tailed). +. Correlation is significant at the 0.1 level (2-tailed).
