Conflicts between Indigenous groups and MNEs in Africa : the role of EM firms and land grabbing investments

Master’s Thesis – International Management Track 6314M0251Y.S2

Conflicts between Indigenous groups and MNEs in Africa:

the role of EM firms and land grabbing investments

Supervisor: Dr. Ilir Haxhi

Second reader: Erik Dirksen

Author: Federico Rapetti

Student number: 11148764

Due date: 24

of June 2016

Statement of originality

This document is written by Federico Rapetti, who takes full responsibility for its content.

I hereby 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 the creation of it. The Faculty of Economics and Business is responsible solely for the supervision on completion of the work, not for the contents.

Acknowledgements

With great pleasure I would like to thank my supervisor for this project, Dr. Ilir Haxhi, for the valuable support given during the thesis and the whole academic year.

I have to thank the financial backers of my time in Amsterdam, undoubtedly: deep thanks to my family, for stimulating my intellectual curiosity, giving me an education and allowing me to wander the world.

I met many people during this year in Amsterdam. All of them left a little legacy, and ultimately contributed to this research. I would like to mention particularly the ‘indigenous party’, the colleagues that worked with me on this project, for sharing doubts and grief during the writing process. Shannon, for translating my mumblings into something that remotely sounds like academic English. Julia, for pushing me to the completion of this work with her determination, and also listening my mess. Pietro, for his help in mastering statistics. Cari, for kindly bearing me and showing that the unexpected can happen.

Finally I would like to acknowledge the support from the University of Amsterdam, the Business School and the academic faculty in particular. This has been an intellectually fecund year and the following work results from all the valuable influences I have been exposed to.

Summary

ABSTRACT 5

1. INTRODUCTION 6

2. LITERATURE REVIEW 9

2.1 MNES IN PRIMARY RESOURCE INDUSTRIES 9

2.2 EMERGING MARKETS MNES 10

2.3 INDIGENOUS PEOPLE 12

2.4 CONFLICTS BETWEEN MNES AND INDIGENOUS PEOPLE 13

2.5 INSTITUTIONS IN RELATION TO CONFLICTS 15

2.6 BIOMASS-RELATED CONFLICTS 17

3. HYPOTHESES AND RESEARCH FRAMEWORK 19

4. METHOD 24 4.1 SAMPLE 24 4.2 DATA COLLECTION 24 4.3 VARIABLES 25 4.4 METHOD 29 5 ANALYSIS 30 5.1 DESCRIPTIVE STATISTICS 30

5.2 CORRELATION TABLE AND MULTICOLLINEARITY TEST 31

5.3 REGRESSION 33

6 DISCUSSION 36

6.1 FINDINGS 36

6.2 LIMITATIONS 38

6.3 FUTURE RESEARCH DIRECTIONS 39

7 CONCLUSION 41

Abstract

The research examines conflicts between multinational firms and indigenous communities in Sub-Saharan Africa. Rising importance of Chinese companies introduces a new factor in the dynamics previously dominated by west-based MNEs. The work argues that firms originated in emerging markets experience a comparative advantage over traditional players when operating in countries with weak institutions, resulting in less violent opposition from native groups. The causal link is the competence in dealing with institutional partners matured domestically. Findings back the overall trend of this relation, supporting the advantage of EM MNEs but missing to identify its source.

The research also investigates biomass exploitation projects. The so-called ‘land grabbing’ is deemed to have explosive effects on local communities, whose lifestyle is threatened by agricultural modernization. The hypothesis advanced is that biomass-caused conflicts cause higher levels of violence compared to traditional projects. Analysis shows that there is no empirical backing for this proposition, demonstrating that the opposite is true.

The investigation of the two global trends is performed on 156 conflict cases in Sub-Saharan Africa, between MNEs and indigenous people. The self-constructed database is statistically analyzed with an Ordinary Least Squares regression. This method aims to achieve a global picture of the phenomena, going beyond the case study approach chosen so far in the literature. The purpose is to identify the determinants of violence and highlight approaches to avert it.

1. Introduction

Globalization promotes operations from Multinational Enterprises (MNEs) in African countries previously at the periphery of the global economy. For industries targeting primary resources, either for extraction of minerals, fossil fuels or exploitation of biomass, investment opportunities arise in naturally endowed Sub-Saharan countries, where deposits of resources lie unexploited.

As these areas become accessible, invasive Foreign Direct Investment (FDI) leaves a mark on previously uncontaminated regions and disrupts the way of living of local human groups that fall under the name of indigenous populations (Ballard & Banks 2003). Existence of these communities is strictly linked with the land they populate, for sustenance and identity (O’Fairchellaigh, 2013).

MNEs’ encroachment poses an existential threat to them and ensuing conflicts are multiplying (Laplante & Spears 2008). Lacking an effective legislative protection by central governments due to their marginalization (geographical, economical and cultural), indigenous communities revert to the use of violence to protect their interests (Ballard & Banks 2003). Under the lens of International Business (IB), conflicts with local stakeholders are a source of uncertainty and ultimately cause costs for the MNE (Calvano 2008; Kemp, et al; Stefani 2009). Opposition must be addressed and managed successfully to avoid destruction of value (Murphy & Arenas 2010; Bond 2014).

In Africa, globalization is also related with the rapid growth of China. As the Chinese economy massively expands, the demand for primary resources surges: China is currently Africa’s first trading partner with 220 USD Billion of interchange in 2014 (Xinhua 2015), the vast majority of which is made of minerals (The Economist 2013).

After the modernization, Chinese MNEs started to directly operate overseas, together with other Emerging Markets (EM) competitors (Berning & Holtbrügge 2012). Their intrinsic nature introduces new ways of doing business in Africa (Cuervo-Cazurra & Genc 2008).

EM MNEs are the vanguard of agricultural investments (GRAIN 2016; Land Matrix 2016), which is gathering momentum with the high prices of food commodities (Erten & Ocampo 2013). The purchase of arable surfaces and biomass reservoirs by foreign firms fosters productivity in underdeveloped areas. However, it also causes disruptions in traditional societies, linked to land for identity and way of living (Martiniello 2015). This triggers conflicts similar to more invasive of FDI types and raises global awareness on the land grabbing trend (Cotula et al 2009; Franco 2012).

In this setting, the research aims to investigate the emergence of a new factor, EM firms operating in Sub-Saharan Africa, and its implication for indigenous populations. Chinese ascendancy is recently attracting the interest of International Business studies (Broadman 2007; Kaplinsky & Messner 2008). EM MNEs break the monopoly that post-colonial investors had in Africa (Ajayi 2006) and question the traditional models of internationalization of the firm (Dunning 2006). Studies cover the pushing factors of this FDI (Kolstad & Wiig 2012; Zhang & Daly 2011). However, conditions making FDI in the host country attractive in first place are not investigated. But a growing activity by Chinese firms indicates that they thrive in developing settings. This advantage over traditional players is linked with the nature of the EM companies and needs investigation.

