Militarization by refugees can have problematic outcomes. It can undermine the sovereignty and stability of the host state, perpetuate a transnational conflict and obstruct international efforts to resolve it, and present difficulties in the provision of humanitarian assistance to needy populations. Existing literature privileges structural explanations for militarization while neglecting the agency, interests and internal politics of refugee groups. In this paper, I offer a comprehensive theory of refugee militarization that emphasizes the importance of endogenous factors, including political and economic motivations, in the context of broader structural factors, including political opportunities and resource mobilization, mediated by the presence of militancy entrepreneurs. This theory helps integrate the motivation of refugees, and the discursive framing used by militancy entrepreneurs to mobilize them, with capacity for militant activity. The need for case studies and specific policy recommendations for host states, non-governmental organizations and international stakeholders are discussed.
Militarization by refugees can have problematic outcomes. It can undermine the sovereignty and stability of the host state, perpetuate a conflict with the country of origin (CoO) and obstruct international efforts to resolve it, and present difficulties in the provision of humanitarian assistance to needy populations.
Why do some refugee groups militarize while others do not? Existing literature addressing the question of refugee militarization has focused almost exclusively on exogenous factors, neglecting their agency, interests, internal politics and orientation. While acknowledging the importance of exogenous factors such as political opportunity and resource mobilization potential in facilitating the possibility of militarization, I aim to fill this gap by proposing a comprehensive theory of refugee militarization including political factors endogenous to refugee groups, which will help explain the motivation of refugees to militarize and the framing used by militancy entrepreneurs to mobilize them. The theory has potentially important policy implications for a variety of actors, including host states, donors, non-governmental organizations (NGOs), the international refugee regime and the United Nations High Commissioner for Refugees (UNHCR).
Refugee militarization is the involvement of groups of refugees in militaristic activities, including political violence, armed resistance, military training, explicit support for combatants, storage and diffusion of weapons, and/or military recruitment. I refer here to instances that are “persistent” (occurring over consecutive years) and/or “intense” (based on casualties and narrative descriptions), following the lead of Lischer (2000).
Most characterizations of refugees in the academic literature view them as passive objects, generally either as cases of humanitarian need – what Nyers (2006) terms “refugeeness” – or as unfortunate by-products of power politics (Morris and Stedman, 2008). Stedman and Tanner make a brief exception to this trend by acknowledging that “there are cases in which refugees grant legitimacy to the warriors who militarize their camps and see them as either protectors or liberators” (2003: 4) but the remainder of their volume revolves around the manipulation of refugee groups and attendant international aid by external militants, and their theory perpetuates the notion of refugees' passivity.
Such a static characterization is incomplete. Refugee groups may be seen as dynamic social actors who make decisions which affect their political destiny, including actions that involve the CoO and are likely to affect the security of the host countries in which they have found refuge. For example, refugees can facilitate militants or rebels fighting against a government in the CoO, support them with food and supplies, or their camps can serve as a base for insurgents and rebels (Loescher and Milner, 2005); and they can make raids on their CoO, drawing their enemy into host-country territory. Exactly this sort of activity took place in Guinea in 2000, when UNHCR was forced to relocate the Liberian and Sierra Leonean refugee camps due to their militarization and involvement in armed conflicts with their CoO (Andrews, 2003; Loescher and Milner, 2008). Other instances of refugee militarization include Palestinians in Lebanon and Jordan, Rwandan Tutsis in Uganda, Rwandan Hutus in the Democratic Republic of the Congo (Zaire), Kosovars in Albania, Khmer and Karen refugees in Thailand, Sudanese in Ethiopia, Nicaraguans in Honduras, Eritreans in Sudan, Kurds in Iraq and Saharawis in Algeria (Grare, 2003; Lischer, 2005; Muggah, 2006; Nyers, 2006; Terry, 2002; Zolberg et al., 1986). There is also some emerging evidence that the Taliban operating in the Northwest Frontier province of Pakistan has been recruiting from refugee camps there. Studies have provided qualitative and quantitative evidence that refugees can increase the chance of engagement in conflict by the host state (e.g. Loescher and Milner, 2005; Loescher and Monahan, 1989; Muggah, 2006; Salehyan and Gleditsch, 2006), and have found that political violence involving refugees occurs in about 15 per cent of states hosting refugee populations of 2,000 or more.1
On the other hand, most refugees do not participate in violent political activity. For instance, Hindus fleeing religious and political persecution in Bangladesh in 2001 did not militarize in India against their state of origin (Bhaumik, 2001); Palestinians in Syria have not organized militarily, with some rare exceptions; overcrowding and poor living conditions at Somali refugee camps in Kenya have not led to militarization against Somalia; and in the 1980s millions of Afghans hosted in Iran and almost a million Mozambicans in Malawi did not take up arms (Lischer, 2005).
