The effect of MNE experience in conflict situations with indigenous communities

(1)

Master Thesis

MSc Business Administration – International Management

Student: Pepijn Rogier Janssen Student number: 11408561 Supervisor: Dr. Ilir Haxhi

Second reader: Drs. Erik Dirksen Date: 26-01-2018

(2)

2

Statement of originality

This document is written by student Pepijn Janssen who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

List of tables and figures

Table 1. Regression on length of conflict Table 2. Regression on type of violence Table 3. Regression on degree of violence Table 4. Descriptive statistics and correlation Table 5. Multicollinearity for conflict length Table 6. Multicollinearity for type of violence Table 7. Multicollinearity for degree of violence

Table 8. Logistic Regression on dependent variable length of conflict Table 9. Linear Regression on dependent variable type of violence Table 10. Linear Regression on dependent variable degree of violence

(6)

6

Abstract

Due to increasing global pressures, multinational enterprises (MNEs) seeking lower costs and abundant natural resources are expanding their business activities worldwide into areas inhabited by indigenous communities. These activities often result in conflicts between MNEs and indigenous communities, as these conflicts negatively affect both the MNE’s performance and these communities’ living conditions. A few previous studies have explored these conflicts from a single-case perspective. However, they fail to examine the firm- and community-level drivers behind these conflicts on a broad, global scale. These cases have never been researched in an extensive, quantitative study that combines multiple concepts. In the current study, we address this gap first by exploring how MNE’s experience influences conflict resolution (i.e. length and conflict severity) and second, by examining the extent to which the geographical and cultural isolation of the indigenous community moderates this relationship. We test these relationships using a sample of 544 cases from 92 different countries around the world. My results show that MNE experience negatively influences both the length of the conflict and its degree of violence, and that these relationships are moderated negatively by the geographical and cultural isolation of the indigenous community. This thesis contributes to existing literature theoretically by combing MNE experience, conflict length, conflict type, and geographic isolation and cultural isolation between MNEs and indigenous communities in a quantitative manner. Practically, this thesis provides valuable information and insight for MNEs, governments, NGOs, and all other stakeholders involved in these conflicts. The empirical results of this thesis can help stakeholders avoid or resolve conflicts effectively.

Keywords: MNE experience, indigenous communities, conflict length, conflict violence,

(7)

7

Introduction

According to Calvano (2008), there has been a shift in the globalised world of MNEs. It used to be common for MNEs to do business, and discuss certain deals and their conditions, with the governments of their host countries. However, this communication has shifted toward dealing directly with the indigenous communities present in the area where the MNE wants to execute its business activities. The last financial crisis caused an enormous downfall in foreign direct investments (FDI) worldwide. However, the amount of FDI is rising again toward its level in 2007 (OECD, n.d.). This rising FDI amount practically means that MNEs are increasing their international activities and will start more projects in areas where indigenous communities are present. Conflicts can negatively affect both the indigenous community (Foster, 2012) and the profitability and the reputation of MNEs (Laplante & Spears, 2008). Since a conflict negatively influences both parties, the outcome of this research can benefit both parties.

An indigenous community is significantly different from the dominant societies by which it is surrounded, in multiple aspects, including in cultural, social, economic, and political aspects as well as in language, knowledge system, and belief (UN, n.d.). These differences are potential triggers for conflicts. In order to deal with conflicts, the MNE management should pay attention to the community perspective, and on the other hand, the communities should increase their power (Calvano, 2008). MNEs consider indigenous communities to be stakeholders (Bruijn & Whiteman, 2010). The majority of conflicts between MNEs and indigenous communities arise from poor communication and other preventable accidents (Whiteman, 2009). Therefore, both the MNE and the indigenous community benefit from excellent communication. This can be realised when indigenous communities are represented by a person who is recognized by the entire community and is culturally literate.

(8)

8

Existing literature shows that the isolation of people and communities can lead to difficulties and conflicts. The more isolated a community, the larger the cultural distance becomes (Gupta & Ferguson, 1992). This makes effective communication difficult which, as described earlier, is one of the most important reasons for conflict (Whiteman, 2009). Dowie (2011) states that when cultures are different from one another, opposition arises. He even states that diverting cultures can be a true catalyser for violent conflicts. Furthermore, a study by Harlow, Dodsworth and Harlow (1965) show that isolation can make people more hostile.

Companies tend to overlook the costs that result from manpower being directed to managing conflicts (Davis & Franks, 2011). Conflicts are primarily managed by senior executives and, in some cases, even the CEO, making it an even more costly situation. A study by Goldman Sachs (2011, in Davis & Franks, 2011) concluded that the start-up time of projects by international oil companies has almost doubled in the last decade. Furthermore, the study indicated that indigenous communities are in the single, largest category of risk. This makes conflicts with indigenous communities a very interesting subject of research, as it highly relevant in international business.

Conflicts between MNEs and indigenous communities are a more modern problem than generally thought. Jurna (2017) illustrates the conflict between the Wajapi Indians and Brazilian miners, who legally and illegally search for gold on the Wajapi’s traditional land. Due to Brazil’s financial crisis, the government is more willing to let mining companies extract natural resources such as copper, iron, magnesium, and gold. The global media platform Viceland (2016) has made an extensive documentary about the conflicts several indigenous communities have with MNEs. A few examples include: the Apache Nation (USA) fighting to stop the transfer of sacred land to a mining company, an indigenous community in Hawaii resisting a telescope being placed on a sacred mountain; and the Krenak people (Brazil) fighting against a toxic spill that contaminated their river.

(9)

9

There seems to be a vast amount of literature on concepts like MNE experience, indigenous communities, isolation, and conflicts (Hilson, 2001; Bruijn & Whiteman, 2010; Johanson & Vahlne, 2009; Vermeulen & Barkema, 2002; Cohen & Levinthal, 1990; Benito & Gripsud, 1992; Zeng, Shenkar, Lee and Song., 2013; Calvano, 2008; Whiteman, 2009; Lertzman & Vredenburg, 2005; Davis & Franks, 2011; Peters & Andersen, 2013; Cornell, 2002; Harlow et al., 1965). These studies all adopt a single-case perspective, thus omitting the broader picture of the firm- and community-level drivers behind these conflicts on a global scale. However, these cases have never been researched in an extensive, quantitative study that combines multiple concepts. This is the first gap in the literature that I identified. Therefore, I propose the first research question:

RQ1: To what extent does MNE experience influence conflict resolution (i.e. length and conflict type) with indigenous communities?