The focus of this study is the conflicts between MNEs and native communities in Africa. While previous studies analyze single cases studies, identifying determinants of the conflict and policy recommendation to avert it, this work adopts a different approach. A cross-sectional analysis of multiple cases highlights international trends and their implications, according to IB principles. This is achieved through the statistical exploration of a self-built database of 156 conflict cases. The events are sourced from press, NGOs websites and MNEs reports and involved MNEs and indigenous groups, in the Sub-Saharan area.

The research claims a significant difference in conflict intensity related to the area of origin of the MNE, comparing those based in emerging markets with those from developed economies. The preliminary assumption is that EM firms are more successful in averting violence from indigenous groups, thanks to the experience in informal settings (Cuervo-Cazurra & Genc 2008).’

The degree of opposition from stakeholders affected by land exploitation is also investigated, given the rising importance of agricultural investments. The research suggests that biomass-related conflicts are more violent compared to other types, because they threaten the social frame of the community (Franco 2012).

The outcome of the research contributes to the growing stream of IB literature focused on EM MNEs, confirming their perceived strengths in the context of Africa. Furthermore, examining land grabbing, the study gives some boundaries to a phenomenon whose impact still escapes academic investigation. More practically, the research supports decisions related to operations in Africa: emergence of a competing model either reinforces the western way, or points to a more effective path by EM MNEs in dealing with indigenous people.

The plan of the research is to review previous academic publications in the first section, then highlight the gaps found in the second. In the third section, literature is used to back the hypotheses and develop a research framework. An explanation on the structure of the database is present in the fourth section and the statistical analysis is displayed in the fifth. In the sixth section, results are interpreted under the light of the initial hypotheses and conclusions are drawn in conjunction with the relevant limitations and additional research gaps identified. The seventh and final section concludes the research, summarizing the findings.

2. Literature review

The scope of the research is to study conflicts between indigenous groups and private firms. A copious amount of literature investigates specific events with a case study approach, framing determinants and managerial best practices to avert confrontation in local settings. However, the phenomenon is on the rise on global basis (Laplante & Spears 2008) and hence requires a broader approach. Adopting the tools of IB, the research argues that local factors that cause conflicts are shaped by transnational causes. Therefore these interactions can be described and understood using universal models, described in the following paragraphs, to identify shared characteristics.

2.1 MNEs in primary resource industries

Historical path dependency dating back to colonial times sees foreign firms operating in Africa to exploit the abundant resources of the continent (Nkrumah 1968).

Multi-National Enterprises and their decision to locate value-adding operations across borders are explained by the influential Eclectic Model (Dunning 1980). Investment choices are dependent upon three groups of advantages: ownership of assets and know how (O-type), benefits from location (L-type) and benefits from internalization of activities (I-type).

Despite a recent wave of mergers and acquisitions that involved the mining industry caused by declining margins(EY 2015), efficiency-seeking localization is subordinate to the resource-seeking behavior in commodity industries. Countries are naturally endowed with primary resources and mineral deposits, oil fields and tropical forests need a physical presence in loco to be exploited: FDI follows presence of the target resource.

Existence of MNEs and magnitude of FDI are strictly linked with the broader phenomenon of globalization (Garrett 2000). In this new global setting, trans-national firms acquire a powerful role, influencing institutions and even daring to challenge them (Newell 2002): their contribution to

prosperity of the host country is a belief shared by the vast majority of global policy-makers, that interact directly with private groups. (Parkinson 1995).

Preferential treatment for companies sparks debate around the opportunity of this choice.1 In remote areas where national institutions lack proper representation, private firms override local stakeholders, backed by legislation that does not take into account the environmental impact. MNEs’ legitimacy is not backed by any democratic consensus, causing opposition and protests (Denon et al 1996). A proven record of lackluster socio-environmental performance is the cause of a rising global movement that tries to curtail MNE activities (Korten 1995).

2.2 Emerging Markets MNEs

Another consequence of globalization is the emergence of new countries in the global economic landscape. While traditional models (Hymer 1976; Buckley & Casson 2009) explain existence of MNEs due to direct ownership of certain advantages (either technological, reputational or managerial), the paradigm is challenged by the increased cross-border activity of firms located in emerging markets, with the Chinese firms at the forefront.

Literature questions the existence of an O-type advantage for EM firms. Rather, it hypothesizes that in some cases FDI is a tool to acquire previously unavailable resources (Amighini

et al 2009) – an idea developed already in the Springboard Perspective to Internationalization (Luo

& Tung 2007).

The Linkage, Leverage and Learning (LLL) theory is introduced to explain the success of EM MNEs with attention to the larger trend of alliance capitalism. In this view, late movers access to resources through joint ventures and partnerships, paving the way to alternative expansion in mature sectors (Mathews 2002). However, the LLL model is not adequate to explain EM FDI in

1 _{For example, Latin-American countries have embraced a promotion of extractive industries in the last two} decades, with generous tax breaks and progressive revision of mining codes towards more complacent legislation. The attraction of foreign capitals was ultimately achieved with considerable socio-environmental costs for the collectivity (Bebbington & Humpreys Bebbington 2011).

developing countries: refined resources are located in the free-market economies (advanced technologies, established brands, strong legislative protection), while Sub-Saharan Africa has little to offer under this perspective.

Another stream of literature indicates that EM MNEs have a differential advantage over established players in developing countries (Dunning 2006). Chinese management is for example advantaged in weak institutional environments due to domestic experience. This is a resource unavailable to MNEs located in countries where formal institutions are stronger (Cuervo-Cazurra & Genc 2008).

It seems not, however, that EM MNEs have a different motivation in choice of FDI locations (Xiaohua & Farrell 2013). Moreover, resource-seeking behavior is possibly the stronger pushing factor: Chinese MNEs, often suspected to act on behalf of the government when not directly state-owned, are actively involved in securing primary resources to keep the domestic manufacturing sector competitive, coupling geo-political concerns with economic motivations (Shinn 2007).

Independently from the originating area of the firm, it is safe to assume that technological constraints make the commodities industry converge towards a similar operations model (EY 2015). Due to the impoverishment of natural endowments in their host location, firms must change locations in the span of few decades, to invest in areas where untapped resources are available. Expansion to peripheral regions that up to few decades before considered economically non viable is a compelling force. This set up the basis for the interaction between MNEs and the second factor of the analysis, indigenous communities.

2.3 Indigenous people

Definition of ‘indigenous’ is the task of disciplines beyond the domain of IB (Cunningham & Stanley 2003). For the purpose of this research, a distinctive lifestyle and approach to Nature and resource management is adopted as a fundamental characteristic (Marsden 1991).