Awareness of the politically active and potentially militant role played by refugees is relatively recent. Ferris noted in 1985 that failure to integrate and assimilate refugees into the host state, in situations in which return to the CoO is not available, has led to instances of refugees engaging in militant activity, citing the examples of Afghans in Pakistan, Khmers in Thailand and Miskitu Indians in Honduras (1985: 19). In 1986 and 1989, Zolberg et al. were the first refer to such groups as “refugee-warrior communities”, and Loescher and Monahan noted that “refugees often live on, or very near, disputed borders; they either reside among combatants in an ongoing conflict, or are perceived to be materially assisting guerrilla forces attempting to overthrow the government from which they have fled” (1989: 3). Nonetheless, academic interest in the topic was not strongly stimulated until after the realization that international humanitarian relief actually supported the perpetrators of the Rwanda war in the mid-1990s (Harpviken, 2008).
Most scholars have sought explanatory variables exogenous to the refugees, usually structural explanations. This is curious given that some challenge the traditional portrayal of refugees as passive, decrying “the striking absence of political ‘voice’ or agency on the part of refugees” (Nyers, 2006: xiv). Harpviken, for example, argues for “a need to reinsert agency into the study of war-related migration”, yet he explains the agent–structure balance by stating that migrants “are conscious and capable actors, actively engaged in social networks that define their response opportunities, but they are also subject to dangerous structural forces far beyond their direct control” (2009: 18, emphasis added). Counterintuitively, agency is demonstrated as social networks that delimit their choices. In this paper, by contrast, I use the term agency to emphasize refugees' structurally transformative potential, rather than just the limited set of choices – including choices related to perception and interpretation – which reify and reproduce existing structures. In other words, to exercise agency is to exercise the capacity to make choices that challenge existing social structures (Hays, 1994).
Nonetheless, it is important to recognize that, as Morris and Stedman (2008) briefly note, refugee militarization is sometimes less than voluntary, as they are coerced to support or participate in violent political activity by militancy entrepreneurs. In such cases the agency of refugee groups' involvement is reduced, but that does not mean that their interests, orientation and perceptions are irrelevant, nor that they are ignored as framing devices by militancy entrepreneurs; rather, the relative weight of those dimensions in affecting their militarization may be diminished in such cases.
Scholarly attention to the phenomenon of ethnic conflict, however, has included significant consideration of factors endogenous to the groups involved – factors such as nationalism, ethnicity, identity and economic interests – in addition to structural factors and institutions. Yet, surprisingly, these attributes have not yet been applied to the study of refugee militarization. The founding text on ethnic conflict, Horowitz's Ethnic Groups in Conflict (1985), suggests that a comprehensive theory must allow for both exogenous and endogenous factors. Horowitz acknowledges explanations based primarily on broader structural factors such as modernization theory but finds them lacking; instead, he focuses on attributes of the groups themselves, including their affiliation, organization and centralization, in the context of institutions of governance. Likewise, Varshney (2007) discusses the importance of institutionalist explanations (structural) as well as the essentialism–instrumentalist divide (endogenous) as factors leading to ethnic conflict. Other studies of ethnic conflict have emphasized the importance of elites' and masses' perspectives on “comparative group worth and legitimacy” based on traditions, symbols and historical experiences, while also acknowledging structural causes such as the security dilemma, and the catalysing presence of “political entrepreneurs” to take advantage of ripe circumstances (Wolff, 2006).
One of the few scholars to discuss the importance of endogenous characteristics of refugee groups, including ethnicity and nationalism, in affecting refugees' engagement in conflict is Milica Bookman (2002). Her focus is on the political–economic factors that lead to nationalism in relation to inter-ethnic competition and conflict between refugees and the CoO as well as the host state. These factors include competition for resources, economic niches and power. However, her study is limited to protracted refugee situations, placing them within regime-differentiated economic frameworks of their host states and the globalized international arena. Conflict is only tangential to that broader focus, and includes more analysis of criminal activity than political violence directed towards the CoO. She does not seek to explain conflict catalysed by refugees, nor the spread of civil war across borders.