According to several studies, MNEs expand in a step-by-step manner. This means that each sequential FDI is made in a more distant culture or country (Johanson & Vahlne, 2009). Furthermore, according to Cohen and Levinthal (1990, in Vermeulen & Barkema, 2002), an MNE needs time to absorb experiences, this concept is called absorptive capacity. The absorbed experiences can be applied to future situations. Therefore, I expect there to be negative relationships between MNE experience and both conflict length and conflict type. This means that the more experience an MNE has in dealing with indigenous communities, both in the number of projects and number of years, the shorter and less violent a possible conflict will be.

This thesis examines MNE expansion on two levels. The first level represents studies which take into account the experience of an MNE in doing business abroad (Johanson & Vahlne, 2009; Benito & Gripsud, 1992; Vermeulen & Barkema, 2002). Secondly, a body of literature studies regarding the isolation of indigenous communities and how this can impact their behaviour is presented (Altman, 2004; Gupta & Ferguson, 1992; Hilson, 2001; Cornell,

(10)

10

2002; Harlow et al., 1965). Until now, it is unclear whether geographic isolation and cultural isolation moderate the relationship between MNE experience and conflict length and conflict type. These two levels of analysis have not yet been combined in a single study. By going through the previous literature, I expect that both levels are intertwined. Therefore, it makes sense to check whether geographic and cultural isolation moderate the relationship between MNE experience and conflict length and type. A second and crucial question arises:

RQ2: How is the effect of RQ1 moderated by geographic isolation and cultural isolation?

The goal of this second research question is to check for a possible moderating effect of geographic isolation and cultural isolation on the relationship tested in the first research question. Gupta and Ferguson (1992) state that when the isolation of a community increases, it becomes more culturally distant. Isolated communities will be dissimilar to communities with which an MNE possibly has experience; therefore, its experience might be not relevant. The more geographically isolated a community is, the higher the chance that a MNE would not know how to communicate and act with this community (Huertas Castillo, 2004). This could possibly lead to difficult communication between MNEs and indigenous communities, which is one of the most crucial factors in conflicts (Hilson, 2001)

I have been provided with a cross-sectional dataset, which consists of 706 individual cases. For my research, all necessary variables are present in the provided dataset. My final dataset consists of 544 individual cases from 92 countries. A total of 162 cases have been removed from the dataset because they were incomplete.

The independent variable is MNE experience. The dependent variables are conflict length and conflict type. The moderator variables are the geographic isolation and cultural

(11)

11

isolation of indigenous communities. The statistical analysis has been controlled by three control variables on three levels: country level, firm level, and community level.

Initially, I ran a test to analyse the mean, standard deviation, and correlation of the dependent, independent and moderating variables. Subsequently, I checked for multicollinearity among the same variables. Finally, I used a binary logistic regression for the dependent variable, conflict length. Both measures for conflict type (type of violence and degree of violence) were tested by a linear regression. I expected to find multiple, significant relationships between the independent variable and dependent variables. Furthermore, I expected to find several, significant moderating effects by testing the interaction effect of the moderating variables.

This research has methodological, theoretical, and practical contributions to the status quo. Methodologically, this is the first extensive, quantitative research study executed regarding the effect of MNE experience in conflict situations with indigenous communities Theoretically, this study adds to current literature by combining multiple aspects: MNE experience, conflict length, and conflict severity. Since this primary relationship has not yet been researched, neither has the possible moderating effect of geographic isolation and cultural isolation. Practically, this research contributes to the three main stakeholders in conflict situations. This information can serve all stakeholders involved in conflicts between MNEs and indigenous communities, and can help stakeholders avoid or resolve conflicts effectively.

This thesis is structured as follows. It begins with a literature review examining conflicts that arose between MNEs and indigenous communities, and how MNE experience could possibly influence the process of conflicts arising. Furthermore, I analyse literature on the geographical isolation and cultural isolation of indigenous communities. Then, a theoretical framework that includes the research questions and hypotheses is presented. Subsequently, a conceptual model is provided and presented visually. In the research design, I elaborate on the

(12)

12

statistical procedures in this thesis, after which I present the results and analysis. Finally, the discussion reviews the findings, theoretical implications, practical implications, limitations, future research possibilities, and the conclusion of this thesis.

(13)

13

Literature review

In this chapter, the current literature and concepts relevant to MNEs doing business with indigenous cultures are critically reviewed and portrayed. The concept of indigenous communities is discussed, followed by MNE experience. Indigenous conflicts are then examined more closely. Finally, the literature concerning geographical isolation and cultural isolation is reviewed.

Indigenous communities

This subchapter gives several interpretations of what an indigenous community is. There is a substantial amount of literature on the concept of ‘indigenous communities’ and its definition (UN, ILO). For MNEs to know how to deal with indigenous communities as local stakeholders, they must be aware of what indigenous communities entail. According to the United Nations (UN) (n.d), there are over 370 million people who can be classified as ‘indigenous’ living in over 70 countries around the world. Indigenous communities have unique traditions and are significantly different from the majority of people in the societies in which they live in terms of their cultural, social, economic, and political characteristics. Indigenous cultures have unique languages, knowledge systems, and beliefs. Indigenous people are descended from a certain group of people who lived on a certain place on the earth. The UN (n.d.) has defined seven possible characteristics of indigenous communities; however, considering the diversity of indigenous communities there is no official definition. Indigenous people should identify themselves individually and should be accepted by the indigenous community. Second, there should be a historical continuity with pre-colonial societies and/or pre-settler societies. Third, indigenous peoples’ strong link to territories and local natural resources is mentioned (UN, n.d.). An indigenous community can never be a dominant group of society. One of their goals is to maintain and reproduce ancestral environments as an

(14)

14

independent people and community. The International Labour Organisation (ILO) has a definition for ‘indigenous’ and ‘tribal,’ which was presented during the indigenous and tribal peoples convention in 1989. The ILO proposes a twofold definition. The first part describes the living environment of indigenous communities as socially, culturally, and economically different from other parts of the national community.