Natives2 populate geographically remote areas, beyond the reach of the institutional constructs regulating modern society. Isolation leaves them at the margins of the globalized world. Their unique lifestyles do not rely upon industrial exploitation of resources and little differ from that of the remote ancestors (Diamond 2012).

Far from the forces of homogenization and institutional convergence, native communities developed over millennia an unmatched cultural richness: accounting to only 5% of the world population (Bodley 2014), they represent by far the widest plurality in spoken languages, way of living and cultures – a heritage recognized and protected by the United Nations (UN General Assembly 2007).

As indicated previously, the definition of ‘native’ is related to the perception of land and nature: many communities identify themselves on the base of the area they inhabit and this increases sensitivity towards the usurpation of their rights (Ballard & Banks 2003). Far from being a static and reactionary value as the Primordialist view pictures (Rex 1995), indigenous culture is in constant change (Hook & Ganguly 2000). This evolution can be caused by the interaction with MNEs, with different groups of natives adapting more successfully than others (Igoe 2006).

In this process an active role is played by Non-Governmental Organizations (NGOs) (Jenkins & Yakoleva 2006) and by the same globalization seen as a primary cause of MNE’s encroachment. The emergence of the ‘glocalization’ movement is a direct consequence of it, with a growing network of indigenous groups lobbying together for their rights (Bruijn & Whiteman 2010).

Together with identity, natives drive subsistence from the land they inhabit (O’Fairchellaigh 2013): in fact, their approach to resource management is of foremost importance for this study.

Industrialized economies rely upon technologies to exploit resources on a massive scale and sustain the consumption of densely populated regions. Global trade enables Consumption and Production phases to be distinctly separated in time and space. Indigenous societies have a different relation with resources: management of the environment is linked to the survival of the community and nature is sustainably exploited to guarantee long-term subsistence (Diamond 2012).

Putting aside ethical considerations over legitimacy of the divergent lifestyles (Lertzman & Vredenburg 2005) and economistic views on the matter (De Soto 2000), the study focus on the situations when the two worlds interact. As the MNEs push in virgin areas inhabited by natives, colliding interests spark conflict.

2.4 Conflicts between MNEs and indigenous people

For the convenience of this research, conflict is defined as encompassing events ranging from minor disagreements peacefully expressed, to open violent confrontation (Kemp et al 2011). Clashes between indigenous groups and MNEs are on the rise on a global basis (Laplante & Spears 2008), due to the proven track of environmental damage that MNE operations cause (Korten 1995).

From the native side, legitimate concerns about livelihood of the communities, water access, claims on use and degradation of land, human rights violation and inequality in profit distribution are all motivations to engage confrontation with MNE’s representatives (Bebbington et al 2008). These can be traced to diverging expectations in resource management, as pointed in the previous paragraph.

In societies where institutions are strong, mismatching interests are settled through the rule of law. Indigenous people, however, are located in areas where national institutions fail to achieve legitimacy, either because corrupt or totally absent (Denon et al 1996). In this situation of

competition over a resource, one of the parties (the indigenous community) does not feel in control of its own interest and is forced to revert to open conflict (Ballard & Banks 2003).

On the MNE side, conflicts with traditional communities raise the uncertainty of the project and ultimately account as a cost, related to court litigations, delay of the commercial operations, in

loco protection of the investment and also indirectly causing increased regulation by government

bodies (Davis & Franks 2011; Calvano 2008; Kemp et al 2011). Not least, the relevance of reputation costs – a damaged image hampers access to capital and poses threats in the longer term (Stefani 2009). Under this contingent threat, MNE behavior is explained under the Stakeholder theory (Zingales 1998): managing expectations of fringe stakeholders such as natives is necessary to avert destruction of economic value (Murphy & Arenas 2010).

A recent trend in industries that operate in environmentally sensitive areas is the adoption of Corporate Social Responsibility codes (Haalboom 2012; Jenkins & Yakovleva 2006). Ethical approach to native management is deemed to have a positive impact on conflicts (Kolk & Lenfant 2010; Jamali & Mirshak 2010). Additionally, active negotiations and responsibility sharing are successful in curbing conflict intensity (Bond 2014; Castro & Nielsen 2001). CSR is a powerful instrument that empowers indigenous people, but can also be used for less idealistic goals: MNE subsidies are used to silence communities, isolate activists and ultimately perform a greenwash over less than sustainable operations (Calvano 2008).

Despite this drawback, a stream of literature promotes the role of the MNEs towards ethical exploitation, filling the gap wherever national institutions fail to deliver effective environmental protection (Hertz 2001). Theorizing the involvement of MNEs as brokers of public goods requires investigation into their alternative: institutions and their role in shaping economic activities.

2.5 Institutions in relation to conflicts

Institutions are the body of constraints set up by societies to regulate intra-group interactions, both explicitly and implicitly (North 1991). They are an important factor to interpret conflicts, ultimately caused by a divergent expectations shaped by cultural settings.

Standards of behavioral acceptance shape firm’s actions, with coercive pressures to legal compliance and incentives to adapt to a standard that attributes legitimacy. Under the Institution Based View of the firm, socio-political rules establish the basis of production, exchange and distribution (Peng 2002). Literature separates formal institutions (briefly, those sanctioned by laws) from informal (those linked with broader cultural standards). While the former have stronger roots in developed societies where the rule of law is fairly established, the latter assumes importance where governments are less pervasive and traditional authorities assume many of their prerogatives (North 1991).

Due to their transnational structure, MNE operate under different institutional frameworks depending on the country of location. The variety of settings causes uncertainty and risk, hampering the economic imperative of standardization (Porter 1986). The construct of Distance is variously codified. The Scandinavian school stresses the psychic distance over informative flows and attributes primary importance to cultural perception (Johanson & Vahlne 1977). The influential CAGE model includes multiple sources of distance, either deriving from cultural, administrative, geographical or economical disparities (Ghemawat 2001). This latter codification has the merit to match with the dichotomy formal/informal institutions, encompassing both.

The frame of Liability of Foreignness (LOF) (Zaheer 1995) categorizes obstacles faced by MNE while operating abroad. These fall under two sources (Sethi & Judge 2009). The first is ‘incidental LOF’, caused by the very foreign origin of the firm: lack of experience, misunderstanding of local laws and customs all pose a threat to the investment. Organizational experience reduces this obstacle (Davis & Franks 2011) and in the mining industry active social

learning can avert conflicts (Schusler et al 2003)3. The second is ‘discriminatory LOF’, caused by policy-induced effects that target the non-domestic origin of the MNE. Host-country political risk assumes the shape of tariffs and restrictive legislation, with the ultimate threat of nationalization of MNEs’ assets. Literature agrees on the moderating role of experience to reduce uncertainty derived by distance, with a bigger effectiveness in reducing incidental LOF (Benito & Gripsrud 1992).

