Refugees,ethnic power relations,and civil conflict in the country of asylum

Abstract

Many countries that face forced migrant inflows refuse to admit these uprooted people premised on negative externalities such as increased insecurity associated with refugee presence. Also, the academic literature on civil conflict identified refugee movements as a factor contributing to the regional clustering of war. Case-based evidence suggests that refugees can disturb the ethnic setup in the country of asylum and thereby trigger instability. To enhance the yet limited systematic understanding of the role of refugees in violent conflicts, this study examines the linkage between forced migrants, transnational connections, and ethnic civil conflict in the country of asylum with a large-N analysis, 1975–2013, arguing that ethnic power politics in the asylum state are determinant for intrastate conflict onset after a refugee influx. Statistical analysis finds that groups are particularly prone to conflict if they are excluded from governmental power and simultaneously host ethnic kin refugees, because a co-ethnic refugee influx enlarges the demographic and political leverage of the kin group, possibly resulting in clashes with other groups in the country. Hence, refugees alone do not consistently influence armed violence – only in combination with political tensions in the receiving country. Therefore, host governments should pursue inclusionary policies towards their population, to prevent dangerous instability, instead of closing borders or blaming refugees for domestic problems.

Keywords

civil conflict ethnicity refugees

Introduction

Although the Kurdish minority in Turkey welcomes co-ethnic Kurdish refugees from Syria, they are worried that the large influx of Syrian Sunni Arabs will upset the delicate ethnic balance in Eastern Turkey (International Crisis Group, 2016). Similarly, the Lebanese government and population fear that the Syrian refugee influx will disturb that country’s fragile sectarian balance (International Alert, 2015; Itani & Grebowski, 2013).

Refugee migration is a prominent topic in today’s world. Faced with the prospect of forced migrant inflows, many potential host governments place restrictions of varying levels on prospective admits. Potential host governments are worried that a higher refugee presence brings with it negative externalities such as increased insecurity. In particular, in countries of asylum where refugees differ from the local population in terms of religion, language or phenotypical features, governments may adopt anti-immigration policies. Yet, such policies assume that refugee flows generally increase instability. This overlooks the observation that most refugees never participate in violent activities.

Recent years have witnessed a heightened political debate regarding the effects of refugee inflows on domestic stability. Simultaneously, research on the risks associated with forced displacement has gained practical relevance. To date, however, the literature on refugees and intrastate conflict does not offer a clear answer to the question of how refugees affect domestic stability. This is particularly since the literature mainly consists of qualitative studies that focus on a few prominent cases (Adamson, 2006; Krcmaric, 2014; Lischer, 2005; Loescher & Milner, 2005; Newland, 1993; Whitaker, 1998). In an often-cited comparative study, Salehyan & Gleditsch (2006) find that refugee-receiving countries have a higher risk of experiencing civil war, owing to the destabilizing consequences for the host state’s economy, society, and security. More recently, the effect of internal and cross-border displacement on the risk of terrorist attacks has received academic attention (Bove & Böhmelt, 2016; Choi & Salehyan, 2013; Choi & Piazza, 2016; Milton, Spencer & Findley, 2013). Yet, even the studies that employ quantitative analysis across many countries fail to explain how refugee flows affect the risk of ethnic conflict. By focusing on the country level, such studies do not offer a view into the effect of different refugee groups on within-country ethnic balance. In particular, these studies neglect the decisive role of the host regime to prevent insecurity and have been hampered by a lack of systematic data on the ethnicity of refugees, the latter of which has recently become available.

Still, many examples from different world regions suggest that refugees increase ethnic instability in the receiving state. For instance, the massive influx of Afghan refugees into the Balochistan province of Pakistan in the 1980s was followed by ethnic conflict (Jamal, 2016). Belize’s ethnic Creole-Mestizo balance was complicated by the influx of Salvadorian refugees (McCommon, 1989). In Zaire, the arrival of over one million Hutus from Rwanda in 1994 led to a deterioration of the relationship among local Hutus, Tutsi, and other groups (Lischer, 2005: 13). Indeed, many political leaders that oppose hosting refugees base their claims on these cases. Thus, it is crucial to ask whether these cases were exceptions to a broader pattern of positive or null effects on stability, or if refugee inflows increase the risk of interethnic tensions on average.

Ethnic identities are fundamental in conflict processes because violence often escalates between different identity groups, owing to political or economic grievances. Consequently, I argue that the ethnicity of refugees is well worth systematic examination. Therefore, this article explores how refugee flows may affect the risk of armed ethnic conflict in the receiving state, depending on local ethnic politics and ethnic ties between refugees and the host population. By answering this question, I elucidate the risks of refugee-related violence and contribute to an important gap in conflict studies. The question of whether refugees disturb the ethnic balance in host states holds significant policy relevance. Knowing how to anticipate and prevent tensions in the context of refugee immigration clarifies where refugee protection is most needed and helps advance a more efficient international refugee response regime. Also, by disaggregating refugee caseloads according to their ethnic identity, I increase the knowledge on different refugee groups and better account for the considerable variation that characterizes global refugee flows.

Leveraging an updated version of the ethnicity of refugees (ER) dataset covering 1975 until 2013 (Rüegger & Bohnet, 2018), I show that the vast majority of refugees worldwide do not cause conflict in the asylum state. Yet, I also show that if refugees arrive in countries with pre-existing exclusion of a refugee group’s co-ethnics, there is a higher risk of exacerbated domestic tensions. My study attributes this reason to a two-step process: a refugee influx in this scenario leads to an increase in the size and political leverage of their ethnic group in the host country, including by sharing rebellious ideas or increasing the manpower of the ethnic group as a whole. This step, in turn, is perceived as a threat by politically dominant groups, who may pre-emptively or reactively suppress either refugees or their ethnic kin in the host state. The combined effect of co-ethnic refugees and political marginalization is robust to a number of confounding factors and alternative explanations.

This article is organized as follows. First, I review the literature on forced displacement and political violence. Next, I introduce the theoretical framework, focusing on how the ethnicity of refugees and ethnic power relations in the asylum state impact the likelihood of intrastate conflict onset. In the following section, I test the risk of civil conflict in the context of refugee inflows at the ethnic group level. I examine the claim that a refugee influx itself does not cause violence, but only does so in combination with the political marginalization of their local co-ethnics. The article concludes with a discussion of policy implications.

Previous research on forced migration and political violence

In simplified terms, the United Nations High Commissioner for Refugees (UNHCR, 2007: 3) defines a ‘refugee’ as ‘someone who is unable or unwilling to return to their country of origin owing to a well-founded fear of being persecuted for reasons of race, religion, nationality, membership of a particular social group, or political opinion’. Customarily, refugee status is granted to individuals who flee general violence since it is otherwise difficult to prove risks of facing persecution. Refugee movements have been the subject of intensive study, but it remains unclear how they impact the asylum state’s ethnic power politics. Besides the risk of an upset ethnic balance, previous research has argued that refugees contribute to regional instability through economic decline, pressures on public health or the import of rebels and weapons (e.g. Martin, 2005; Salehyan & Gleditsch, 2006; Zolberg, Suhrke & Aguayo, 1989).