Joly (2002) comes close to addressing the importance of refugees' political agency in her ideal-type dichotomy of refugee experiences in the land of exile. According to her theory, refugees belong to either an Odyssean or a Rubicon typology. Odyssean refugees are those who “were positively committed to the political struggle and to a project of society in their homeland; they also brought this project with them into exile … Return is their objective with the aim of continuing the project” (2002: 9). Rubicon refugees maintain “no collective project of society… Return for the purpose of settling back home is not envisaged within the framework of options for the future” (2002: 16). There are important differences in the way the two ideal-types integrate into host countries, but her theory is targeted towards refugee experiences in developed third-country resettlement states rather than states of first refuge, where opportunities for engagement in militarized conflict with the CoO are far more likely (Salehyan and Gleditsch, 2006).
Like Joly, an ethnographic study by Malkki (1995a) on Burundian Hutu refugees investigated different constructions of identity, meaning and symbology among those residing in a camp and those in an urban setting in Tanzania. Her study revealed stark differences: camp residents held more militant and polarized attitudes constructed around displacement, had adopted the refugee label, and identified themselves as Hutu first; while those in the urban setting had mostly assimilated into Tanzanian society, rejected the refugee label, and identified first as Burundian or even Tanzanian (a national identity) rather than as Hutu (an ethnic identity). While her work is an important contribution to understanding identity construction, it is mostly descriptive rather than theoretical, focusing primarily on the camp/urban distinction, with little explanation of the differences, and it does not address political violence or militarization explicitly. Nonetheless, it does highlight an intersection between endogenous and exogenous factors not addressed by the literature on refugee militarization: the relationship between residential environment and identity construction.2
Sociological study of dynamics of group engagement in violence can also offer important insights to the study of refugee militarization. First, a group is most likely to become a “conflict group”, one ideologically orientated towards and able to engage in organized conflict, when it achieves conflict solidarity – that is, “members not only recognize that their goals are incompatible with those of their opponents, but also have many grievances against them and are frustrated” – in addition to sufficient conflict resources (Bartos and Wehr, 2001: 81). In other words, there exists the motivation and resources for militarization. Both of those conditions could be met in refugee camps and communities in host states bordering the CoO. Next, identities are fluid, multiple and constructed (Brubaker, 2004; Goff and Dunn, 2004; Schlee, 2002, 2008a), and camps and communities full of refugees who were expelled due to persecution or participation in a civil war – circumstances in which a single common identity was an important element in the conflict – also facilitate other conditions for transformation of a group to a conflict group: increased interactions, similarity of culture and shared identity (Bartos and Wehr, 2001: Ch. 5), and collective memory of conflict (Cairns and Roe, 2003), which are further reified and made more salient by the presence of a threatening external enemy (Woehrle and Coy, 2000: Ch. 1).
The study of refugee militarization that has come closest to a comprehensive explanation incorporating both exogenous and endogenous factors, and that laid the theoretical foundation for this project, is Lischer's (2005) Dangerous Sanctuaries. The most important contribution made by her study is a rigorous theoretical and methodological treatment of refugee militarization. She theorizes that the likelihood of war diffusion across borders through refugees is most affected by: (1) circumstances surrounding the origins of the refugee crisis – whether from war and chaos, group persecution, or defeat in civil war; (2) the capability and will of the host state to provide security and demilitarize refugee camps; and (3) the presence of undifferentiated international humanitarian aid, which could be used to assist and support rebel movements. Lischer rejects earlier dominant explanations, especially socio-economic ones. Her theory brings refugee militarization towards a contentious politics framework, though without explicitly using such terminology. For instance, the securitization and demilitarization of camps reflects a broader question concerning political opportunities. Whether humanitarian aid is differentiated and monitored, as well as whether support is available from other actors such as a co-ethnic minority group in the host, is an example of resource mobilization. Her theory also highlights the presence of a militant group prior to expulsion from the CoO, which I subsume under a discussion of militancy entrepreneurs.
Other factors can be considered under these broader categories as well. Freedom of movement, organization and communication across refugee communities or camps must be included as another type of political opportunity; and militancy entrepreneurs can develop in exile, like the Palestinians, rather than only prior to expulsion. Next, although Lischer does begin to address the importance of endogenous factors with her consideration of the circumstances of expulsion, she neglects two more such motivating factors that are critical: first, whether there exists among the refugees an ethno-nationalist project that centres on the CoO; and second, whether the refugees are economically orientated towards the CoO, both of which will be discussed below.