Accordingly with IBV, national institutions are not only an obstacle for industry. From the grant of contracts to exploit resources of public domain to the taxation of those activities, and finally through regulation of environmental hazardous operations, they are also active players. In many cases they promoted mining investment (Helwege 2015) and for Chinese MNEs there are signs of robust government support to private enterprises that seek FDI (Zhao & Wang 2008).

National institutions are hence an important variable in MNE’s strategic decisions, theorized by the need to deploy ‘political resources’ to steer legislation towards hospitable investment environment (Feinberg & Gupta 2009). This is achieved through the engagement of managers in the public debate, in the support of government initiatives and also direct lobbying. However, the result risks to illegitimately trespass the boundary of legality. Facing national institutions notoriously porous to private meddling, MNEs in the Sub-Saharan Africa are infamous to revert to corruption, a behavior that fosters negative reactions by natives and hampers conflict resolution (Ikelegbe 2005). Chinese MNEs are perceived to have an advantage, being able to deploy tactics of “institutional taming” already adopted in the domestic setting (Cuervo-Cazurra & Genc 2008).

Under the category of informal institutions, national culture plays an important role in explaining firm’s behavior (Hofstede 1994) and related management of indigenous conflict. Managerial orientation is the outcome of educational paths and use of bi-culturally trained employees assures firms to better understand native issues (Lertzman & Vredenburg 2005). Cultural

3 _{Chinese firms seem aware of this. While 1 million Chinese nationals moved in Africa in the past two} decades to work for foreign firms (The Economist 2015), is also true that a sample of 400 Chinese firms operating in the continent shows that 80% of their workforce is local (Sautman & Hairong 2015), a higher percentage compared to western firms.

propensity towards confrontation, measured by masculinity of the culture, is another factor in alternative paths of conflict resolution (Avruch 1998). Also CSR preferences seem determined by the cultural context, with some societies pressuring private actors to operate for the common good with a different intensity than in countries where government take this duty4.

2.6 Biomass-related conflicts

As stated above, conflicts between MNEs and indigenous people are caused by the industrialized exploitation of natural resources. The substitutability of commodities makes trade truly global, with the development of the Chinese economy impacting extensively on the prices. The secular increase in demand has triggered a ‘super-cycle’ of inflated prices that only recently has shown signs of deceleration (Cashin et al 2002).

Food commodities are not spared by the price increase (Erten & Ocampo 2013) due to prioritization of cash crops, biofuel productions and cereals allocated to animal consumption (Borras et al 2011). Agricultural investments become a profitable type of FDI (Cotula et al 2009), fostering the purchase of arable surfaces by foreign investors. This land grabbing (Franco 2012) has a specific relevance for Sub-Saharan Africa, where the majority of virgin soils are located.5

Introduction of modern agricultural techniques impacts traditional communities, causing loss of traditional jobs, displacements and impoverishment of biodiversity, all factors that contribute to indigenous conflicts (Martiniello 2015).

The magnitude of the phenomenon is likely exaggerated (Bruckner 2015; Brautigam 2015), but EM MNEs are very actively involved (Land Matrix 2016). Recent protests in different

4 _{For example, US-based firms are pressured by public opinion to be responsible towards the collectivity} more than in Europe, where welfare provides for the shortcomings of the private sector (Crane et al 2013). 5 _{About 200 million hectares, 75% of the global surfaces that can still be exploited for agriculture (Cotula et}

countries6 drove the attention of this research, due to characteristics similar to other indigenous conflicts.

6 _{In April 2015, the palm plantation owned by Socapalm in Dimombari, Camerron, was blockaded by the} angry residents, complaining about the unfair practices of land management. Similar protests were coordinated with farmers in Sierra Leone, Liberia and Cambodia, against the plantations of Socapalm. In December 2015 5 protestants died in the mobilization against the purported plans to expropriate lands under customary law in Oromia region, Ethiopia.

In April 2016, young residents of Samburu county in Kenya rose against the Kirimon group holding, accused of privatizing pasture lands.

3. Hypotheses and research framework

Given the rising importance of EM firms’ operations in Africa, this research argues the existence of a comparative advantage that EM MNEs have when operating in developing countries, especially when dealing with indigenous groups. This is linked with familiarity to institutionally weak settings and reflects in a lower degree of violent opposition from natives.

Moreover, the research argues that arable lands and biomass-related projects cause stronger opposition from local communities, compared to more traditional types of investment. The reason is the impact on the indigenous group’ social frame, whose agrarian identity and livelihood are fundamentally threatened.

Hence, the object of investigation is the conflict between firm and indigenous community. Confrontations create an uncertain outlook for the FDI and account as costs for the MNE, making operations less profitable and threatening the reputation of the company. (Davis & Franks 2011; Kemp et al 2011; Stefani 2009). Successful managerial decisions are implemented in a way that does not harm indigenous stakeholders and avert conflict (Bond 2014; Castro & Nielsen 2001). When this is not achieved, escalating levels of violence reflect the original cause of the conflict (Hart & Sharma 2004; Calvano 2008). The degree of violence is hence taken as a proxy for successful management: when the violence is mild, an effective mediation is reached or the underlying causes are not of capital importance. On the other hand, severe confrontation with resort to armed conflict is taken as a proof of lackluster management or of vital interest threatened. In the following paragraphs the framework of the research is summarized, explicating the underlying assumptions and the hypotheses to test.

Recent increases in activity of EM MNEs in Sub-Saharan Africa questions traditional views on MNE activities (Dunning 2006) and lead the claim of an advantage that the new-comers have in developing regions, due to experience matured in the originating country (Cuervo-Cazurra & Genc

with weak institutions but important traditional authorities (North 1991), thanks to effective use of political influence (Feinberg & Gupta 2009). Used to interpret underlying interests through informal channels, EM MNE are better in managing the expectations of local parties and avoiding costly conflicts. The statement is made explicit in the following hypothesis:

Hp 1: MNEs originated from Emerging Markets are more successful in managing conflicts with indigenous groups compared to traditional MNEs, resulting in a lower degree of violence.

Conflicts caused by EM MNEs are expected to be less violent due systematic reliance upon informal channels to solve divergences, reinforcing the perception of a comparative advantage (Cuervo-Cazurra & Genc 2008).

It is undeniable that a salient characteristic of African governance is the lack of transparency and perceived inefficiency of the bureaucracy. Sub-Saharan African countries score systematically low in the ratings for quality of governance (WGI 2016). The relation between government effectiveness and citizenship welfare is well documented (Kaufmann et al 2011) and supports the importance of institutional quality to promote social peace and positive economical environment.