To understand the role of refugees in armed conflict processes in asylum countries, it is important to understand why groups engage in conflict within countries. Conventional war studies find that ethnic groups that suffer from political marginalization are more prone to conflict (e.g. Cederman, Wimmer & Min, 2010). In this regard, the literature on external factors that impact internal conflict dynamics helps understand the role of refugees in such processes. Previous research shows that transborder ethnic kin (TEK) groups or forced migration movements contribute to the regional clustering of civil conflicts (Cederman et al., 2013; Cederman, Girardin & Gleditsch, 2009; Forsberg, 2014; Gleditsch, 2007; Salehyan & Gleditsch, 2006). According to this logic, if refugees contribute to instability in the asylum state, they only do so in combination with precipitating internal factors. However, the question of whether domestic structures mediate the effect of refugee flows on the conflict risk has received insufficient attention in the literature. Specifically, we know little about the role of the refugees’ ethnic group membership in conflict processes.

One important departure point is an extensive qualitative literature on refugees, which uses case studies to pinpoint the conditions under which forced migrants militarize in the asylum state. The refugee crisis in the Great Lakes region in the 1990s raised public and scientific awareness that refugees can become participants in conflict: perpetrators of the genocide in Rwanda mixed with refugees in Zaire and recruited in the refugee camps, which finally led to the outbreak of civil war in the years that followed (Whitaker, 2003: 212). Non-committed humanitarian relief that is provided in refugee camps can attract foreign fighters (Barber, 1997; Lischer, 2005; Terry, 2002). Refugees may also reinforce instability by spreading weapons or military skills across borders (Mogire, 2004; Muggah & Mogire, 2006; Weiner, 1992). These studies commonly find the host state’s willingness and capacity to prevent refugee manipulation play a decisive role in determining if rebels recruit refugees (Adelman, 1998; Gerdes, 2006; Lebson, 2013; Lischer, 2005; Zolberg, Suhrke & Aguayo, 1989). Likewise, I expect the risk of political tension to be determined by the asylum state’s ethnic power constellation and consequent willingness to accept co-ethnic refugees or reject ethnic alien refugees irrespective of whether refugees engage in violence.

On this note, the research that approaches international migration, including refugees, from a security perspective helps understand the host government’s reaction. In particular, host regimes disapprove of immigrants that allegedly engage in violence, constitute a cultural threat or imply an economic burden (Weiner, 1992). Consequently, host governments could enforce stricter border controls, demand assistance from the international community for ensuring security in the refugee settlements or use diplomatic channels, financial support or military power to decrease emigration from the source countries (Dowty & Loescher, 1996; Milner, 2000; Teitelbaum, 1984; Weiner, 1992).

A comparative study by Salehyan & Gleditsch (2006) links refugees to the onset of intrastate conflict in the host country. They find that states receiving refugees from neighbors have a higher risk of experiencing civil war. As a potential explanation of these statistical findings, they argue that refugees diffuse rebel networks, import weapons and ideologies, increase economic problems, and change the ethnic structure. Similarly, case-based evidence suggests that states where refugees disturb the ethnic balance are more prone to experiencing conflict (Adamson, 2006; Goldstone, 2002; Loescher & Milner, 2005; Newland, 1993; Weiner, 1992). This work has no doubt considerably advanced the literature on refugees and conflict. While earlier research on refugee militarization explains instances where refugees directly and actively engaged in violent activities, Milner (2000), Steputat (2004), and Salehyan & Gleditsch (2006) also consider indirect threats for the asylum state, such as economic decline and ethnic shifts, where the passive presence of refugees increases the conflict risk. However, existing research neglects the role of the ethnic identity of refugees and the host population, thus offering little direct insight as to how identity might mediate conflict risk. Moreover, due to the country-level approach used in most of the extant literature, we know little about the relationship between refugees and different population groups in the host state. Drawing on new quantitative data on the ethnicity of refugees (Rüegger & Bohnet, 2018), this study helps to fill this gap by testing whether refugees alter the ethnic composition in asylum states and, if so, how such a change affects conflict processes. I expect that changing demographics can create concerns about the identity of a country and also can shift the intergroup power balance.

More recently, several scholarly articles have examined the link between refugees and terrorism with large-N analyses. These studies support earlier case-based research on refugee violence. First, they find that terrorist attacks, similar to other forms of political violence, are more common among poorly treated refugee groups (Milton, Spencer & Findley, 2013), and that terrorist groups tend to misuse humanitarian aid that is intended to help refugees (Choi & Salehyan, 2013). Second, they show that overall, states that host refugees and immigrants are not associated with a higher risk of terrorism (Bove & Böhmelt, 2016). Yet, violence directed against refugees becomes more likely after terrorist attacks owing to a scapegoating effect (Savun & Gineste, 2019).

I argue that refugee flows and conflicts are generally local and subnational phenomena, and that it is therefore crucial to use more disaggregated measures. My focus on ethnic identities of refugee groups builds on the forced migration studies that test the negative effects attributed to refugees beyond the state level. Concentrating on the geographic location of refugee or IDP settlements in Africa, both Fisk (2014) and Bohnet, Cottier & Hug (2018) find an increased risk of conflict onset around these locations, as well as a higher likelihood of indiscriminate violence against civilians and communal conflict (Fisk, 2018, 2019).

In sum, previous scholarship on forced displacement and conflict suffers from several shortcomings that complicate the formulation of effective policy recommendations on how to prevent destabilization and violence in refugee-receiving countries. A better insight in these dynamics is critical to help governments to understand that the actual risk of accepting refugees is far lower than is too often perceived. Although the disturbance of the ethnic balance is often mentioned as one of the negative externalities associated with refugee migration, the theoretical understanding of the refugees’ impact on local ethnic politics is limited.

The logic of ethnic conflict in refugee-receiving countries

Do refugees increase the risk of ethnic conflict in the asylum state? And if so, can this be attributed to changes in the ethnic balance? To answer these questions, this section theorizes how refugees with ethnic kinship ties to the local population impact the risk of ethnic conflict onset. In sum, the likelihood of intrastate conflict following sustained refugee immigration strongly depends on ethnic power relations in the host state.

The constructivist approach to ‘ethnicity’ offers a good starting point to explain why the ethnicity of refugees is important. Ethnic groups are defined as self-perceived communities with a shared culture and a common ancestry. This socially constructed ethnic group membership can be based on a common language, religion or somatic features (see Cederman, Wimmer & Min, 2010: 98–99). The group-based persecution and the experience of flight often strengthen the ethnic identity of a refugee group (Lebson, 2013; Lischer, 2005). Consequently, I suppose that refugees not only are strongly aware of their ethnic group membership but also are perceived as a distinct ethnic community in the country of asylum.