An indigenous community is substantially created by its own customs, traditions, laws, and regulations (ILO, 1991). The second part describes the communities retaining their own social, economic, cultural, and political institutions at the time of conquest, colonisation, or establishment of present state boundaries (ILO, 1992). According to Epps and Brett (2000, in Hilson, 2001), indigenous people are among the poorest of a country’s population. They earn their money through local economic activities such as farming, forestry, and fishing. Indigenous communities are not only dependent on their land for economic activities, but also this land holds great cultural value as their ancestors also lived on it. This makes it unrealistic to assume that indigenous communities can be paid to leave a certain location. Indigenous groups do not classify themselves as a stakeholder in the MNE, but rather see themselves as ‘relatives’ of the land (Whiteman, 2009, in Bruijn & Whiteman, 2010). So far, there is very little empirical evidence available on how MNEs actions affect how indigenous communities define themselves. Even more importantly, there is very little empirical evidence on how an organization’s actions contribute to conflicts (Bruijn & Whiteman, 2010). The word 'indigenous' is a generally accepted and well-known term. However, some people might prefer other terms such as: tribes, first peoples, aboriginals, hunter-gatherers, nomads, peasants, and hill people (UN, n.d.).

(15)

15

MNE experience

The factor, MNE experience, is broadly discussed in business literature, for example on business expansion and learning processes. There are a few theories and concepts I would like to address concerning MNE experience as well as previous research on this topic. To state, I would like to begin with literature on MNE expansion. According to the Uppsala model (Johanson & Vahlne, 2009), the internationalisation process is a cumulative and path-dependent process primarily based on international experience and knowledge. The Uppsala model indicates that an MNE expands first to places that are more similar geographically, institutionally, and culturally and then to places that are more distant. This theory complies with the absorptive capacity theory of Cohen and Levinthal (1990, in Vermeulen & Barkema, 2002), which states that a company needs time to absorb experiences and use these experience in further developments.

This view of expansion by way of experience is contradicted by Benito and Gripsud (1992), who argue that each FDI decision is made by a MNE on a case-by-case basis and that a FDI is not made in locations necessarily closer to the home country. Zeng et al. (2013) present multiple arguments both in favour of and against MNE experience. The authors provide empirical evidence that MNEs can learn incorrectly, leading to experience that is possibly not beneficial for future FDI. Host-culture experience only becomes beneficial when a MNE has a certain amount of this experience. Until now, there is no conclusion on whether MNE experience helps in dealing with cultural differences or whether it influences MNEs’ FDI decisions. Furthermore, there are no published articles and only one student thesis available, which discusses the relationship between MNE experience and conflict with indigenous communities. The outcome of this previous thesis indicates no significant relationship between MNE experience and conflict with indigenous communities. However, there are some aspects of the paper that make this outcome open to debate.

(16)

16

Indigenous conflicts

Pondy (1967, in Calvano, 2008) describes a conflict between an MNE and a community as a dynamic process that unfolds over time, rather than a discrete event. Calvano (2008) describes a conflict as a situation in which inequalities exist between MNEs and communities. In such a conflict, the indigenous community will always be at a disadvantage and should ‘hope for the best’. On the other hand, MNEs can decrease the threat of a potentially violent conflict by devoting time and money to prevention. The author concludes that the solution for conflicts between MNEs and communities is two-fold: MNE management should be more aware of community perspectives, and communities should increase their power to challenge corporate actions (Calvano, 2008). Since the number of conflicts with indigenous communities has increased, these communities are now considered important local stakeholders by MNEs and other stakeholders (Bruijn & Whiteman, 2010). In 2009, Shell made the statement:

‘We work with our Aboriginal neighbours to build relationships based on trust and mutual understanding. We benefit Aboriginal communities through a variety of local investments and national programs. We seek to understand the views of Aboriginal people to gain from their traditional environmental knowledge by helping to preserve it and integrating it into our business plans’ (2009, in Bruijn & Whiteman, 2010, p. 479). According to Banerjee, Calvano and Whiteman (2000; 2008; 2009, in Bruijn & Whiteman), natural resource development in a situation where an indigenous community is a stakeholder typically leads to power imbalances. According to Hilson (2001), most unavoidable environmental problems that occur at mining locations, including erosion, sedimentation, and vegetation removal, are mostly dismissed by the local community. The root causes of conflicts between MNEs and the local community stem from poor communication and highly preventable environmental accidents. MNE business developments, like natural resource development on indigenous communities’ lands, are often perceived as fundamentally unjust, despite corporate codes on corporate social responsibility (CSR), stakeholder relation programs, and community consultation (Whiteman, 2009). Subsequently, Lertzman and

(17)

17

Vredenburg (2005) have identified several factors that should be considered when interacting with indigenous peoples. The indigenous community should be represented by a person who is recognized by the entire community, not a consultant speaking on behalf of the indigenous community. Furthermore, to ensure efficient interaction between the two parties, the representatives of both parties must be culturally literate. When a company fails to build a sustainable relationship with a local community, the company’s performance suffers, for example because of supply chain problems and an unreliable workforce (Davis & Franks, 2011). So far, research on conflicts between MNEs and communities focuses primarily on community responses to negative outcomes of MNE activities, instead of on the dynamics of the relationship between MNEs and communities (Calvano, 2008).

Geographical and cultural isolation

A common misconception is that indigenous people always live in isolated locations with both a large geographic and cultural distance, such as places as the rainforest in South America, the Sahara Desert, or up high in the Himalayas. However, indigenous people do not necessarily live in remote locations. Altman (2004) makes a distinction between metropolitan, urban situations and remote, regional situations. Altman describes situations in which cultures come into conflict with stakeholders from a totally different culture. The article describes that the majority of conflicts may be caused by conflicting cultures. Dowie (2011) states that opposition arises when cultures are different from one another. More specifically, Dowie (2011) states that diverting cultures can be a catalyser for conflicts. In many developed countries, most indigenous people live in urban areas (Peters & Andersen, 2013). Increasing isolation of communities lead to a larger cultural distance (Gupta & Ferguson, 1992). According to Bebbington (1999), communities that are geographically isolated generally have poor access to resources, like labour, markets, skills, credit, and social organisation. This lack of access is a significant issue for indigenous communities when dealing with MNEs, as resources give the

(18)

18

community bargaining power (Bebbington, 1999). It is hard to come into contact with geographically isolated communities on a mutual level of understanding and respect. One could state that the more isolated a community is, the lower their bargaining power. This makes it more difficult for a MNE and community to engage with each other and come to a quick, nonviolent agreement. This view is supported by Huertas Castillo (2004), who states that the more geographically isolated a community is, the higher the chance that a MNE does not know how to communicate and deal with it properly. Communication problems between indigenous communities and MNEs are one of the most crucial factors when it comes to conflicts (Hilson, 2001). Furthermore, Cornell (2002) states that minority communities that are granted autonomy have a higher chance of acting in aggressive manners, thus indicating a higher chance of conflict. This view is shared by Harlow et al. (1965), who state that isolation leads to hostile views and behaviours toward the people that cause isolation. A community is culturally isolated when it strongly differs from its largest, nearby society on multiple aspects, such as language, norms, and values.