A comprehensive theory: integrating motivation and structure
There are important distinctions of refugee militarization from social movements: First, social movements are traditionally domestic while refugee situations are inherently transnational and international; second, for those seeking to militarize refugees, there is no need for active politicization prior to mobilization, since the acts of war, expulsion and exile that led to asylum-seeking are inherently politicizing. By seeking asylum in a foreign country, groups are engaging in political behaviour and become political entities.
Refugee groups typically “desire to return to their homelands or find a permanent home elsewhere” (Bookman, 2002: 184). But who seeks the homeland and who aims to move elsewhere? To borrow Hirschman's (1970) famous terminology, the refugees' option of loyalty to the authorities in the CoO has been nullified by expulsion; the only choices remaining are exit – effectively abandon claims for citizenship and property left behind in the CoO, and seek resettlement3 – or voice, which represents indefinite perpetuation of the project of return through diplomatic channels, violent means or both.
Ironically, the theory that I propose in this paper is most closely summarized by the earliest attempt to describe refugee militarization, offered in a brief paragraph by Zolberg et al.: “Individuals in exile find that the most socially meaningful and economically rewarding activity is to join the warriors, and consequently move from the category of mere displaced persons into that of the politically active and conscious … [The refugee situation] also offers a new set of resources in a new situation which can be used by innovative political entrepreneurs to establish themselves” (1986: 166). Several important factors here are noteworthy. First, militarization is “socially meaningful”, which is related to a political motivation; second, there is an economic motivation; third, there are militancy entrepreneurs who bridge the gap between motivation and action; and fourth, there is a new and unique set of resources available for militant ends. The only major element of contentious political activity added in this paper is the existence of permissive political opportunities for militant activity.
The theory that I propose in this paper is summarized in Figure 1. Variables concerning motivation are under-theorized in the field of refugee studies and have traditionally been ignored by practitioners.4 The first two subfactors, a war of exclusion and an ethno-nationalist project, provide the basis for political motivation: grievances around which militancy entrepreneurs are able to “frame” mobilizing arguments and shape reality for potential supporters.5 In particular, in refugee situations it is easy to frame arguments around injustice and identity, while militarization offers hope of agency.6
Political motivation: war of exclusion
First, I use the terminology employed by Morris and Stedman (2008: 74–75) to ask whether the conditions under which civilians fled the CoO were a “war of exclusion”. Lischer (2005) specifies this variable with three possible values: situational, persecution or state-in-exile. However, her definition of situational refugees, the only ones not expected to militarize, is distinguished from the other categories primarily by their lack of political interest and involvement. A two-category scheme is more clear as it distinguishes between the war itself and motivations, which are better considered separately.
If one central purpose of the war was to displace a population, as in cases of ethnic cleansing, then refugees are more likely to be politically motivated for militarization. As Morris and Stedman note, however, this is not a necessary condition for militarization, since refugee groups that fled other types of wars can still militarize over time.
Political motivation: ethno-nationalism
The ethno-nationalist project centred on the CoO is one of the strongest motivations for refugees. An ethno-nationalist project is more relevant as a motivating factor for militarization by refugees than other identities because the definition of “refugee” is contingent on the existence of states, which are inextricably tied up with the concept of “nation” (Malkki, 1995b; Zolberg et al., 1986); while an ethnic dimension is often the identity on which the refugees were expelled. Other identities, such as those based on tribe, are less relevant to the study of refugee militarization because they do not necessarily tie in to conflict centred on physical inclusion and exclusion from the state.
The drawing of a connection between ethno-nationalism and increased likelihood of violent conflict is not a novel idea (e.g. Bookman, 2002; Brown et al., 2001; Brubaker and Laitin, 1998; Harff and Gurr, 2004; Varshney, 2002, 2003). Nationalism is defined in large part by distinction from the “other” and inter-ethnic competition for power and resources (e.g. Lake and Rothchild, 1998). A politicized ethnic group that has constructed an historical narrative and symbology of nationhood connected to their territory of (prior) residence has already established the foundations for nationalist aspirations towards self-sovereignty. The acts of persecution and expulsion that resulted in them becoming refugees then serve to solidify processes of inter-group conflict and intra-group unity, which further consolidate their ethno-national identity. Like other conflict-generated diasporas, such refugees maintain and perpetuate identities centred around “a very specific, symbolically [rather than instrumentally] important, and territorially defined ‘homeland'” (Lyons, 2006: 115). In a symbolic example, the Polisario leaders of the Saharawi in Algeria, refugees from the conflict with Morocco who have militarized, named all four of their refugee camps in Algeria after towns in Western Sahara in order to remind residents of their project of return (Bookman, 2002: 192, fn. 29).