Country borders do not restrict the sense of being an ethnic community and the loyalty towards group members (Davis & Moore, 1997; Saideman, 2002). In fact, refugees often flee along ethnic or colonial ties to neighboring countries relying on established networks and low assimilation costs (Moore & Shellman, 2007; Newland, 1993; Schmeidl, 1997). Of the 21.3 million people registered as refugees at the end of 2015, more than 85% resided in countries neighboring their home state (UNHCR, 2015) and more than 40% found co-ethnics in the asylum state (Rüegger & Bohnet, 2018: 8). Thus, it is important that refugee flows along ethnic linkages and their consequences are studied. Ethnic kinship raises the receptiveness of refugees among group members in the host country and facilitates integration. Simultaneously, it reinforces pressures if the country is inhabited by more than one ethnic group. State authorities usually welcome ethnic kin refugees because they strengthen their superior position towards both internal challengers and foreign regimes, while they often refuse to accept co-ethnic refugees of minority groups (Jacobsen, 1996; Loescher & Milner, 2005; Newland, 1993). For example, Kenya repeatedly planned to close the border with Somalia owing to alleged rebel infiltration and illegal immigration and to shut down the Dadaab camp, where more than 300,000 mostly Somali refugees settled in 2016 ( The Guardian, 2016). Similarly, the Pakistan government refuses to host Urdu-speaking Bihari refugees from Bangladesh in the Sindh province out of fear of ethnic clashes between the majority Sindhis and minority Mohajirs, who also speak Urdu (Khalidi, 1998: 346).

These two examples represent the many countries that deter refugees. Yet, empirical evidence suggests that the realities on the ground frequently do not reflect the government’s rhetoric: forced migrants still arrive. Indeed, the vast majority of countries admit refugees, either because they signed the UN Refugee Convention and observe the principle of non-refoulment (UNHCR, 2007: 32), or because they are unable to control their borders. Border closings often are temporally limited. Furthermore, when immigration controls intensify, borders nevertheless remain porous. Thus, refugee groups that are perceived as dangerous for the host’s security often enter the country.

I argue that analysis of the civil conflict risk in asylum states should consider the ethnicity of refugees. This builds on the many studies that highlight the importance of ethnicity in conflict research. Some of the negative externalities ascribed to refugee communities, including the impact on the ethnic balance and the diffusion of rebellious ideologies or insurgent networks, presumably depend on ethnic connections between refugees and the host population, as well as on the ethno-political power constellation in the asylum state. In particular, the negative factors associated with kin refugees influence politically marginalized ethnic groups, as I elaborate below. Consequently, only analysis at the ethnic group level, as opposed to the state-level approaches in previous research on refugees and conflict (e.g. Salehyan & Gleditsch, 2006), allows us to capture the effect of refugees on domestic political processes in receiving countries.

While all refugees alter the demographic constellation in a country, as non-citizens, they usually do not concern national power politics. Only refugees who share ethnic kinship ties with the local population, or a part of it, impact ethnic power politics owing to prospects of long-term displacement and future naturalization. Refugees demographically strengthen their co-ethnic groups, which increases the kin group’s leverage in state politics but simultaneously challenges the other groups in a country. This imperils the ethnic stability and increases the risk of political violence. First, the refugees’ kin group could revolt thanks to increased political strength. Second, other groups could attack the refugees’ co-ethnics owing to the threats implied by demographic changes. Thus, the effect of co-ethnic refugees on the ethnic balance is most salient in multi-ethnic states, as in Malaysia in the late 1970s, when the ethnic Chinese boat refugees from Vietnam were accused of troubling the political equilibrium between local Chinese and Malays (Stein, 1979: 717). Overall, this is an indirect mechanism, where co-ethnic refugee presence causes destabilization, without refugees actively engaging in violent activities.

Moreover, the local co-ethnic host’s confrontation with the suffering of transnational kin refugees can increase dissatisfaction. These grievances transfer to the asylum state if the authorities are incapable or unwilling to support the refugees, which can escalate to local political conflict. For example, in Macedonia in 1999, the question of whether or not to accommodate the Kosovar refugees led to intra-elite fighting between the majority Macedonians and the minority Albanian governmental representatives (BBC News, 1999). The Macedonian authorities, who saw their demographic predominance at risk, were hostile towards the refugees. This increased discontent among local Albanians, which in turn boosted support for Albanian nationalism and the insurgency two years later, notably after virtually all Kosovar refugees had returned home.

Furthermore, kin refugees can directly threaten the host state’s security: if refugees import rebellious ideologies or weapons, they are more likely to share them with co-ethnics thanks to a common language, habits, and mutual trust that facilitate interaction. For the same reasons, rebel groups tend to recruit among co-ethnic refugees. Also, new information on political opportunities can increase discontent with the status quo.

Co-ethnicity between refugees and the host population is a precondition for demographic changes and an upset ethnic composition. I expect local kin groups to have a higher chance of being both the target and initiator of violence during phases of refugee immigration. Therefore, I derive the first hypothesis:

Hypothesis 1: Groups with co-ethnic refugees in their country have a higher risk of experiencing civil conflict than groups without co-ethnic refugees.

As stated above, the risk of ethnic clashes is higher in multi-ethnic states with a fragile balance between the constituting groups. In this regard, a group’s political status, that is, exclusion or inclusion in the state government, is a primary cause of conflict involvement. In particular, large marginalized groups are prone to conflict (Cederman, Wimmer & Min, 2010). Co-ethnic refugees of excluded groups can more easily disrupt political stability by intensifying grievances among their co-ethnics and increasing their opportunities to exert violence, which will be explained in the following. Hence, refugees do not necessarily create tensions, but their presence intensifies previously existing ones. Therefore, I distinguish three refugee types: first, refugees without ethnic ties to the country of asylum who might be seen with suspicion but are unlikely to impact domestic politics; second, refugees with ethnic linkages to a politically powerful group who mostly are welcomed and adequately accommodated; third, ethnic kin refugees of politically marginalized groups who are perceived as a peril to internal stability and often suffer from neglect.

A host government is more inclined to assist ethnic kin refugees who do not challenge its dominance, whereas other ethnic refugee groups are likely to receive less support.¹ This might increase grievances among the refugees and their allegiant local kin. Consequently, the political context determines refugee accommodation and thus, the likelihood of a conflict, because the receiving state must be willing and capable to shelter the refugees (Lischer, 2005: 18–19).² For instance, the host government can determine whether or not to allow humanitarian agencies to help the refugees, such as the Indian government that banned NGOs from working in refugee camps hosting Tamils from Sri Lanka in 1993 (Refugee Council, 1999: 12–13). The lack of relief is one of the main reasons for insecurity and violence among refugees (Crisp, 2000: 70). Additionally, insufficient assistance creates economic inequalities when the political elite is unwilling to share resources with a different ethnic refugee group. Consequently, the refugees, their co-ethnics, and other groups in the country may fight over resources.