(19)

19

Theoretical framework

This thesis involves theory-driven research. Thus, I based my research questions and hypotheses on the current, relevant literature in the field.

MNE experience and conflict situations

Results of theories and studies contradict each other regarding the value of experience in international expansion. On the one hand, there is the Uppsala model (Johanson & Vahlne, 2009), which describes a step-by-step process of expansion. The absorptive capacity theory by Cohen and Levinthal (1990, in Vermeulen & Barkema, 2002) complies with the Uppsala model in stating that a company needs time to absorb and use experience in further developments.

On the other hand, Benito and Gripsud (1992) state that each FDI decision is made individually and that former experience or cultural distance is not significant. Johanson and Vahlne’s view can be described as an internationalization process, whereas mainstream economic theory (Hirsch, 1976) treats individual investment decisions as discrete phenomena. Benito and Gripsud’s (1992) view fits with mainstream economic theory. In their view, a lack of experience is described as an extra cost in controlling FDIs. As argued by the internationalization theory, the experience of an MNE increases over time and with a greater number of expansions. The MNE gains experience in countries and with cultures that are relatively similar in culture and geography to the MNE’s home country. Over time, the MNE is able to expand to more distant countries because of the experience the MNE has gathered throughout its incremental internationalization process. This might very well influence possible conflict with an indigenous community. Hilson (2001) states that poor communication between MNEs and local communities is the root cause of conflict. Other crucial factors to avoiding or mitigating conflicts are described by Lertzman and Vredenburg (2005). A MNE’s communication skills and knowledge (Hilson, 2001) and other factors (Lertzman &

(20)

20

Vredenburg, 2005) might increase the MNE’s experience, thus impacting the type of conflict and its length.

MNE experience, conflicts and isolation

Both geographical and cultural isolation have an impact on communities. Increasing isolation of communities leads to larger cultural distances (Gupta & Ferguson, 1992). Furthermore, Cornell (2002) states that minority communities that are granted autonomy have a higher chance of acting in aggressive manners, which indicates a higher chance of conflict. This view is shared by Harlow et al. (1965), who state that isolation leads to hostile views and behaviours toward the people that cause isolation. The isolation of communities might influence the relationship between MNE experience and conflict situations. Isolated communities can be different in such a way that renders MNE experience irrelevant. An example is communication between the MNE and an indigenous community: when the method of communication in an isolated community is not similar to that of communities the MNE dealt with previously, this past experience will not help the MNE.

Research questions

Current literature fails to describe the relationship between MNE experience and conflict situations with indigenous communities. These concepts have been researched independently, but not together until now. A second gap I identified in the literature is the lack of knowledge about the possible effect of the isolation of indigenous communities, both geographically and culturally, on possible conflicts with MNEs. From these specific and distinct concepts, I proposed two research questions. The first research question examines a possible relationship between MNE experience, in both years of experience and in number of projects, and the duration and type of conflict. This question is followed by a second research question,

(21)

21

which considers the moderating effect of geographic and cultural isolation on the relationship described in the first question.

A total of six hypotheses are tested in this thesis. A conceptual model that illustrates the hypotheses can be found in Figure 1. It is important to state that both MNE experience and conflict type each have two measurements. Since adding more hypotheses makes the overview less clear, I decided not to include these in the hypotheses, nor in the conceptual model. For both MNE experience and conflict type, adding the second measurement to the hypotheses and conceptual model does not make a difference in the outcome of the research. Chapter 7 offers a more detailed overview of the variables I refer to. In this chapter, each hypothesis is elaborated on in three parts. First, previous studies the relate to the hypothesis are discussed. Secondly, the core argument of why variables are related is justified. Lastly, the core argument is illustrated with an example from another study.

H1: MNE experience has a negative influence on conflict length.

I hypothesize that MNE experience, both in number of conflicts and in years of experience, is a predictor of conflict length. Previous studies have made several claims on this topic. According to Furlong (2005), building a relationship contributes to establishing peace. Furthermore, and more importantly, the author describes relationship building as a very important factor in reaching a solution in a conflict situation. By enhancing the communication competences of a MNE toward a community, the tension will drop and the MNE’s corporate image will be improved (Asawo, 2011). The importance of communication has also been described by Hilson (2001). Experience in communicating with indigenous communities increases the chances that a MNE will successfully work toward a healthy relationship with the indigenous community. However, it is important that both parties are willing to participate in the process of learning in relation to managing resources (Berkes, 2009). Berkes (2009) concludes that joint problem solving is a must in resource management; this only takes place

(22)

22

when both the MNE and the indigenous community share experiences and ideas. Therefore, it is expected that the better the communication and interaction between the MNE and indigenous community, the shorter a conflict will be. This shorter duration will be achieved through the MNE’s experience in dealing with indigenous communities.

H2: MNE experience has a negative influence on conflict type.

The second hypothesis states that MNE experience, both in number of conflicts and in years of experience, is a predictor of conflict type. I hypothesize that the more experience a MNE has, the lower the level of violence will be in a potential conflict with an indigenous community. Kemp, Owen, Gotzmann, and Bond (2011) describe the resource extraction industry and the most common drivers of conflicts in this industry, that is the unequal distribution of risks, impacts, and benefits. The article states that MNEs find it difficult to translate high-level CSR policy into practice during the operation. According to Laplante and Spears (2008), companies that sustain unpopular regimes or sustain conflicts by funding are likely to become the target of violence. When communities have participated equally in the decision-making process throughout the project, they are more likely to accept the project. This co-management between a MNE and community gives the community the chance to peacefully discuss issues, eliminating the need to take other actions such as violence, blockades, or others. Insufficient public participation and inadequate regulation can lead to frustration and violence from the communities. Laplante and Spears (2008) conclude that including the community in the decision-making process reduces the most common reason for communities to use any type of violence. Lertzman and Vredenburg (2005) state that cultural literacy can be learned and can help facilitate successful interactions between a MNE and an indigenous community. Cohen and Levinthal (1990, in Vermeulen & Barkema, 2002) state an MNE needs time to absorb experiences and use these experience in further developments (absorptive capacity). Therefore, the experience of a MNE affects its communication and participation capabilities, which in turn

(23)

23

affect the degree of violence of a possible conflict. Calvano (2008) states that previous studies about MNE and community conflicts primarily focus on the responses of the community to negative outcomes of MNE activities, instead of focusing on the dynamics of the relationship between MNEs and indigenous communities. The dynamics of the relationship between MNEs and indigenous communities are expected to be influenced by MNE experience and subsequently decrease the conflict type.