Narrowing to the scope of this research, the degree of corruption in the public sphere negatively influences the outcome of conflicts (Ikelegbe 2005). Corrupt bureaucracies dissipate the financial resources allocated to solve social problems. The failure in the task to redistribute economical benefits causes negative reactions from the damaged communities (Denon et al 1996).

Given the pervasiveness of the institutions in the projects we examine, it is of the utmost importance for the MNEs to find a way to successfully interact with public stakeholders (Helwege 2015). This task is theorized with the need to deploy political resources (Feinberg & Gupta 2009). It would be naïve to ignore that use of these “political resources” in African countries often means corruption of public officials. An accusations moved to Chinese firms is that they operate with ease under African regimes notoriously opaque (Cuervo-Cazurra & Genc 2008).

Hence the previous hypothesis of a comparative advantage with the moderation action of corruption control is refined in the following way:

Hp 2: The control of corruption in the host country negatively moderates the correspondence between EM origin, and degree of violence of the conflict.

EM MNEs are hypothesized as more successful at managing conflicts in countries that loosely enforce control on corruption, due to the peculiar reliance upon informal networks, hence supporting the superiority of their approach (Helwege 2015).

The initial assumptions advance a supposed peculiarity of EM MNEs, whose less established nature constitutes an advantage in certain settings. Their motivations for investment in Sub-Saharan Africa, however, do not differ from those of traditional players (Xiaohua & Farrell 2013). Independently from their origin, MNEs are driven by the imperative of profitability through the exploitation of natural resources (Shinn 2007).

Native communities, on the other hand, have a significantly richer approach to the environment they inhabit (Lertzman & Vredenburg 2005): they derive identity (Ballard & Banks 2003) and means of subsistence from it (O’Fairchellaigh 2013). Mismatching expectations around the management of the resources is the ultimate cause of conflicts (Murphy & Arenas 2010).

Divergent perceptions derived from cultural otherness (Hook & Ganguly 2000) are also linked to the resource target of the investment and impact on the type and length of conflict. For example, extractive industry has a proven track of exploitation (Jenkins & Yakovleva 2006) and mobilizations against mines accompany the preliminary exploration of sites, even before projects are made official. Similarly, dam construction attracts extensive attention due the access the primary resource: water (Kolk & Lenfant 2013). The long-term horizon of river management infrastructure curbs down destructive content of the conflict that tend to remain at a low intensity (Getz & Oetzel 2009).

Alike, investments in agricultural lands have specific consequences. The establishment of private property to biomass reservoirs under customary laws reduces the means of subsistence of the community. Land grabbing also forces relocations and is deemed to have an explosive consequence on overpopulated areas (Franco 2012). Therefore, the following hypothesis is stated:

Hp 3: Biomass-related projects cause conflicts with a higher degree of violence, compared to other types of investment.

Conflicts driven by investments in biomass, either for agriculture, logging or animal farming, are expected to cause an elevated level of violence due to the perceived threat to the identity and livelihood of the community (Cotula et al 2009; Franco 2012)

Rule of law in the western meaning is seldom existent in non-institutionalized groups (North 1991). Indigenous societies rather rely upon hybrid cultural-social authorities to regulate intragroup interactions. Without a formal entity exercising the monopoly of force, held by the government in modern states, natives are kept in check by relation-based links that sanction the use of violence (Diamond 2012). Customary institutions evolve over time and achieve legitimacy during the process (Rex 1995). The forced interaction of native groups with MNEs challenges this long-term process and wears down the social fabric (Hook & Ganguly 2000).

Compared to other projects, land grabbing investments have the potential to disrupt even more radically the delicate equilibrium of isolated communities (Martiniello 2015). The introduction of new agricultural technologies causes the loss of traditional jobs and threatens food security. Privatized areas are interdicted to natives, even if exploited before by the community under a customary regime. Social interactions are revolutionized, resulting in the redefinition of traditional roles and marginalization of women and elders. Finally, new problems arise with substances abuse, prostitution and cultural disorientation. It is sadly common that new contacted groups are all but wiped out in a matter of years, after the first interaction with external actors (Calvano 2008).

While their social frame enters a crisis, indigenous people find traditional authorities ineffective. The institutionalized limits of violence are not valid anymore and coercion becomes the only way indigenous groups can prevail against culturally unconstrained enemies (Diamond 2012). When the living system is under threat, natives act accordingly with extreme reactions. Hence, the hypothesis is formalized under these terms:

Hp 4: Consequences on the social frame of the community positively moderate the relation between the biomass content and the degree of violence of the conflict.

Conflicts impacting the traditional social structure of the community result in more intense violence due to the weakening of the informal institutions that keep violence in check (Calvano 2008; Diamond 2012).

4. Method

Multiple studies focus upon conflicts between MNEs and indigenous groups, investigating them predominantly under a qualitative perspective with description of specific cases. The very nature of the research questions, namely if EM MNEs significantly differ from MNEs based in more developed countries when dealing with indigenous conflicts in Africa, calls for a trans-national approach.

To fill this gap, a cross-sectional conflicts database is developed. 156 cases of conflict between indigenous groups and firms are selected in the sub-Saharan region. The cases for Africa are a part of a bigger global database, result of the shared effort with other students on the same project using a common coding scheme. The entries are chosen to cover the widest geographic diversity, mitigating the immediate availability of data for certain regions7 to ensure plurality. Furthermore, considering the research hypothesizes a role for the industry in the degree of violence, the database covers a wide variety of projects and resources involved.

4.2 Data collection

Entries are built through secondary data gathered from sources available online, with prominence for Environmental Justice Atlas. The project maps and describes cases involving a threat for environmental rights (EJA 2016). Many entries include indigenous groups and thus fall within the scope of the research. The definition of ‘indigenous people’ (Cunningham & Stanley 2003) is not taken adamantly, rather encompassing all ethnic groups whose livelihood is influenced by a project

7 _{For example, North America is a significantly represented thanks to wide legislative recognition of} indigenous groups and NGOs presence. Other areas, such as China or the Middle East, fail to achieve numeric consistency due to absence of indigenous guarantee or limited presence of NGOs that usually report the conflicts.

run by a firm, without strictly applying the already broad UN definitions for native groups (UN 2007).

MNE definition is relaxed as well. The introduction of an ownership structure variable allows to include private firms, both publicly traded and privately owned, and state-owned companies. Additionally, domestic firms are examined and for this reason the database includes conflicts caused by domestic investment projects. For MNEs characteristics, official corporate information available on the company websites is used.