Furthermore, co-ethnic refugees increase the likelihood that marginalized groups revolt for two main reasons. First, as an indirect mechanism, the excluded group gains demographic and political weight thanks to a larger size and might finally feel strong enough to initiate an insurgency. Also, constituting a higher share of the country population thanks to co-ethnic refugees might increase the perception of unequal treatment and underrepresentation, which is a major cause of grievances and consequent insurgencies. In contrast, an already powerful governmental ethnic group is not motivated to start violence solely because of increased group size. Second, aggrieved ethnic groups are more receptive to new, possibly violent ideologies that are imported by refugees or other immigrants. Likewise, marginalized groups have incentives to use weapons or manpower for their own claims against the government; hence, they are amenable to diffused military equipment and direct support by co-ethnic refugees. For example, Rwandan Tutsi refugees aligned with the insurgents in the Mulelist uprising in Zaire in the 1960s (Gerdes, 2006: 39).

Simultaneously, by increasing their kin’s leverage, co-ethnic refugees challenge the predominance of other groups in a country. Consequently, the powerful groups could increase pressures and exclusionary policies on the refugees’ kin, including the use of violence against the perceived contester. Thus, groups with co-ethnic refugees in the asylum state have a higher risk of experiencing aggression by the authorities. For instance, the discrimination of Kenyan ethnic Somali intensified after the immense influx of Somali refugees (Kagwanja & Juma, 2008: 223). Also, refugees that are connected to a controversial group might directly be attacked, such as in Iraq in 2006, when approximately 170 Palestinian refugees were killed by Shiia militias who associated the Palestinians with the larger Sunni insurgency (US Committee for Refugees and Immigrants, 2007).

Ethnic power politics in the country of asylum determine the effect of kin refugees on internal conflict. The local co-ethnics could be either the perpetrators or the victims of political violence, which leads to the following hypothesis:

Hypothesis 2: Politically excluded groups with co-ethnic refugees in their country have a higher risk of experiencing civil conflict than politically excluded groups without co-ethnic refugees or included groups.

Analysis

I test the hypotheses using logistic regression models with the onset of ethnic civil conflict as the dependent variable. The unit of analysis is the ethnic group-year, with the standard errors clustered at the country.³ The analysis covers the years from 1975 to 2013. The sample includes all ethnic groups listed in the EPR-ETH dataset (Vogt et al., 2015).

Operationalization

The dependent variable measures if an ethnic group experiences civil conflict onset in a given year, as obtained from the ACD2EPR dataset (Wucherpfennig et al., 2012) that links ethnic groups to conflicts coded in the UCDP/PRIO Armed Conflict dataset (Gleditsch et al., 2002). Armed conflict is defined ‘as a contested incompatibility that concerns government or territory or both where the use of armed force between two parties results in at least 25 battle-related deaths. Of these two parties, at least one is the government of a state’ (Gleditsch et al., 2002: 618–619). If a conflict began during an ethnic group-year, the dependent variable was coded 1, and if not, 0. Ongoing conflict years were removed from the analysis. A total of 174 conflict onsets are recorded for the 20,736 group-years included in the analysis, among these, 25 group-years experienced a conflict onset while also receiving ethnic kin refugees.⁴

To test Hypothesis 1, the independent variable indicates whether or not a group hosts co-ethnic refugees from at least one country of origin. Data on the ethnicity of refugees were drawn from the ER dataset (Rüegger & Bohnet, 2018) and linked to the EPR-ETH and TEK datasets. For example, the Ogaden minority in Ethiopia has hosted co-ethnic refugees from Somalia since the late 1980s. Given that the original UNHCR dataset provides end-year figures, the variable is lagged one year to account for temporal dependence.

Hypothesis 2 assumes the impact of kin refugees to be conditional on the power status of a group. Therefore, I include an interaction term of the political status with refugee presence. Information on the ethnic groups’ political power was obtained from EPR-ETH. I distinguish between politically included groups with power at the national level, who are the groups that EPR-ETH codes as monopoly, dominant, senior or junior partner, and excluded ethnic groups without access to state power, that are recorded as powerless, discriminated or self-exclusionist.⁵

To assess the size of ethnic kin refugees, based on the information on the estimated refugee group sizes in the ER dataset, I created a categorical variable that measures whether the group hosts a small (> 0 and < 10,000), medium (>= 10,000 and < 100,000) or large (>= 100,000) number of ethnic kin refugees, or none.⁶ If a group received co-ethnic refugees from more than one country, the sum of all countries of origin was taken. For example, in some years, the Kurds in Iraq hosted kin refugees from Turkey and Iran. Considering the limitations of quantitative refugee data, such as imprecision owing to unregistered refugees and political incentives to over- or underestimate refugee numbers, and the problems of categorizing refugees into socially constructed ethnic groups, I refrain from using absolute figures. This variable is highly right-skewed, as most ethnic groups do not host any kin refugees. Out of 730 ethnic groups in the sample, 122 received ethnic kin refugees, during a total of 1,119 out of 20,736 group-years. Of these, 36% host a small number of co-ethnic refugees, 45% between 10,000 and 100,000 refugees, and 19% receive more than 100,000 co-ethnic refugees, including the Pashtuns in Pakistan who hosted more than a million co-ethnic refugees from Afghanistan during several decades.

Numerous control variables account for alternative explanations of the outbreak of armed conflict. In addition to political exclusion or inclusion, I control whether the group was downgraded recently since such reversals are known to increase grievances and thus, the risk of conflict (Petersen, 2002). Further, the relative size of the group compared to the groups that have access to central state power is added, because large excluded groups have an increased risk of conflict (Cederman, Wimmer & Min, 2010). The war history of the ethnic groups is considered by including a dichotomous covariate measuring whether the group participated in previous conflicts. Earlier conflicts explain current tensions and consequent reoccurring war. To check established findings that refugees generally increase the risk of conflict in asylum states (Salehyan & Gleditsch, 2006), I include a binary measure of whether the country hosted any non-co-ethnic refugees from neighboring states in the previous year. Further, I oversee whether a civil conflict involving any other group is ongoing in the country during the preceding year, which is a strong indicator of political instability. In addition, I control for ongoing conflicts in neighboring countries. This variable is lagged by one year. Conflicts may spread owing to instability in the region, neighboring conflict may directly impact internal conflict (Sambanis, 2001: 275), and neighboring conflict may increase the number of refugees. Further, at the country level, I control for GDP per capita (logged) and population (logged) (Feenstra, Inklaar & Timmer, 2015). Poor countries have a higher risk of civil war and less capacity to accommodate refugees, since providing shelter and food depends on financial and natural resources. Finally, I include a peace-years variable with squared and cubed terms to control for temporal dependence (Carter & Signorino, 2010).

Results

Table I presents the results of the regression models. Model 1 includes the binary coefficient for co-ethnic refugees and the control variables, as suggested by Hypothesis 1. Model 2 interacts ethnic kin refugees with political exclusion to test Hypothesis 2. The third model incorporates the categorical variable of different co-ethnic refugee group sizes. In Model 4, I added the binary indicator of whether the group’s country receives other, non-kin refugees.

Table I.