H3: Geographic isolation will negatively moderate the influence of experience on conflict length.

H4: Geographic isolation will negatively moderate the influence of experience on conflict type.

According to Pringle (2014), communities specifically choose geographical isolation because of their anxiety about modern influences on their ancient and traditional ways of living. Communities are influenced by modern technologies and ways of living, which do not fit into their traditional ways of living. Furthermore, these indigenous communities fear unknown diseases, which could threaten the existence of the community. Geographical isolation is very important when there is the desire to preserve traditional cultural elements. Huertas Castillo (2004) states that the more geographically isolated a community is, the higher the chance that a MNE does not know how to communicate and act with this community; this could have consequences that relate to conflicts. Tang and Tang (2001) describe a situation in which a geographically isolated community allows outsiders to enter the area. In the original situation, the community caught fish in way that sustained the fish population for future generations. Outsiders came to the area with new fishing techniques. This new technique resulted in a rapid decrease in the amount of fish in the waters. Confrontation on this matter could have possibly led to a violent conflict. When there is an encounter between an isolated indigenous community and another less isolated party, as described in articles by High (2013) and Lawler (2015), there

(24)

24

is change of the conflict becoming violent. Geographically isolated communities are hard places to reach for MNE representatives. This makes it hard to establish the right communication and consent, which could possibly lead to a long-lasting violent conflict.

H5: Cultural isolation will negatively moderate the influence of experience on conflict length.

H6: Cultural isolation will negatively moderate the influence of experience on conflict type

Cultural isolation is closely connected to geographical isolation. Cultural isolation can be a result of geographical isolation. Geographical isolation makes it harder for indigenous communities to become familiar with other communities’ languages, norms, and values. This argument also works vice versa: it is hard to understand the cultures of communities that are hard to reach. According to Dowie (2011), opposition arises when cultures are different from one another. It is even stated that diverting cultures can function as a catalyser for conflicts. This has been shown in the example of the fishing village by Tang and Tang (2011), where outsiders were not able to understand the indigenous community’s way of thinking. Shivakumar (2002) states that culture provides a certain amount of direction for a community. However, when this culture is unique, conflicts may arise from cultural differences between the isolated community and other parties. Harlow et al. (1965) study primates in isolation and made some interesting discoveries. First, social isolation led to aggressive and violent behaviour. Furthermore, social isolation might cause fear of the unfamiliar. Another study by Cacioppo and Hawkley (2009) links social isolation to an increased sensitivity to social threats.

(25)

25

Conceptual model

The hypotheses described in the previous part of this thesis have led to a simplified conceptual model, which can be found below. A '-' represents a predicted negative

relationship or moderating effect.

(26)

26

Research Design

Sample and data collection

A previous thesis determined that there was no significant relationship between MNE experience and conflict length and type. This has to do with the fact that this thesis only used cases from Africa and Latin America, with 68 cases in total. Therefore, I use a data set that is as large as possible. All the data for the variables described in the conceptual have been made available by my supervisor. A total of 706 cases were available. However, some cases were not complete or had variables that must be recoded. A total 162 cases were removed from the dataset due to incompletion. My final dataset consisted of 544 cases that took place in 92 countries worldwide.

Variable recoding

I was asked by my supervisor to recode the variable, conflict length (interval), as either short or long. The precise point at which a conflict becomes either short or long was decided on by checking the median of conflict length of ended cases, both in the total sample of 706 cases (median 48 months) and the final sample of 544 cases (median 46 months). After discussing these medians with my supervisor, we agreed on a tipping point of 46 months. I realise this reduces variation and that this is a limitation of this study. MNE experience in years also needed to be recoded, since only the actual year of the first conflict with an indigenous community was provided in the dataset. Therefore, I made a new variable that displays the absolute number years of experience. The dataset was most recently updated in mid-2016; hence the new variable was calculated as 2016.5 minus the year of the MNE first experiencing a conflict with an indigenous community. I recoded the control variable, country. In the dataset provided, this variable was presented as actual country names, I coded each country with

(27)

27

number. In order to be able to check for moderation, a total of 4 interaction variables were created.

Dependent variables

The dependent variables in this research were conflict length and conflict type. According to Getz and Oetzel (2009), conflicts can either be identified as short or long. The length of the conflict is measured by comparing the date (month and year) of the start of the conflict with the end date of the conflict, resulting in a total number of months. This thesis used Getz and Oetzel’s method. Conflict length is an interval variable (UCLA, n.d.). In order to define the on-going cases in the data set, the median of the ended conflicts was determined. The decision on where the tipping point would be is based on the median of 46 months. All cases, including the on-going cases, were recoded. The on-going cases that lasted longer than 46 months were coded as long conflicts. In order to run the statistical test, a new variable was created, where conflict length was either short or long.

Conflict type is the second dependent variable and has two measurements: type of violence and degree of violence. The measurement, type of violence, is coded as an ordinal variable (UCLA, n.d.). A conflict is coded as having a certain type of violence: low-level violence, court actions, mid-level violence, high-level violence, or maximum-level violence. The conflict concerning the Kalgoorlie project in Australia between the Barrick Gold Corporation and the indigenous Aboriginal community of Ninga Mia was a low violence conflict. The indigenous community peacefully protested against further developing the mine (Andrusiak & Taylor, 2008). A conflict in which court actions were filed is the Mande Norte project in Colombia. The conflict between Rio Tinto (company), the Embera indigenous peoples, and the Afro-Colombian community of Jiguamiando was caused by the development of mineral ores and building of an extraction mine (Environmental Justice Atlas, 2014). An

(28)

28

example of a mid-level violence conflict is the Bayano Dam conflict in Panama between AES Corporation and the Madungandi Kuna's indigenous people. The construction of this hydro project required the entire village to relocate. The indigenous community tried to stop this development by protesting. However, their actions led to human rights violations and state-sponsored violence (Finley-Brook & Thomas, 2010). A far more extreme case is the high-level violence at the Xolobeni mine in South Africa between Mineral Commodities Ltd. and the AmaMpondo people. In order to close one of the world’s largest mineral ore mines, the indigenous community protested and filed lawsuits. This resulted in a high-level violence case (Environmental Justice Atlas, 2016). A conflict with maximum-level violence occurred in the Philippines at the Marinduque project between Placer Dome and the Marinduque indigenous community. The Marcopper Placer Dome mining disaster is one of the largest environmental disasters in the Philippines’ history and resulted in flooded villages, toxic rivers, ill residents, and dying children. Currently, the fight for justice is ongoing (Environmental Justice Atlas, 2015).