Conflicts span from the 1920s to current times with the majority starting less than ten years ago. A common pattern sees conflicts ending when a court implements a sentence that interrupts the project. If this does not happen, communities keep the conflict awake almost indefinitely, even after the end of the commercial exploitation phase if the project still damages the environment.

Finally, for country parameters the research relies upon data from the World Bank and their project for the Worldwide Governance Indicators. The most recent data available is selected, predominantly for 2014. For some African countries, no official data currently exists. The database is populated with the most reputable estimations from official organs.

4.3 Variables

Dependent Variable Degree of Violence

The variable derives from qualitative descriptions of the conflict and can take six categorical dimensions: 1, latent; 2, low; 3, mild, 4, medium; 5, high; 6, extreme.

Latent degree of violence describes conflicts that are still at a dormant stage, with preliminary initiatives from a limited part of the citizenship. Usually these are: self-called community consultations, sensitization campaigns and petition to government agencies.

Low degree is identified when the conflict involves a wider share of the civil society and activism starts to appear. Examples are: non-binding consultative votes, peaceful rallies and attempts to bring the case to court.

Mild degree finds communities actively pursue opposition techniques with higher impact but still non-violent means. Commonly these are: temporary occupation of properties, symbolic protest actions, limited injuries and occasional arrest of activist under minor charges.

Medium degree of violence includes active use of coercive tactics to influence the outcome of the conflict with resulting collateral damage to both people and assets. Examples are: intense protests and strikes, road-blocks, prolonged property occupation, targeted arrest of activists and heavy injuries requiring hospitalization.

High degree depicts conflicts fully escalated with systematic use of violence and intentional damages. Commonly this encompasses: arson and forced demolition of buildings, sabotage of operations, mass arrests, criminalization of activists and targeted assassinations.

Extreme degree sees widespread violation of human rights with aspects of a proper insurgency. Observed characteristics are: military repression, kidnapping, torturing, paramilitary actions, systematic killings and ethnical cleansing.

Despite the scale measures the degree of mobilization from the community, no distinction is made between violence perpetrated by either MNE agents, government representatives or indigenous individuals.

Independent Variables Area of origin of the MNE

The MNE is described as originating from either: 1, Emerging Market, or 0, Developed Country. A dichotomous variable is chosen to achieve numerically consistent groups. More detailed categorization discriminating on the base of the country of origin would result in non-representative clusters. The criterion is set at home country level with no distinction on specific MNEs and their

characteristics. For the home country is chosen where the global headquarter is located, even if the subsidiary is local. Home country membership to the Organization for the Economic Co-operation and Development (OECD) is chosen as a proxy for high quality institutions, strong market economies and thus commonly described as Developed Country (OECD 2016). Remaining cases are coded as Emerging Markets. Two countries avoid this classification: Jersey Island, a Crown Dependency of the United Kingdom, is coded as ‘developed’ due to the OECD membership of the parent state. Singapore, despite not being an OECD member, is universally considered a developed economy and coded accordingly.

Biomass-related project

The nature of the MNE’s project, linked with the exploitation of a specific resource, is coded with this dichotomous variable. The project can either be: 1, Biomass-related; 0, Other project. In case of multiple causes, the most relevant is chosen. Biomass includes conflicts developing around forestry resources, arable surfaces and pasture areas, both for commercial use and for natural conservation. The management of territory is related to them and overlaps with the phenomenon of interest, land grabbing.

Moderators

Host country corruption control

The governance quality in the host country is measured by the Worldwide Governance Indicators. The World Bank aggregates personal views from a large number of representatives belonging to civil society, academia and business, on the perceived quality of the governance in specific countries (Kaufmann et al 2011; WGI 2016). The Index ranges from 0 to 100 percentiles and rates six different governance areas. The one considered to moderate the model is corruption control, which is the perceived “extent to which public power is exercised for private gain” (WGI 2016). The values used for the research are those for 2014.

Impact on community Social frame

The variable describes the impact from the conflict on the social frame of the community, according to NGO’s report. It is either: 0, absent; 1, present. The scope is broad and commonly includes discrimination of indigenous groups, marginalization of women, alcoholism, prostitution, increased crime rates, contestation of traditional authorities, increased police presence, corruption and bribes, co-optation of leaders, loss of traditional jobs and rituals.

Control variables

To evaluate the effects of the variables on the model, control parameters are chosen – one for each of the levels of the analysis.

At MNE level, the ownership structure describes the property of the firm, i.e. who exercises the indirect power of control. Firms are either: 1, Government owned; 2, Privately owned; 3,

Publicly traded. In case of mixed structures, the majority stakeholder is taken into account.

At community level, the geographic isolation of the community is described as either: 1, Isolated; 2, Not isolated. Communities living in remote regions with a low population density such as jungles, mountainous area or deserts are coded as isolated, while the remaining falls under the residual category.

At conflict level, the length is measured from beginning of the first mobilizations to end date. Duration can either be: 0, Short; 1, Long. In 126 of the cases, more than 80% of the total cannot be considered closed at the time of the investigation. For the sake of analysis the research considers them ended by July 2016. The median length with this assumption is 90 months (7,5 years). The split between long and short conflicts is set at 84 months (7 years), assuming that conflicts currently ongoing will likely continue for more than 6 months.

4.4 Method

The relation described by the hypotheses can be expressed in the following equation:

Y = α + βX

+ γX

+ ε

Y stands for the degree of violence and α the intercept. β and γ are the inclination of the lines

corresponding each value of the dependent variable Y to the respective independent variables. These are X1, the area of origin of the MNE, and X2, the biomass content of the conflict. ε indicates the

error, the difference between the Xi value observed and the value forecasted (Field 2013).

An Ordinary Least Squares (OLS) linear regression links the categorical dependent variable and the dichotomous independent variables. The method still returns consistent results even if the outcome is categorical. With more than five categories, scale measurement is acceptable and the method is thus effective (Rhemtulla et al 2012). OLS aims to reduce the divergence between observed values and those forecasted through the linear approximation. A small difference between them, calculated by the coefficient of determination (R2) attributes a higher predictive power to the model. The analysis is run with a stepwise approach to test the influence of the variables separately. The method regresses multiple variables while removing the weakest correlated one. OLS is an adequate technique when the empirical backing of a research is not strong, which is the case due to lack of previous quantitative studies on indigenous conflicts (Field 2013).