Logistic regression

	Model 1	Model 2	Model 3	Model 4
Co-ethnic refugees	0.604*	0.195		0.651*
	(0.254)	(0.497)		(0.259)
Co-ethnic refugees * exclusion		0.553
		(0.549)
Co-ethnic refugees no.			0.297**
			(0.108)
Refugees non-co-ethnic				0.083
				(0.200)
Excluded group	0.841***	0.756***	0.834***	0.839***
	(0.204)	(0.218)	(0.207)	(0.204)
Downgraded	1.630***	1.633***	1.633***	1.635***
	(0.224)	(0.221)	(0.224)	(0.226)
Relative group size	1.096**	1.070**	1.126**	1.087**
	(0.375)	(0.377)	(0.369)	(0.383)
War history	1.113***	1.124***	1.122***	1.108***
	(0.278)	(0.275)	(0.275)	(0.279)
Ongoing conflict	0.871***	0.872***	0.859***	0.874***
	(0.222)	(0.222)	(0.222)	(0.224)
Neighbor conflict	0.145	0.145	0.141	0.141
	(0.251)	(0.253)	(0.253)	(0.250)
Population	0.075	0.073	0.073	0.071
	(0.087)	(0.086)	(0.087)	(0.086)
GDP	−0.274**	−0.275**	−0.272**	−0.263**
	(0.091)	(0.090)	(0.091)	(0.086)
Peace years	−0.313***	−0.310***	−0.314***	−0.313***
	(0.049)	(0.050)	(0.049)	(0.049)
Peace years²	0.011***	0.011***	0.011***	0.011***
	(0.002)	(0.002)	(0.002)	(0.002)
Peace years³	−0.000***	−0.000***	−0.000***	−0.000***
	(0.000)	(0.000)	(0.000)	(0.000)
Constant	−2.798**	−2.729**	−2.775**	−2.879**
	(0.977)	(0.977)	(0.973)	(0.942)
Observations	20,280	20,280	20,280	20,280
χ²	389.097	441.872	379.207	391.456

Country-clustered standard errors in parentheses. ^† p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001.

The results in Model 1 suggest that a group with kin refugees has an increased chance of experiencing an ethnic conflict ceteris paribus: the logit coefficient is positive and significant at the 0.05 level. This supports Hypothesis 1 postulating that ethnic kin refugees can act as a conflict trigger. Figure 1 displays the predicted probability of civil conflict for groups with and without co-ethnic refugees, including 95% confidence intervals. Although there is a wide error band for groups that host ethnic kin refugees, owing to fewer observations, their substantial risk of conflict, which is on average 5.82%, is almost three times higher compared to groups without refugees that have a net conflict risk of 2.10%. This being a binary measure, we learn that the presence of any co-ethnic refugees increases the risk of conflict in the country of asylum. Also, Model 1 confirms several results from previous studies: politically excluded, recently downgraded, and relatively large ethnic groups are significantly more likely to revolt. Past conflict and ongoing conflict in the country in the previous year yield the same effect. Moreover, group involvement in conflict appears to be more likely in poor countries.

Model 2 includes the interaction coefficient of kin refugees and political exclusion. Figure 2 shows the difference in the probability of conflict onset between hosting and not hosting kin refugees for politically excluded and included groups with the other covariates held at median or mean. The difference between receiving or not receiving kin refugees is negligible for politically included groups. Yet, the impact of kin refugees on the

Figure 1.

Net effect of co-ethnic refugees on predicted probabilities of civil conflict onset

Figure 2.

First difference in the effect of co-ethnic refugees on the probability of civil conflict for included and excluded groups

probability of conflict is significant for marginalized ethnic groups. This supports Hypothesis 2 that the effect of refugees on conflict is mainly driven by political exclusion of the refugees’ co-ethnic groups, that already have a higher risk of conflict. According to the theoretical expectation, a politically excluded ethnic group that is demographically supported by kin refugees gains political leverage, thereby challenging the other groups in a country.⁷

To further examine the impact of refugees on the ethnic balance in asylum states, Model 3 incorporates a categorical measure of the co-ethnic refugees’ size, differentiating between small, medium, and large co-ethnic refugee groups, or none. Owing to the imprecision of refugee count variables, especially in combination with ethnic categorizations, this rough variable is chosen over actual numbers, which would only pretend to be more precise.⁸ The variable’s positive and significant effect hints that larger refugee groups have a stronger impact on conflict, which supports the idea that cases where the ethnic balance is heavily upset, in particular, observe a higher likelihood of conflict.⁹

To compare the effect of co-ethnic refugees to the effect of other refugees in the group’s country, in Model 4, I added the covariate for non-kin refugees. The insignificant coefficient reveals that non-co-ethnic refugees in a state have no effect on the probability of conflict at the ethnic group level, which contradicts Salehyan & Gleditsch’s (2006) country-level finding that refugees in general increase the risk of violence. Yet, the effect of co-ethnic refugees remains positive and significant. This is an important finding as it shows that refugees do not randomly cause conflict in the asylum state. Similarly,

Table II.

Sensitivity regression, part 1

	Model 5	Model 6	Model 7	Model 8	Model 9	Model 10
Co-ethnic refugees > 10,000	0.701**	0.193
	(0.266)	(0.767)
Co-ethnic refugees > 10,000 * exclusion		0.652
		(0.785)
Co-ethnic refugees from neighbor			0.677**	0.300
			(0.261)	(0.523)
Co-ethnic refugees from neighbor * exclusion				0.508
				(0.553)
Co-ethnic refugees					0.592^†	0.251
					(0.316)	(0.520)
Co-ethnic refugees * exclusion						0.469
						(0.564)
SE Europe & FSU					−0.309	−0.289
					(0.791)	(0.787)
Sub-Saharan Africa					0.145	0.137
					(0.817)	(0.814)
Arab world					0.340	0.324
					(0.783)	(0.783)
Asia					0.207	0.211
					(0.788)	(0.783)
Latin America					−0.503	−0.506
					(0.853)	(0.849)
Refugees non-co-ethnic	0.053	0.051	0.088	0.087	0.031	0.035
	(0.196)	(0.197)	(0.200)	(0.200)	(0.226)	(0.225)
Excluded group	0.833***	0.769***	0.834***	0.758***	0.872***	0.802***
	(0.209)	(0.219)	(0.205)	(0.218)	(0.204)	(0.217)
Downgraded	1.628***	1.635***	1.635***	1.638***	1.642***	1.644***
	(0.223)	(0.220)	(0.226)	(0.223)	(0.231)	(0.228)
Relative group size	1.161**	1.155**	1.075**	1.054**	0.960*	0.948*
	(0.370)	(0.371)	(0.385)	(0.385)	(0.376)	(0.374)
War history	1.126***	1.130***	1.105***	1.115***	1.043***	1.056***
	(0.275)	(0.272)	(0.281)	(0.278)	(0.300)	(0.298)
Ongoing conflict	0.856***	0.863***	0.874***	0.875***	0.825***	0.828***
	(0.222)	(0.221)	(0.224)	(0.223)	(0.232)	(0.231)
Neighbor conflict	0.147	0.149	0.141	0.140	0.083	0.084
	(0.253)	(0.254)	(0.250)	(0.252)	(0.262)	(0.264)
Population	0.071	0.069	0.071	0.069	0.056	0.053
	(0.086)	(0.086)	(0.086)	(0.086)	(0.084)	(0.084)
GDP	−0.262**	−0.264**	−0.263**	−0.263**	−0.221*	−0.224*
	(0.087)	(0.086)	(0.086)	(0.085)	(0.093)	(0.092)
Peace years	−0.314***	−0.312***	−0.313***	−0.310***	−0.316***	−0.314***
	(0.049)	(0.049)	(0.049)	(0.050)	(0.049)	(0.049)
Peace years²	0.011***	0.011***	0.011***	0.011***	0.011***	0.011***
	(0.002)	(0.002)	(0.002)	(0.002)	(0.002)	(0.002)
Peace years³	−0.000***	−0.000***	−0.000***	−0.000***	−0.000***	−0.000***
	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)
Constant	−2.858**	−2.809**	−2.880**	−2.817**	−2.970*	−2.892^†
	(0.954)	(0.950)	(0.942)	(0.948)	(1.476)	(1.478)
Observations	20,280	20,280	20,280	20,280	20,280	20,280
χ²	374.373	470.929	394.476	445.746	456.726	492.657