The second measurement, degree of violence, is coded in the same way as the other violence variable. The variable, degree of violence, is an ordinal variable (UCLA, n.d.). The degree of violence is either: no degree of violence, destruction, occupation, road block, kidnapping, injury, or, at the most extreme, death. Conflicts between MNEs and indigenous communities are perceived to be most severe when there is an outbreak of violence; this occurs when the intensity of the conflict is rather high (Getz & Oetzel, 2009). These conflicts can differ in severity, including violence, wars, human rights violation, or even death (Whiteman & Mamen, 2002). An example of a conflict with no degree of violence is the one between G4 Industries Limited and the Giriama indigenous community in Kenya. The board of G4 Industries Limited decided to pull out of a biofuel project when increasing environmental evidence revealed the project would destroy a wetland ecosystem that is critical to regional

(29)

29

wildlife (Cernansky, 2011). An example of a destructive conflict is the one between Jan Mangal Uganda Ltd. and the Karamojong indigenous community in Uganda. In this case, mineral resource exploitation led to drought, land disputes, and land dispossession. This left the community deprived of water, food, healthcare, and education (Datzberger & Malagala, 2015). A case of occupation is the Eunapolis project between Veracel Celulose and the Pataxo indigenous community in Brazil. The traditional territory of the Pataxo indigenous community was taken from the community and then they took it back again (World Rainforest Movement, 2000). Road blocks are a common phenomenon in conflicts between MNEs and indigenous communities. This occurred at the Dayuma Oil Extraction project in Ecuador between Petroecuador and several indigenous communities, including the Quechuas, Shuar, and Huaoranis. In order to demand a fairer distribution of oil resources, the indigenous communities blocked the Aucas oil street (Environmental Justice Atlas, 2014). More severe cases include kidnapping and injuries. This occurred at the Inga Hydropower Project in Congo between ESKOM and the Makhuku Vundi and Makhuku Manzi indigenous communities. This project could potentially lead to floods, water pollution, decreasing water quality, and other health-related issues (Environmental Justice Atlas, 2016). The most extreme degree of violence is death. An example of a conflict that involved death by violence was the Bioko Island case in Equatorial Guinea between the Exxon Mobile Corporation and the Bubi indigenous community (Human Rights Watch, 2009).

Independent variables

The independent variable in this research is MNEs’ experience in operating in locations where they deal with indigenous communities. I decided to use two measurements for MNE experience. The first measure of MNE experience was to count the number of projects around the world in community territory or affecting communities since start of operations. This measurement is an interval variable, more specifically a ratio variable (Leard Statistics, n.d).

(30)

30

The second measurement of MNE experience was the year in which first project with an indigenous community started. In order to make the variable suitable for statistical analysis, it was recoded into years of experience instead of the first year of a relevant project. This measurement is an interval variable. According to Davis and Franks (2011), MNE experience can be considered a predictor of conflict. Their article states that MNEs’ knowledge on how to communicate with indigenous communities is one of the most crucial factors in preventing or dealing with conflicts.

Control variables

The internal validity of this research was controlled by three control variables. The three control variables served to eliminate external factors, meaning that the results of the research can be linked directly to the research variables. Each of the three control variables exists on a different level: the country level, firm level, and community level. On the country level, the control variable, country, was used. The variable, country, tells the country in which the conflict took place. On the firm level, the MNE experience in specific country was used as a control variable. On the community level, the community population was used as a control variable.

Moderating variables

A moderating variable can be described as a variable that modifies the strength or direction of a causal relationship between a dependent variable and an independent variable. This moderating effect is also known as the ‘interaction effect’ (Wu & Zumbo, 2007). As moderating variables, I decided to research the potential moderating effect of the geographic isolation and cultural isolation of communities with whom MNEs do business. The more isolated a community is, the larger its cultural distance becomes (Gupta & Ferguson, 1992). Furthermore, minority communities with high levels of autonomy have a higher chance of acting in aggressive manners, which practically means that there is a higher chance of conflict.

(31)

31

It is hypothesised that the more isolated a community is, both culturally and geographically, the less influence MNE experience has on decreasing the conflict length and degree of violence. The geographic isolation of the community was coded as follows: 1. Highly isolated: i.e. community members live within the community with little contact with the outside world. 2. Most community members live within broader municipalities with mixed population. The cultural isolation of the community was coded as follows: 1. Highly isolated: i.e. the community is from an isolated culture, has a different language, and is not connected to a broader ethnic group. 2. Community connected to a broader ethnic group, but isolated within the group. 3. Community embedded within a broader, mainstream ethnic group.

Method and analysis

In order to analyse the data for this thesis, the statistical software SPSS was used. Multiple tests were performed in order to test the effect of the independent variable on the dependent variables and to test the effect of the moderating variable on the dependent variables. The data was structured as a cross-sectional dataset with 544 cases. Cross-sectional data is data about different aggregates in the same time period (University of Washington, n.d.). Due to the fact that this research had multiple dependent variables, multiple separate regression models were required. In order to check for a relationship between the dependent variables and the independent variable, I used a binary logistic regression for the dependent variable, conflict length, and a linear regression for the dependent variables, type of violence, and degree of violence (Institute for Digital Research and Education, n.d.). For all three dependent variables, I first checked for a significant, direct effect, after which the moderators were separately added. Subsequently, the interaction between the independent variable and the moderator were added (Jaccard, Wan & Turrisi, 1990).

(32)

32

The regression models are presented in three tables, with one table for each dependent variable. These models can be found subsequently in Table 1, Table 2 and Table 3.