Table 2 summarizes the process: first control variables are tested, followed by the independent variables alone. Subsequently moderators are introduced as interaction terms, repeating

5 Analysis

5.1 Descriptive statistics

The cases for the African continent amount to 156, distributed in the following manner: - Northern Africa (2): Sudan (2)

- Western Africa (44): Nigeria (29), Liberia (5), Mali (3), Sierra Leone (2), Senegal (2), Niger (2) and Guinea-Bissau (1)

- Eastern Africa (70): Kenya (22), Tanzania (13), Uganda (11), Mozambique (7), Madagascar (6), Ethiopia (6), South Sudan (2), Malawi (2) and Somalia (1)

- Central Africa (18): Zimbabwe (8), Democratic Republic of Congo (4), Cameroon (2), Equatorial Guinea (1), Congo (1), Gabon (1) and Zambia (1)

- Southern Africa (22): South Africa (15), Namibia (4), Botswana (2) and Lesotho (1) Table 3 shows the result of a descriptive analysis on the dataset.

The degree of violence has a mean of 3,35 with a relatively normal distribution. A variety of conflicts are observed with a slight prevalence of lower intensities. MNEs found have mostly a private ownership, with a majority of publicly traded firms. One third of the communities identified lives in areas hard to access. The mean length for a conflict is 135 months (11 years and 3 months) and the median is at 90 months (7 years and half). With the division set at 7 years, 69 conflicts can be considered short and the remaining 87 are long. The MNEs involved are mostly based in an OECD country (100) with the remaining located in emerging markets (56). 39% of the conflicts are caused by biomass-related projects, while the remaining 99 have other causes. WG indicator for corruption control has a mean at the 22,34th percentile. The range is broad, because in the sample are the countries of southern Africa with a good quality of governance, together with some of the so-called ‘failed states’ (Call 2008) with the lowest scores on global scale. A meaningful impact on the social frame of the community is identified in 119 of the cases.

5.2 Correlation table and Multicollinearity test

Pearson bivariate correlation is run to test preliminary correspondences between the variables (Field 2013). Table 3 includes the results, with the 4 significant correlations found (p < 0,01 and p < 0,05) discussed below.

The relation between area of origin of the MNE (IV) and degree of violence (DV) is significant (β = -0,166, p = 0,039) and coherent with the hypothesis. EM MNEs cause less violent reactions compared to OECD-based firms, confirming a meaningful difference. Projects exploiting biomass approach statistical significance (β = -0,139, p = 0,083), when related to the degree of violence caused. However, the direction of the correspondence goes against the hypothesis. Land investment seems to trigger lower levels of opposition compared to other projects, an important finding that contradicts the model.

An obvious trend (β = 0,149, p = 0,064) is present between the ownership structure of the MNE and the violence of the conflict. State-owned firms appear to cause milder opposition from local communities. This finds additional support in the highly significant correspondence (β = -0,577, p = 0,000) between area of origin of the MNE, and ownership structure. EM MNEs are commonly government-owned, while West-based firms are almost universally private and publicly

in domestic investments. Conflicts with indigenous groups in this case have a lower degree of violence because local firms are not hampered by the distance foreign investors experience.

A strong statistical significance (β = -0,180, p = 0,024) is identified in the correlation between ownership structure of the firm and the biomass-content of the conflict. The sign is negative, meaning that private companies are more frequently involved in this type of conflict. Agricultural investment is a relatively new business and domain of private investors, while state-owned enterprises usually operate in traditional industries.

Public firms are also meaningfully related to shorter conflicts (β = 0,178, p = 0,032). An attempted explanation is that state-owned companies can leverage their connection with government bodies to end violence, leading to a shorter timespan in the conflicts in which they are involved. Despite lacking statistical significance (β = -0,112, p = 0,164), there is a trend linking violence and geographic isolation of the community.

To further investigate whether linear relations exist between the independent factors of the model, the variance inflation factor (VIF) is examined. The values returned on table 4 with degree of violence as dependent variable are well below the threshold of concern of 5 (Field 2013), showing very weak collinearity. The tolerance level similarly highlights that no linear relations is identified, supporting the model validity and the inclusion of all the variables without risk of collinearity.

5.3 Regression

Five models are tested with the Ordinary Least Squares regression: the first only with the control variables, the second and the fourth with the control variables and one of the independent variables, the third and the fifth with control variables, independent and the related moderators, set as interaction terms. Results are on table 5 below.

The first model tests the effects on the degree of violence from the ownership structure of the MNE, the geographic isolation of the community and the length of conflict. The model finds a highly significant constant (β = 3,191, p = 0,000), which is the intercept of the function with the

Y axis. There is also the significant presence of the ownership structure variable (β = 0,421, p =

0,044) as positive determinant of the degree of violence, already explained in the comment to the correlation table. Geographic isolation of the community and length of conflict are negatively related to the dependent variable, despite not being significant. The F-test on the model determines that the variance explained is twice as big as the non-explained (F-statistic = 2,194), indeed an

encouraging result. However, the model illustrates only 2,3% of the observed variance of the degree of violence (Adjusted R2 = 0,023), a low value.

The second model introduces the area of origin of the MNE as a determinant of the degree of violence. The impact of the control variables is constant. The influence of the new variable is not highly significant but indeed confirmed, with EM origin contributing to reduce the violence (β = -0,411, p = 0,289). The Adjusted R2 remains stable compared to the model including only control variables, but the explanatory power worsens (F-statistic= 1,930) and thus the hypothesis is rejected.

The introduction of the interaction term between the above tested area of origin and the hypothesized moderator, corruption control in the host country, achieves a better outcome. The model achieves the explanation of 3,9% of the variance in the dependent variable (Adjusted R2 = 0,039) and the proportion of forecasted variance is 2,254 compared to the unexplained, at a level little short from significance (p = 0,052). The coefficient for the interaction term is negative and statistically almost acceptable (β = -0,027, p = 0,066). Institutional quality is positively linked with milder violence, in accordance with the literature. It is worth to point that the effect of area of origin has an identifiable shift in direction (β = 0,372), impacting positively on the degree of violence: a probable explanation is advanced in the next part of the research.

The fourth model tests the control variables together with the content of the conflict, whether biomass-related or not. The latter is found negatively related to the violence (β = -0,451) as previously described in the correlation table. This clearly contradicts the hypothesis forecasting a stronger response. Empirical backing of the model achieves a negligible explanatory power of 2,9%,with an F-statistic of 2,160 (p = 0,076) and thus the hypothesis is discarded.

The introduction of the moderator of the impact on the community social frame does not improve the situation. The model loses explanatory power (Adjusted R2 = 0,024) and the F-test goes below the critical threshold of 2. General directions of the variables are confirmed and the social

impact is found to have positive impact on the level of violence (β = 0,271). However, the model does not reach enough significance and is rejected as well.

In the sixth section of the research the statistical results are analyzed in the context of the literature, interpreting the findings and the new gaps identified.

6 Discussion

The empirical test of the database shows that the hypotheses require a reformulation to be valid under statistical criteria.