Country-clustered standard errors in parentheses. ^† p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001.

Table III.

Sensitivity regression, part 2

	Model 11	Model 12	Model 13	Model 14	Model 15	Model 16
Co-ethnic refugees	0.634*	0.264	0.517^†	0.176	0.650*	0.239
	(0.256)	(0.506)	(0.280)	(0.516)	(0.259)	(0.517)
Co-ethnic refugees * exclusion		0.502		0.463		0.557
		(0.558)		(0.559)		(0.548)
TEK group	−0.123	−0.118	−0.407	−0.400
	(0.212)	(0.213)	(0.318)	(0.321)
TEK conflict	0.287	0.267	0.295	0.276
	(0.212)	(0.217)	(0.212)	(0.216)
Relative TEK size			2.594^†	2.562^†
			(1.391)	(1.385)
Relative TEK size²			−2.591*	−2.562*
			(1.229)	(1.220)
Refugees from rival state					0.889	0.897
					(0.704)	(0.699)
Refugees non-co-ethnic	0.070	0.069	0.055	0.055	0.077	0.076
	(0.201)	(0.202)	(0.204)	(0.204)	(0.202)	(0.202)
Excluded group	0.825***	0.750***	0.824***	0.755***	0.832***	0.746***
	(0.202)	(0.216)	(0.203)	(0.215)	(0.206)	(0.220)
Downgraded	1.647***	1.649***	1.647***	1.650***	1.634***	1.636***
	(0.224)	(0.222)	(0.232)	(0.229)	(0.227)	(0.224)
Relative group size	1.057**	1.035**	1.121**	1.100**	1.109**	1.084**
	(0.384)	(0.385)	(0.378)	(0.379)	(0.388)	(0.389)
War history	1.097***	1.107***	1.098***	1.109***	1.108***	1.120***
	(0.276)	(0.274)	(0.283)	(0.281)	(0.279)	(0.276)
Ongoing conflict	0.864***	0.867***	0.858***	0.860***	0.873***	0.875***
	(0.225)	(0.224)	(0.225)	(0.224)	(0.224)	(0.223)
Neighbor conflict	0.119	0.121	0.118	0.119	0.137	0.137
	(0.245)	(0.248)	(0.242)	(0.245)	(0.250)	(0.253)
Population	0.070	0.068	0.095	0.093	0.075	0.073
	(0.082)	(0.082)	(0.094)	(0.094)	(0.086)	(0.086)
GDP	−0.261**	−0.262**	−0.267**	−0.267**	−0.268**	−0.269**
	(0.083)	(0.083)	(0.085)	(0.085)	(0.085)	(0.085)
Peace years	−0.312***	−0.310***	−0.307***	−0.305***	−0.312***	−0.310***
	(0.049)	(0.050)	(0.048)	(0.048)	(0.049)	(0.050)
Peace years²	0.011***	0.011***	0.011***	0.011***	0.011***	0.011***
	(0.002)	(0.002)	(0.002)	(0.002)	(0.002)	(0.002)
Peace years³	−0.000***	−0.000***	−0.000***	−0.000***	−0.000***	−0.000***
	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)
Constant	−2.793**	−2.733**	−3.080***	−3.020**	−2.879**	−2.810**
	(0.902)	(0.906)	(0.933)	(0.938)	(0.941)	(0.945)
Observations	20,280	20,280	20,280	20,280	20,280	20,280
χ²	409.349	450.494	486.402	527.686	421.154	457.308

Country-clustered standard errors in parentheses. ^† p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001.

Buhaug & Gleditsch (2008: 228) find that the number of refugees from countries that experience conflict has no significant effect on conflict contagion to neighboring states. My results highlight that refugees may contribute to increasing interethnic tensions if they have linkages to groups that sustain political marginalization.

Sensitivity analysis

To check the robustness of the findings, I estimated several additional models displayed in Table II and Table III. Based on the sensitivity models that include interaction terms, Table IV discloses the first difference in the effect of co-ethnic refugees on the probability of civil conflict for included and excluded groups.

Table IV.

Sensitivity models, first difference in the effect of co-ethnic refugees on the probability of civil conflict for included and excluded groups

	Model 6	Model 8	Model 10	Model 12	Model 14	Model 16
Included	0.001	0.002	0.001	0.001	0.001	0.001
	(0.004)	(0.003)	(0.003)	(0.003)	(0.002)	(0.003)
Excluded	0.012*	0.011*	0.010^†	0.010*	0.008†	0.011*
	(0.005)	(0.005)	(0.006)	(0.005)	(0.005)	(0.004)

Standard errors in parentheses. ^† p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001.

First, in Model 5, I coded the binary coefficient for co-ethnic refugees more restrictively, in that the variable is 1 when the number of refugees is estimated to be more than 10,000 kin, and otherwise 0. This allows me to minimize the risk of overestimating the effect of kin refugees that were sampled based on imprecise estimation techniques by either the UNHCR or the ER dataset and to exclude the smallest refugee groups that apparently have less impact on the ethnic balance. The effect of kin refugees on conflict is positive with significance at the 1% error level. Further, the interaction of this restricted co-ethnic refugee variable with political exclusion in Model 6 supports the claim that only excluded groups with kin refugees have a higher risk of rebellion, as opposed to governmental groups that are not affected by kin refugees. In Model 7, the refugee variable only includes ethnic kin refugees from contiguous countries, as opposed to kin refugees with a maximal distance of 950 kilometers between home and host state that were considered in the previous models, because notably refugees from neighboring states are associated with a higher conflict risk (Salehyan & Gleditsch, 2006). Also, this variable is positive and significant, as is the conditional effect of kin refugees from neighbors and political exclusion in Model 8.