A CoO without a rent-producing resource but which is identified as a national homeland can become even more precious in the eyes of exiles – and more worth fighting for – than a non-homeland CoO possessing such resources. The difference stems from their respective replacement value: substitute sources of revenue might be found elsewhere, bitter though the change may be, but a homeland can never be replaced. The dominant group in the CoO is often well aware of this dynamic. They see refugees' intent to return as a potential source of conflict and future instability:
Involuntary displacement was usually followed by the state-sanctioned destruction of homes and villages in order to eliminate the economic or emotional pull they might exert on the migrants (this formula was applied to the Kurds in Turkey, they Mayan peasants in Guatemala, and the Karen in Myanmar, among others). Needless to say, such displacement and concomitant destruction fosters anger and frustration coupled with a sense of injustice and a desire for revenge. When properly harnessed by ethnic political leaders, these feelings are a potent of nationalistic sentiment that can be used to develop a cadre of refugee-warriors. As long as the goals of refugee-warriors are to return to their countries and reclaim their lands, home country governments are justified in perceiving a threat from the encampments. (Bookman, 2002: 186–187)
Other refugee groups recognize that they are not a nation, or at least not one attached to the CoO. They make a rational decision that opportunities for security, stability, employment and perhaps education are better somewhere, in some cases anywhere, other than the CoO. Thus, assuming that living conditions are at least safe and stable in the country of first refuge,7 these latter groups will choose to put the CoO behind them and begin their lives anew. Those in this group, without a nationalist project either before or after the expulsion, and with little motivation to fight for the CoO, map on to Joly's Rubicon refugees. Clear examples of these dynamics can be found in the cases of Palestinian refugees, well known for militant activities directed at their non-rent-producing homeland, expelled from their long-time residences in Kuwait in 1991 and Iraq in 2003. In neither of the latter oil-rich cases did they militarize against their CoO (Lebson, 2009).
The next endogenous variable is an economic orientation towards either the CoO or the host state. Orientation towards the CoO will lead to a greater likelihood of militarization, while orientation towards the host will decrease it. This set of variables establishes the microfoundations of refugee groups' motivations by drawing on the “push” and “pull” theories of migration – the rational decision-making process by potential migrants of the relative utility of moving versus staying. Until a refugee situation is politically resolved, they exist in limbo, permanently settled in neither the CoO nor the host. The costs and benefits of orientating towards either the CoO or the host are continuously assessed. It is not unusual for refugee families to send a member back to the CoO periodically to check on security, living conditions and the status of their property. Looking at the host state, refugees consider whether they are able to lead a “normal” existence: Do they have legal access to courts, citizenship, free movement and residence? Do they have legal access to employment? Such accommodation offers refugees increased economic and social integration into the host country and could decrease motivation to return to the CoO. Conversely, political exclusion and repression are factors known to increase the risk of militancy domestically (Hafez, 2003),8 and the same logic could apply to refugees.
In addition to host-state policies, refugees also consider how the standard of living and employment opportunities in the host compare to those in the CoO. Poverty breeds need and has been shown to lead to militant action by insurgent groups (Fearon and Laitin, 2003), and it is usually considered a structural variable. I include it here as an endogenous factor because it is not poverty per se, but rather relative deprivation compared to the CoO, which could motivate support for militancy.9 Living conditions in camps that are substantially worse than in the CoO serve as a constant reminder, from refugees' point of view, of the injustice and violations that were committed upon them by agents of persecution in the CoO, potentially further increasing their motivation to support militancy. If conditions are, or were and could be again, significantly better in the CoO than in the host, the group will orientate towards the CoO.