Table 1: Regression on Length of conflict

Table 2: Regression on Type of violence

(33)

33

Results and analysis

In this chapter the results of the statistical analysis will be presented. The structure is as follows, the first chapter presents the descriptive statistics. Next the data set will be tested for multicollinearity. Finally, the regression analysis will be described. By using these statistical outcomes my hypotheses will be evaluated and either be accepted or rejected.

Descriptive statistics

Table 4 shows the descriptive statistics and correlations of the dependent variables, independent variables and the moderator variables. The dependent variable type of violence has a mean of 2.820 and a standard deviation of 1.475. This means the average type of violence of a conflict is higher than peaceful protest which is coded as 2. The dependent variable degree of violence has a mean of 2.540 and a standard deviation of 1.991. Therefore, it can be concluded that the average degree of violence is higher than injuries which is coded as 2. The third dependent variable, conflict length, has a mean of 0.690 and a standard deviation of 0.464. Conflicts which last 45 months or shorter are coded 0, conflicts which last 46 months or over are coded 1. Therefore, it can be stated that the average length of conflict is longer than 45 months. The independent variable experience in number of project has a mean of 31.100 and a standard deviation of 39.296. The other independent variable, experience in years, has a mean of 26.996 and a standard deviation of 26.274. The moderator geographical isolation has a mean of 1.650 and a standard deviation of 0.477. This mean score tells us that the average indigenous community in this data set lives within broader municipalities with mixed population. The second moderator variable, cultural isolation, has a mean of 1.920 and a standard deviation of 0.651. Therefore, it can be stated that the average indigenous community in this data set is connected to a broader ethnic group, but isolated within the group.

(34)

34

Correlation

Besides the descriptive analysis in table 4, the correlation between the variables is shown. Correlation is a very common statistic and shows a coefficient which describes the relationship between two separate variables (Web Center For Social Research Methods, 2006). This subchapter will describe the significant correlations between the dependent variables, independent variables and the moderator variables. First, there is a significant, positive correlation between degree of violence and type of violence. This makes sense since it are two different measures for the same dependent variable, conflict severity. Furthermore, there is a significant, positive correlation between conflict length and type of violence, this indicates that if longer the conflict lasts the more violent a conflict is. Thirdly there is a significant, positive correlation between conflict length and degree of violence, this indicates that if longer the conflict lasts the more violent a conflict is. However, this correlation is weaker than the correlation between conflict length and type of violence. Looking at the first measure of the independent variable, there is a significant negative correlation between MNE experience in number of projects and conflict length, this indicates that if a MNE has relatively more experience in doing projects with indigenous communities the conflict length will decrease. The second measure significantly correlates with two independent variables; MNE experience in number of years significantly and negatively moderates with degree of violence and with conflict length, this indicates that MNEs with a long history of dealing with indigenous communities have relatively less violent conflicts. A third significant correlation is a positive correlation with the first measure of the independent variable; MNE experience in number of projects. This makes sense since it are two different measures for the same dependent variable, MNE experience.

Before stating any indications regarding the significant correlations between moderators and other variables the coding of the moderators has to be specified. The coding of both

(35)

35

moderators is the other way around: 1 being very isolated, 2 less isolated, etc. Therefore, positive correlations have a negative effect and the other way around. In this chapter I use the outcomes from the correlation table and indicate its meaning. There is a significant, negative correlation between geographical isolation and type of violence and degree of violence, this indicates that that the less geographically isolated an indigenous community is the lower the type- and degree of violence When looking at the correlation table one sees a lot of significant correlations for the moderator cultural isolation. There is a significant, negative correlation between cultural isolation and type of violence, this indicates that that the less culturally isolated an indigenous community is the lower the type of violence. There is a significant, negative correlation between cultural isolation and degree of violence, this indicates that that the less culturally isolated an indigenous community is the lower the degree of violence. There is a significant, negative correlation between cultural isolation and conflict length, this indicates that that the less culturally isolated an indigenous community is the shorter the duration of the conflict. Furthermore, there is a significant, positive correlation between cultural isolation and MNE experience in years, this indicates that that the less culturally isolated an indigenous community is the more experience an MNE has is conflicts. Lastly there is a significant, positive correlation between cultural isolation and geographic isolation¸ this indicates that that the less culturally isolated an indigenous community is the less geographically isolated an indigenous community is. The significant correlations presented for the moderators add to the information found in the literature regarding these two moderators.

(36)

36

(37)

37

Multicollinearity

In order to rule out multicollinearity and increase the reliability of the analysis a test has to be performed. Multicollinearity is present when two or more variables are strongly correlated, in other words have a strong linear relationship (Field, 2009). A first method to test for multicollinearity is by checking for correlations which are very strong, values between 0.8 and 1. In the correlation table the highest correlation to be found is 0.715, which is the correlation score for degree of violence and type of violence. A second test to check for multicollinearity is the Variance Inflation Factor (VIF). When the VIF score is under 10 and the tolerance score is above 0.2 there exists no multicollinearity (Field, 2009). Looking at table 5, which has conflict length as dependent variable, all VIF scores are between 1 and 1.4, the tolerance scores are all above 0.7. This indicates that there are no signs of multicollinearity in Table 1. For table 6, which has type of violence as dependent variable, all VIF scores are between 1 and 1.4, the tolerance scores are all above 0.7. This indicates that there are no signs of multicollinearity in Table 2. For table 7, which has degree of violence as dependent variable, all VIF scores are between 1 and 1.4, the tolerance scores are all above 0.7. This indicates that there are no signs of multicollinearity in table 3. There are no multicollinearity problems with the data which will be used for the regression analysis.

(38)

38

Regression analysis

According to Field (2009) the purpose of a regression analysis is to predict the behaviour of a dependent variable from one or multiple independent variables. This makes it possible to use the collected data in more ways and go beyond the normal statistical relations. A stepwise regression analysis is used in order to detect a relationship between the independent variables and the dependent variables. To be able to check for a possible moderation effect, four

(39)

39

interaction variables have been added to the dataset, two per independent variable. A total number of 21 models are run in order to test for significant results and to check whether the hypothesis can be supported. In the first regression analysis, where conflict length is the dependent variable, the dependent variable was recoded into a dummy variable with either being short or long. For the first dependent variable a binary logistic regression analysis is used since the dependent variable is categorical and the independent variables are continuous. For the dependent variable conflict type, a linear regression model is used. First the three control variables were used in the model, subsequently the independent variables and the moderators were added. Please find the results in tables 8 and 9.