The next sensitivity check in Models 9 and 10 includes a nominal variable indicating the world region, which are South-Eastern Europe and the former Soviet Union, sub-Saharan Africa, the Arab world, Asia, Latin America, and the West as baseline category to control for unit-specific heterogeneity of the error terms. Particularly the refugee crisis in the Great Lakes region in Central Africa in the 1990s awakened scientific interest in refugee involvement in conflict. Thus, it is crucial to control for whether the results claiming global application are biased by regional factors. The impact of the co-ethnic refugee variable remains positive and significant when controlling for the world region, however, with a 10% error margin. Hence, the phenomenon of increased instability after a co-ethnic refugee influx is not restricted to Africa. Also, the average difference in the effect of kin refugees on conflict between excluded and included groups remains positive and significant for excluded groups.

To account for a potential conflict diffusion effect, Model 11 controls for whether the group has any transborder kin and whether this TEK group was involved in civil conflict in the preceding year. Yet, conflict-affected transnational kin groups have no significant impact on the likelihood of intrastate conflict, and the effect of co-ethnic refugees remains positive and significant. While conflicts can spread along ethnic linkages (see Gleditsch, 2007), accounting for co-ethnic refugees shows that conflicts do not spread haphazardly along transborder ethnic ties, but rather in combination with co-ethnic refugees or other cross-border movements, such as transnational rebels (see Salehyan, 2007). Similarly, in Model 12, the effect of co-ethnic refugees on armed conflict remains significantly conditional on political marginalization. Models 13 and 14 include the relative size and squared relative size of the TEK group compared to the incumbent, to control for the curvilinear effect of foreign kin groups on civil conflict as theorized in Cederman et al. (2013). Considering TEK groups, who also are the potential senders of refugees, attenuates the effect of co-ethnic refugees, though the coefficient remains positive and significant at the 10% error level.¹⁰

Finally, in Models 15 and 16, I control for whether the refugee-sending and receiving countries are rivals, as coded in the strategic rivalry dataset (Thompson & Dreyer, 2011). Scholarship contends that rivalry between states increases the size of refugee flows between these states (Moorthy & Brathwaite, 2016), as well as the risk of refugee militarization (Lebson, 2013; Muggah & Mogire, 2006; Stedman & Tanner, 2003), which might indirectly affect refugee-related conflict. However, the effect of kin refugees remains positive and significant.

Conclusion

Focusing on the ethnicity of refugees and power politics in the asylum state, I have investigated to what extent refugees trigger ethnic conflict. My point of departure has been past research indicating that refugees from neighboring states increase the risk of armed conflict in asylum countries owing to an upset ethnic balance and case-based evidence that the host regime plays a decisive role in preventing refugee-related violence. However, previous research has not systematically analyzed the ethnic group membership of refugees and, therefore, has failed to properly test the ethnic balance mechanism. I highlight that refugees per se do not cause intrastate conflict. Yet, if refugees arrive in already tense situations with political inequality, they may indirectly contribute to decreasing stability, particularly if they are co-ethnics of a marginalized local group. Although these findings suggest that refugees increase political instability in some situations, they differ from studies and public opinions that depict forced migrant inflows as inherently harmful. Here, I have presented a disaggregated and more balanced framework that shows that internal problems caused by the marginalization of ethnic groups may intensify in the context of refugee migration.

This study presents the first systematic test of the argument that refugees trigger ethnic conflict by upsetting the cultural balance in a state. I find that refugees may amplify intergroup tensions that already existed before the refugee crisis. Disaggregated data on the ethnicity of refugees allows me to test whether refugees have connections to the local population. This contributes to the emerging literature that examines the effect of refugees on armed conflict at a substate level (see e.g. Bohnet, Cottier & Hug, 2018; Fisk, 2014). Also, highlighting that refugee movements are composed of different ethnicities and identity groups copes with the fact that refugees are not homogenous masses (see Lischer, 2007; Stein, 1981). Moreover, scholarship argues that ethnicity is an important category in war and refugee studies; thus it is a logical and important consequence to combine these fields. The analysis at the ethnic group level contributes to the understanding of the refugees’ role in ethnic civil conflict processes. Distinguishing between co-ethnic and ethnic alien refugee groups, this study could not replicate Salehyan & Gleditsch’s (2006) often-cited finding that refugees from neighboring states essentially increase the risk of conflict. Finally, by assessing the effect of refugees in different political situations in the asylum state, disaggregating between ties to governmental and excluded ethnic groups, this study introduces the host state as the determinant factor for civil conflict outbreak after a refugee influx. Refugee movements consist of diverse groups and individuals and consequently, the asylum state perceives them in various ways, resulting in different policies for distinct refugee populations.

With regards to the ongoing political debate on where the increasing number of refugees can be sheltered, these are important empirical findings. Refugees may exacerbate the risk of violent conflict when they arrive in already tense settings. Consequently, I suggest that governments in refugee-receiving countries can reduce the risk of resentments and consequent insurgencies among its people by providing relief and security to the refugees and the local population and by including political minorities in the state decisionmaking. Having shown that situations where refugees are co-ethnics of politically marginalized groups are more likely to experience violent conflict, because the host government refuses to assist these refugees, I derive that particularly refugees without protection suffer from vulnerability and rebel manipulation. In cases where refugees miss support from political leaders in the asylum state, the international community and humanitarian organizations are called upon to maintain assistance and safety. Also, international stakeholders are prompted to urge asylum states to revise their policies and impartially assist all refugees regardless of their national, religious or ethnic background.

While I believe I have disentangled some important aspects of refugee movements as ethnic civil conflict triggers in this study, there still is room for further research to solve the challenges associated with the record high of forced displacement in today’s world. First, further systematic analysis is critical to understanding how the other negative mechanisms attributed to refugee groups actually relate to violent conflict, such as rebel network diffusion or the refugees’ impact on the local economy. Second, research on specific refugee identities should be advanced to deepen the knowledge on refugee manipulation in conflict, including the settlement location beyond Africa, gender or age (see e.g. Johnson, 2011). Such analysis will require improved quantitative refugee data that is based on more precise figures and categorizations, as well as longer time frames. Third, while researchers have started to focus on lower-level violence involving refugees as victims or as perpetrators, including riots, communal strife, and terrorist attacks, there still is need for a better understanding of how the instability caused during conflicts in neighboring countries and the consequent refugee immigration manifest in different forms of political conflict.

For now, I conclude that previous research and especially the public debate have overestimated the effect of refugees on civil conflict in asylum countries. Immigration-hostile governments have blamed refugees for internal problems without solid cause: refugees that flee along transborder ethnic ties can affect the cultural setup in a country, but only if their co-ethnics experience political exclusion in the first place.

Footnotes

Replication data

The dataset and do-file for the empirical analysis in this article, along with the Online appendix, can be found at .

Acknowledgements

I thank Lars-Erik Cederman, Simon Hug, Idean Salehyan, Heidrun Bohnet, Nils-Christian Bormann, Philipp Hunziker, Sebastian Schutte, Guy Schvitz, Kunaal Sharma, Manuel Vogt, Julian Wucherpfennig, Alex Braithwaite, Burcu Savun, and the anonymous reviewers for helpful comments. Also, I thank Adriana Rüegger for coding assistance.