The refugees' former economic base in the CoO can also affect their attachment to the land. Groups whose property and skills are easily transferable have less binding them to the CoO and consequently less to fight for; whereas refugee groups whose property and skills are less transferable, especially small farmers, are more likely to find it worthwhile to fight to return to the CoO. Likewise, if refugees face significant barriers to entry for employment opportunities in the host state, which is more likely when the economic bases of the two countries are substantially different, refugees will have more motivation to fight for the CoO. Another factor that must be considered is the “greed” motivation: if the CoO, and especially the region from which the refugees fled, relies on an economic base of oil or other rent-producing natural resources, refugees may have greater cause to fight for their home country.10
The final explanatory hypothesis in this set is the duration of stay in a refugee situation, neither repatriated to the CoO nor assimilated into the host state or a third-party resettlement state. Although refugees sometimes recreate social patterns, including class distinctions, upon settling in the host state (cf. Harpviken, 2009), in most cases the more protracted the refugee situation, the less important these original distinctions become. As in the case of Iraqis in Jordan, even the capital reserve of relatively wealthy refugees is eventually drained without incoming revenue to maintain it. This experience can decrease class and other cleavages among the group, create coalitions between formerly opposing clans or subgroups, and promote group cohesion by increasing the salience of their shared ethnic identity, refugee experience and external enemy.
The direct causes of protracted refugee situations are political: the persistence of persecution and/or government policies in the CoO that led to their exile and preclude their return; resistance of the host state to integrating and assimilating them; reluctance of third-party resettlement countries to accept large numbers of refugees; and in some cases, threats by external militant groups who derive legitimacy from the continuing refugee presence (Morris and Stedman, 2008). In protracted refugee situations, they are more likely to militarize, provide a source of rebel soldier recruitment, create tension with locals in the host state and contribute to regional instability (Loescher and Milner, 2005; Loescher et al., 2008). The reasons are partly related to identity construction, as “the longer a conflict goes unresolved the more likely identity-related issues will emerge, since the conflict becomes part of disputant self-understanding and part of how they are viewed by other parties” (Woehrle and Coy, 2000: 6). An example can be found in the case of Palestinian refugees in Jordan, who did not militarize until almost two decades after fleeing their CoO.
There is often some subset of the refugee group, or associated with it, who see the potential to mobilize refugees to support or engage in militant activity. I refer to such actors as “militancy entrepreneurs”. If a militant organization (claimed to have) acted on behalf of the refugee group prior to expulsion, they usually constitute such entrepreneurs. On the other hand, they could also develop in exile, as did the Palestinian Liberation Organization (PLO). As discussed earlier, militancy entrepreneurs are often able to make use of collective action frames based on the perceived injustice faced by refugees due to their persecution and exile, their right to land in the CoO as justified by their ethno-nationalist historical narrative, and the potential to gain agency and reclaim what is rightfully theirs through militarization. However, as noted above, the relationship between militancy entrepreneurs and refugees could be one of coercion, consensus or a combination thereof.
The last two sets of factors, political opportunities and resource mobilization, are well known to students of social movements (e.g. McAdam et al., 1996; Tarrow, 1998), rebellions and insurgencies. The first is the degree of permissiveness of political opportunities, which is affected by factors spanning both domestic (in the host state) and international domains. Host state policies include, as Lischer (2005) emphasized, whether refugee camps or communities are securitized and demilitarized; and whether they allow the political space for refugees and refugee organizations to freely move, communicate, organize and assemble within and between communities. UNHCR and other IGOs and NGOs on the ground in refugee camps can create permissive political opportunities by ignoring evidence of militant recruitment, training and diversion of humanitarian aid towards militant purposes, as they struggle to maintain missions of humanitarianism and neutrality. However, in many cases refugees are actively discouraged from political organization and activity not only by their host state – often in order to preserve positive relations with the CoO government (Mandal, 2003) – but also by the international refugee regime, particularly UNHCR. If the relationship between the CoO and host is one of opposition, then the host will be unlikely to repress refugee militarization, creating a more permissive opportunity environment. Finally, strong international engagement and pressure towards a resolution of the conflict that led to the refugees' expulsion can decrease political opportunities for militarization, although in some cases only temporarily as peace agreements break down.