The first hypothesis states a relation between EM MNEs and conflicts with low intensity. The direction is indeed confirmed (β = -0,411), hence firms originated in non-OECD countries tend to trigger less opposition from indigenous groups. However the level of significance of the relation does not achieve the arbitrary threshold of p < 0,05 and its validity is doubtable. In the database there are not enough evidences to support the claims of a perceived advantage experienced by Chinese firms (Cuervo-Cazurra & Genc 2008) or to individuate the otherness of EM MNEs compared to established players (Dunning 2006), at least in indigenous conflicts. The limited size of the sample has an influence on this outcome, together with the rather rough division between EM versus OECD-based companies. A wider database, possibly extended beyond Sub-Saharan Africa, can produce more consistent sub-clusters. Chinese firms can be tested on their own, compared to European, American and other EM companies. In this scenario, national culture can explain the outcome of conflicts (Avruch 1998) together with the preference for CSR practices (Crane et al 2013), that positively impacts conflict resolution (Kolk & Lenfant 2010; Jamali & Mirshak 2010).

The second hypothesis is presented as an attempt to link the origin of the MNE to the violence of the conflict, through an interaction term of corruption control in the host country. This aims to explicit the claim that Chinese MNEs rely on informal networks to advance their projects (Cuervo-Cazurra & Genc 2008) and thus strict enforcement of corruption laws limits this channel. The result is indeed significant, but with an unexpected outcome. Introducing the WG Indicator for corruption control, the findings show that EM MNEs experience more violent conflicts compared to established players. The negative impact of high levels of corruption policing over the violence is meaningful but negligible (β = -0,027, p = 0,066). Therefore, the second hypothesis is rejected

because it does not back the perceived advantage of EM firms (Cuervo-Cazurra & Genc 2008). Instead, it supports the claims that good governance promotes social peace (Kaufmann et al 2011) and that deterrence of corruption reduces the chances of conflict escalation (Denon et al 1996; Ikelegbe 2005).

The third hypothesis advances the claim that conflicts involving the appropriation of biomass and arable surfaces are more violent. The result of the empirical test firmly contradicts the assumption, showing how land grabbing conflicts have a significantly lower violence compared to the rest (β = -0,451). Despite involving issues crucial for the survival of the community such as food security, cultural identity and management of the land (Franco 2012; Martiniello 2015), biomass-related conflicts have a rather low intensity. This is likely due to lack of visibility of the direct cause. Massive land purchases are usually made in peripheral areas and projects experience long waiting times before the start of commercial exploitation, due to the absenteeism of the investors (GRAIN 2016). Moreover, land ownership is not as visible as a surface mining pit or a dam that perturbs the morphology of a whole region, and for this reason agrarian investments tend to draw less publicity and opposition. Rejecting this hypotheses comes with the acknowledgement that the composition of the sample can determine a biased outcome. Coding criteria can contribute to a downgrading of the violence8. More importantly the database suffers from lack of representativeness, coupling agriculture, logging and pasture, in opposition with a wide category of projects including mining, drilling, water management and civil infrastructure. A higher number of cases can be split into more homogeneous and representative categories, testing for example the perceived bad reputation of mining (Jenkins & Yakovleva 2006) or the characteristics of conflicts threatening water access (Kolk & Lenfant 2013).

8 _{As databases on land grabbing show, a consistent number of deals are announced but never implemented} (Land Matrix 2016). They fail due to the excessively ambitious proportions and related doubts on the economic feasibility; but also due to public outcries and following governmental obstructionism, concerned by national food security and sovereignty on the land. A number of these purported land grabbing deals are

The fourth and last hypothesis states that damages of the social frame of the indigenous community provoke higher degrees of violence in the context of biomass exploitation. The relation is, quite intuitively, positive (β = 0,271) but fails to achieve a good degree of significance (p = 0,638). This last hypothesis is set aside, despite confirming that the wearing down of the social frame of a community negatively impacts on the propensity to compromise, resulting in widespread recurrence to coercion (Diamond 2012).

Regarding the control variables, the direction is coherent with the hypotheses and finds support in the literature, despite the limited statistical value. Ownership structure is positively related to the degree of violence. As stated in the comment to the correlation table, public firms have a direct connection with the government and can more easily control conflicts. Geographic isolation also is positively linked with more intense conflicts. The remote location of isolated groups constitutes an obstacle to communication and peaceful settlement (Calvano 2008). Finally, length and violence are negatively related in all the models. Shorter conflicts are characterized by higher violence. Two complementary explanations are possible: visible acts of violence attract the attention from institutions that intervene to end the confrontation (Helwege 2015). Also, recent conflicts9 have a higher destructive content due.

6.2 Limitations

Together with the above stated limitations in the database, the research design lacks some efficacy in testing the hypotheses.

Primarily, the attention of the research is towards presence of a conflict: visible outbursts are a preliminary condition to draw the interest of the sources used, media or NGOs. This requirement ignores the alternative paths that lead to confrontation. Due to inherent costs for all the stakeholders involved, peaceful negotiations are the preferred way to settle issues and usually take place under

the radar (Calvano 2008)10. Paradoxically, successful companies fail to be part of the sample because they avert open conflicts, thus weakening the conclusions of the research.

Availability of data is another relevant limitation. Conflict cases are derived from secondary data by NGOs. This might induce a bias, because advocating groups have interests in depicting events in a way favorable to their agenda and adopt a narrative of victims and perpetrators. Besides, natives rarely stand as independent parties with organs of communication separate from NGOs, while MNEs prefer not to put potentially harmful cases under the spotlight.

Related to this, NGOs tend to give more coverage to conflicts caused by MNEs under close scrutiny from the western media. Chinese MNEs in Africa are harder to trace due to the same supposed preference to settle potential conflicts through informal negotiations (Cuervo-Cazurra & Genc 2008). Moreover, a critical public opinion is absent in their home country. This likely results in a bias against West-based MNEs, because sheer amount of reported cases by NGOs returns a negative picture.

The hypotheses on land grabbing are open to criticism as well. The phenomenon is rather broad and only partially overlaps with biomass-related investments, so the coding criteria chosen oversimplify the issue. Moreover, lacking proper academic investigation in lieu of journalistic sources, the size of the phenomenon is most likely overestimated, like the claims of the devastating impacts on native societies (Bruckner 2015; Brautigam 2015).

6.3 Future research directions

The main assumption of this research, albeit only partially verified, is the comparative advantage EM MNEs have when managing indigenous conflicts in Africa. The findings indicate that this relation is true but fail to address why this happens. Besides mentioning relation-based competences in institutionally weak settings (Cuervo-Cazurra & Genc 2008) and experience in deploying

10_{For example: promotion of FPIC (Free Prior Informed Consent) (Lertzmann & Vredenburg 2005),} effectiveness in backroom negotiations and cooptation of actors (Ikelegbe 2005) and role of deterrence,