As of the beginning of this research two measurements were used for the dependent variable conflict type, both type of violence and degree of violence. I did this to check for the measurement with the highest explanatory power. After running the linear regression analysis of both measurements, it is clear that type of violence is the measurement with the highest adjusted R2 scores and therefore has the highest explanatory power. Therefore, as of this moment the dependent variable conflict type is measured by type of violence. The linear regression of degree of violence, table 10, can be found in appendix 1.

The first hypothesis (H1) expects that MNE experience has a negative influence on conflict length. This means that the prediction is that the more experience an MNE has the shorter a possible conflict will be. The outcomes of the statistical analysis which are presented in table X show that MNE experience in number of projects is a significant predictor of conflict length (B -0.004, p<0.1). Therefore, H1 can be accepted.

The second hypothesis (H2) expects that MNE experience has a negative influence on conflict type. This means that the prediction is that the more experience an MNE has the less violent a possible conflict will be. In order to either accept or reject H2 the SPSS outcomes in table X have to be analysed. First looking at the dependent variable type of violence, it can be

(40)

40

stated that only MNE experience in number of years is a significant predictor of type of violence (B -0,092, p <0.05). This negative relation between MNE experience in years and type of violence would mean to accept H2. MNE experience, in number of projects, is not found to be a significant predictor of type of violence. Therefore, H2 can be accepted, however only MNE experience in number of years has the predicted significant, negative effect on type of violence.

The third hypothesis (H3) predicts that geographic isolation will negatively moderate the influence of MNE experience on conflict length. Hence meaning that the more geographically isolated an indigenous community is the longer a possible conflict will take. The results of the statistical test show a significant moderation effect (B -0.231, p<0.05). This is the moderation effect of geographic isolation on the effect of MNE experience in years on the dependent variable length of conflict. It seems to be a negative effect, however since the variable is coded as 1: highly isolated and 2: mixed population, the effect is positive instead of negative. Therefore, H3 is rejected.

The fourth hypothesis (H4) predicts that geographic isolation will negatively moderate the influence of MNE experience on conflict type. Hence meaning that the more geographically isolated an indigenous community is the more violent the conflict will be. For the dependent variable, type of violence, there is a significant moderation effect. The moderation effect is present as a significant moderator for MNE experience in number of projects (B 0.071, p<0.1) for the dependent variable, type of violence. It seems to be a positive effect, however since the moderator variable is coded as 1: highly isolated and 2: mixed population, the effect is negative instead of positive. Therefore, H4 is accepted. Geographic isolation does significantly negatively moderate the influence of MNE experience in number of projects on the dependent variable type of violence.

The fifth hypothesis (H5) predicts that cultural isolation will negatively moderate the influence of MNE experience on conflict length. This practically implies that the more

(41)

41

culturally isolated an indigenous community is the longer a possible conflict will last. For the dependent variable conflict length, there is a significant moderation effect (B 0.240, p<0.05). This is the moderation effect of cultural isolation on the relation between MNE experience in number of project and conflict length. It seems to be a positive effect, however since the variable is coded as 1: highly isolated, 2: connected to broader ethnic group and 3: embedded with broader mainstream ethnic group the effect is negative instead of positive. Therefore, H5 is accepted.

The sixth hypothesis (H6) predicts that cultural isolation will negatively moderate the influence of MNE experience on conflict type. Hence meaning that the more culturally isolated an indigenous community is the more violent the conflict will be. There is no significant moderation effect to be found. Cultural isolation does not significantly moderate the influence of MNE experience in number of projects nor MNE experience in number of years on type of violence. Therefore, H6 is rejected.

For the dependent variable length of conflict, the Nagelkerke R square is 0.051 when only adding the control variables, explaining 5.1% of variance in length of conflict. This increases to 5.4% when adding MNE experience in number of years of experience. For MNE experience in number of years adding only geographic isolation increases the Nagelkerke R square to 6.4%, adding only cultural isolated increases the percentage to 7.0%. Adding the interaction term of geographic isolation results in 6.3% and finally adding the interaction term of cultural isolation results in the highest explanatory percentage of 7.3%.

When only adding MNE experience in number of projects to the control variables this results in explaining 5.9% of variance in length of conflict. For MNE experience in number of projects adding only geographic isolation increases the Nagelkerke R square to 6.0%, adding only cultural isolated increases the percentage to 6.4%. Adding the interaction term of

(42)

42

geographic isolation results in 6.9% and finally adding the interaction term of cultural isolation results in the highest explanatory percentage of 5.4%.

For the dependent variable type of violence, the Adjusted R2 is 0.028 when only adding the control variables, explaining 2.8% of variance in type of violence. This increases to 3.4% when adding MNE experience in number of years of experience. For MNE experience in number of years adding only geographic isolation increases the Adjusted R2 to 4.3%, adding only cultural isolated increases the percentage to 6.7%. Adding the interaction term of geographic isolation results in 3.3% and finally adding the interaction term of cultural isolation results in the highest explanatory percentage of 3.2%.

When only adding MNE experience in number of projects to the control variables this results in explaining 2.7% of variance in type of violence. For MNE experience in number of projects adding only geographic isolation increases the Adjusted R2 to 3.7%, adding only cultural isolated increases the percentage to 6.6%. Adding the interaction term of geographic isolation results in 3.1% and finally adding the interaction term of cultural isolation results in the highest explanatory percentage of 2.9%.

Reviewing the Nagelkerke R square and Adjusted R2 scores there are some interesting observations. Overall the explanatory values are substantially higher for the dependent variable length of conflict than for the dependent variable type of violence. The highest explanatory value is 7.3% for the dependent variable length of conflict, this is in a model with control variables, MNE experience in number of years of experience and the interaction term of cultural isolation. For the dependent variable type of violence, both for MNE experience in number of years and for MNE experience in number of projects, the explanatory values are considerably higher when adding the moderator variables than when adding the interaction terms.

(43)

43

To conclude, the outcomes of the regression analyses initially showed that the measurement type of violence is the preferred one over degree of violence for the dependent variable conflict type. Subsequently the regression analysis for degree of violence has been replaced to the appendix and the rest of the research is done with type of violence as measure for conflict type. The statistical analysis showed results which (partially) support H1, H2, H4 and H5. The remaining hypotheses, H3 and H6, are rejected.

(44)

44

(45)

The effect of MNE experience in conflict situations with indigenous communities

Statement of originality

Table of Contents

List of tables and figures

Abstract

Introduction

Literature review

Theoretical framework

Research Design

Results and analysis