The last set of factors is the potential to mobilize resources, which is a sine qua non of any contentious political action. Internationally, do other powerful states in the world or the region have an interest in supporting militant activity by refugees? For example, the United States supported Afghan refugees to militarize against their CoO in the 1980s; and South Africa supported counter-revolutionary forces in Angola and Mozambique (Zolberg et al., 1986). Other sources of resources include whether international humanitarian aid to the group is differentiated and closely monitored; and whether a diaspora supports militant action by refugees. Domestically, does the host state have an interest in funding militancy by refugees as a proxy against its neighbour?11 Can militants draw resources from other allies, such as a co-ethnic minority group, who might support violent political action?12 If so, then militancy entrepreneurs will be able to capitalize on the refugees' experiences, ethno-nationalism and orientation, facilitated by permissive political opportunities, to mobilize them towards militarization.
A call for case studies
The comprehensive theory proposed in this paper is only a starting point. In the next phase, qualitative case studies are needed to apply and test the theory by comparing, if possible, similar refugee situations that did and did not result in militarization. For example, why did the Palestinians in Jordan militarize in the late 1960s but not in the 1950s or since 1971?13 Case studies are the most appropriate method suggested by this theoretical framework for a number of reasons. First, the question posed seeks to understand how (capacities, opportunities, resources) and why (motivations, mobilization, leadership) refugee militarization is likely to occur. These questions are best answered with case studies. Second, rather than simply the presence or absence of militarization, it is necessary to trace its evolution, including how changing identities and ethno-nationalism affected the likelihood of militarization over time. Third, refugee militarization is a relatively contemporary phenomenon – the concept of refugee was only formalized after the Second World War – and the oldest case of refugee militarization dates back only to Burmese Kareni refugees in Thailand in 1949. Case studies, including interviews of former militants and actual refugees, may therefore be more valuable than archival analysis. Fourth, they can provide direct observation by the researcher of ongoing processes, including motivation for or against militarization, rather than second-hand accounts. Fifth, case studies encompass a wide variety of evidence, including interviews, observations and documents, which can help triangulate and support narratives. Sixth, it is difficult to distinguish the phenomenon of interest, militarization, from the surrounding context. Such distinctions, which are central to the theory presented here, are best achieved by in-depth observation and study. Finally, and related to the latter point, the theory seeks to model and assess complex causal relations with many moving parts, including issues of identity, ethno-nationalism, orientation and broader structural factors, which can best be discerned from case studies (Yin, 2009).
Reducing the risk of militarization
Application of this framework to better understand the factors leading to refugee militarization could help practitioners and policymakers make crucial decisions to decrease the risk of engagement in conflict by refugee groups. It also has the potential to help illuminate processes at work in other situations of group conflict, such as ethnic violence and conflictual state-minority relations.
What are the implications of this theory for host states? First, it is important to reiterate that this study seeks to supplement, not replace, theories on the environmental and structural factors affecting refugees' engagement in militarized conflict. This means that it will probably be found that housing large numbers of refugees in densely populated camps near the border with their CoO, without ensuring their demilitarization and physical security, and without carefully supervising the distribution of goods and aid into the camps, continues to invite danger. When refugee group politics lead to increased motivation and utility for conflict, and militancy entrepreneurs exist who are able to draw on refugees' experiences and ethno-nationalism to frame their arguments, the need to address these structural and environmental issues is even more urgent. The most important lesson to be drawn from this study for host states, as for all stakeholders, is increased awareness and vigilance regarding the increased likelihood of engagement in conflict for refugee groups when certain conditions are met. However, it is also crucial not to refuse entry or asylum to groups of refugees on this basis, which would be discriminatory and a violation of their human rights as well as international law, and could result in refugees' injury or death.
For NGOs and international organizations, including UNHCR, it is important to address the differing needs presented by refugee groups in the context of the ramifications that certain policies can have on the groups' political decisions. Until now, UNHCR has generally treated all refugees under the same rubric, emphasizing return at the earliest safe opportunity. However it is necessary to consider that some refugee groups will seek, even violently, return to their CoO; while others will show a strong preference to resettle in other countries in order to maximize opportunities for economic and educational gain, rather than return. This difference can have important implications for the refugee regime, including challenging the standard conceptualization of refugees as passive actors who universally prefer the status quo ante bellum.
When host states, IGOs, NGOs and other international stakeholders encounter a refugee situation in which a shared (ethnic) identity formed a basis for their expulsion, the refugees maintain an ethno-nationalist project and they are economically orientated to their CoO, then the circumstances will be ripe for militancy entrepreneurs to mobilize refugees towards militarization. Host state and international policymakers may be able to decrease the risk of militarization by:
1.Reducing political motivation:
proactively engaging in the political conflict in the CoO to reach a lasting peace agreement that will enable refugees to return home;
offering culturally sensitive counselling to refugees, especially teenagers and out-of-school youth, to reduce motivation for violent reprisal; and
incorporating training in non-violent conflict management in the educational systems of refugee camps and communities.
2.Reducing economic motivation:
improving living conditions, employment opportunities and legal rights of refugees in the host state to meet the terms of the 1951 Convention Relating to the Status of Refugees, regardless of whether the host state is a signatory; and
offering skill training to refugees when the economic bases of the CoO and host are substantially different, in order to facilitate economic integration in the host state, even temporarily.
3.Deterring militancy entrepreneurs:
training and employing refugees as police and intelligence forces to maintain security, identify and report the presence of small arms, and deter militarization in refugee camps and communities.
4.Preventing political opportunities:
actively demilitarizing and securitizing refugee camps and communities;
monitoring and minimizing political space in refugee camps and communities for militant organizations to organize, train, recruit and operate.
5.Preventing resource mobilization:
intercepting donor sources of financing and arms to militancy entrepreneurs;
using diplomatic resources to push sympathetic donor states and groups to end facilitation of militancy; and
monitoring and differentiating humanitarian aid to ensure that it is not diverted towards supporting militant groups.
While humanitarian concerns must continue to be met, this theory could help raise awareness among policymakers of refugees' personal and group agency. By acknowledging and expecting that refugees are actors who take independent political action, practitioners and policymakers might reconsider policies that seek to deter political activity. For example, they can establish alternative opportunities for political agency, such as town hall forums in camps and communities, refugee-run media, polling, civil society groups, ethnic cultural centres and other traditional forms of political engagement unique to each group. Such opportunities could constitute not only an outlet for political engagement and expression, but also valuable skill training and additional employment opportunities.
Finally, a more rigorous and comprehensive investigation into refugee militarization will offer insights into the study of contentious politics, including other types of group violence and state-minority relations, such as transnational rebellions, ethnic conflict, civil war, and nationalist and secessionist movements.
I offer special thanks to Kristian Berg Harpviken, as well as an anonymous reviewer and numerous other colleagues for their helpful suggestions.
Lischer (2000) found that, on average, there was political violence involving refugees in 11 states hosting refugees each year during the period from 1987 to 1998, out of an average of 82 states hosting 2,000 or more refugees per year.
However, inferring causality from this relationship may be subject to an endogeneity problem, as the refugees in the urban setting in Malkki's study were self-settled, suggesting that those inclined towards “cosmopolitan” perceptions may have self-selected to move to the urban setting.
It is important to note that options for resettlement are usually far beyond the political agency available to refugee groups. Those who choose resettlement may be quickly resettled or they may languish in camps for years while political considerations prevent both their onward migration and their permanent settlement in the country of first refuge.
UNHCR and other intergovernmental organizations (IGOs) and international non-governmental organizations (INGOs) that work with refugees have begun to acknowledge and quietly address the phenomenon of refugee militarization in the past decade, following the situation with Rwandan genocidaires in the mid-1990s (interview with senior UNHCR official, November 2010).
For discussions on framing in the social movement literature see, for example, Brysk (1995), McAdam et al. (1997) and Snow and Benford (1992).
Gamson (1995) discusses these three critical collective action frames in motivating social movement followers.
This assumption did not hold true for Rohingya refugees in Bangladesh, who were so harassed by Bangladeshi government forces that they felt life was actually preferable in their native Burma and so chose “voluntary” repatriation. Some later returned to Bangladesh (Lewa, 2003).
See also Lemarchand (2004) regarding intertwined aspects of exclusion, including political, economic and social dimensions, which he applies to refugees in the Great Lakes region of Africa.
Recalling the groundbreaking work of Gurr (1970), relative deprivation can be an important factor leading to increased motivation for group violence.
Collier and Hoeffler (2004) make the case that economic arguments such as these are the most persuasive predictors of civil war.
Gerdes (2006: 119-120) finds this factor to be the most important in predicting whether a host state will allow and/or facilitate the presence of refugee warriors.
Harpviken (2009) offers a reconceptualization of forced migration through the perspective of social network analysis, including the importance of ties with relatives or co-ethnics in the host state.
This author is currently (2010–2011) conducting in-depth case studies of Palestinian and Iraqi refugees in Jordan to apply this theoretical framework.