Funding

This research was funded by the Swiss Network for International Studies.

Notes

References

Adamson

Fiona B

(2006) Crossing borders: International migration and national security. International Security 31(1): 165–199.

Adelman

Howard

(1998) Why refugee warriors are threats. Journal of Conflict Studies 18(1): 49–69.

Barber

Ben

(1997) Feeding refugees, or war. Foreign Affairs 76(4): 8–14.

BBC News (1999) World: Europe: UN plans refugee evacuation (http://news.bbc.co.uk/2/hi/europe/334772.stm).

Böhmelt

Tobias

Bove

Vincenzo

Gleditsch

Kristian Skrede

(2019) Blame the victim? Refugees, state capacity, and non-state actor violence. Journal of Peace Research 56(1): 73–87.

Bohnet

Heidrun

Cottier

Fabien

Hug

Simon

(2018) Conflict-induced IDPs and the spread of conflict. Journal of Conflict Resolution 62(4): 691–716.

Bove

Vincenzo

Böhmelt

Tobias

(2016) Does immigration induce terrorism? Journal of Politics 78(2): 572–588.

Buhaug

Halvard

Gleditsch

Kristian Skrede

(2008) Contagion or confusion? Why conflicts cluster in space. International Studies Quarterly 52(2): 215–233.

Carter

David B

Signorino

Curtis S

(2010) Back to the future: Modeling time dependence in binary data. Political Analysis 18(3): 271–292.

10.

Cederman

Lars-Erik

Girardin

Luc

Gleditsch

Kristian Skrede

(2009) Ethnonationalist triads: Assessing the influence of kin groups on civil wars. World Politics 61(3): 403–437.

11.

Cederman

Lars-Erik

Gleditsch

Kristian Skrede

Salehyan

Idean

Wucherpfennig

Julian

(2013) Transborder ethnic kin and civil war. International Organizations 67(2): 389–410.

12.

Cederman

Lars-Erik

Wimmer

Andreas

Min

Brian

(2010) Why do ethnic groups rebel? New data and analysis. World Politics 62(1): 87–119.

13.

Choi

Seung-Whan

Salehyan

Idean

(2013) No good deed goes unpunished: Refugees, humanitarian aid, and terrorism. Conflict Management and Peace Science 30(1): 53–75.

14.

Choi

Seung-Whan

Piazza

James A

(2016) Internally displaced populations and suicide terrorism. Journal of Conflict Resolution 60(6): 1008–1040.

15.

Crisp

Jeff

(2000) Forms and sources of violence in Kenya’s refugee camps. Refugee Survey Quarterly 19(1): 54–70.

16.

Davis

David R

Moore

Will H

(1997) Ethnicity matters: Transnational ethnic alliances and foreign policy behavior. International Studies Quarterly 41(1): 171–184.

17.

Dowty

Alan

Loescher

Gil

(1996) Refugee flows as grounds for international action. International Security 21(1): 43–71.

18.

Feenstra

Robert C

Inklaar

Robert

Timmer

Marcel P

(2015) The next generation of the Penn World Table. American Economic Review 105(10): 3150–3182.

19.

Fisk

Kerstin

(2014) Refugee geography and the diffusion of armed conflict in Africa. Civil Wars 16(3): 255–275.

20.

Fisk

Kerstin

(2018) One-sided violence in refugee-hosting areas. Journal of Conflict Resolution 62(3): 529–556.

21.

Fisk

Kerstin

(2019) Camp settlement and communal conflict in sub-Saharan Africa. Journal of Peace Research 56(1): 58–72.

22.

Forsberg

Erika

(2014) Transnational transmitters: Ethnic kinship ties and conflict contagion 1946–2009. International Interactions 40(2): 143–165.

23.

Gerdes

Felix

(2006) Forced migration and armed conflict: An analytical framework and a case study of refugee-warriors in Guinea. Working paper, University of Hamburg (https://www.wiso.uni-hamburg.de/fachbereich-sowi/professuren/jakobeit/forschung/akuf/archiv/arbeitspapiere/migration-gerdes-2006.pdf).

24.

Gleditsch

Kristian Skrede

(2007) Transnational dimensions of civil war. Journal of Peace Research 44(3): 54–66.

25.

Gleditsch

Nils Petter

Wallensteen

Peter

Eriksson

Mikael

Sollenberg

Margareta

Håvard

Strand

(2002) Armed conflict 1946–2001: A new dataset. Journal of Peace Research 39(5): 615–637.

26.

Goldstone

Jack A

(2002) Population and security: How demographic change can lead to violent conflict. Journal of International Affairs 56(1): 3–21.

27.

Guardian (2016) Kenya says it will shut world’s biggest refugee camp at Dadaab. 11 May (https://www.theguardian.com/world/2016/may/11/kenya-close-worlds-biggest-refugee-camp-dadaab).

28.

International Alert (2015) Citizen’s perceptions of security threats stemming from the Syrian refugee presence in Lebanon. Beirut Research and Innovation Center, Lebanese Center for Studies and Research (http://www.international-alert.org/publications/citizens-perceptions-security-threats-stemming-syrian-refugee-presence).

29.

International Crisis Group (2016) Turkey’s refugee crisis: The politics of permanence. Report no. 241 (https://www.crisisgroup.org/europe-central-asia/western-europemediterranean/turkey/turkey-s-refugee-crisis-politics-permanence).

30.

Itani

Faysal

Grebowski

Sarah

(2013) Beyond spillover: Syria’s role in Lebanon’s drift toward political violence. Atlantic Council Issue Brief (http://www.atlanticcouncil.org/publications/issue-briefs/beyond-spillover-syrias-role-in-lebanons-drift-toward-political-violence).

31.

Jacobsen

Karen

(1996) Factors influencing the policy responses of host governments to mass refugee influxes. International Migration Review 30(39): 655–678.

32.

Jamal

Umair

(2016) Pakistan’s Balochs fear minority status in their own province. Diplomat (https://thediplomat.com/2016/02/pakistans-balochs-fear-minority-status-in-their-own-province/).

33.

Johnson

Richard AI

(2011) Refugee camp security: Decreasing vulnerability through demographic controls. Journal of Refugee Studies 24(1): 23–46.

34.

Kagwanja

Peter

Juma

Monica

(2008) Somali refugees: Protracted exile and shifting security frontiers. In: Newman

Edward

Loescher

Gil

Milner

James

Troeller

Gary

(eds) Protracted Refugee Situations: Political, Human Rights and Security Implications. Tokyo, New York: United Nations University Press, 214–247.

35.

Khalidi

Omar

(1998) From torrent to trickle: Indian Muslim migration to Pakistan, 1947–97. Islamic Studies 37(3): 339–352.

36.

Krcmaric

Daniel

(2014) Refugee flows, ethnic power relations, and the spread of conflict. Security Studies 23(1): 182–216.

37.

Lebson

Mike

(2013) Why refugees rebel: Towards a comprehensive theory of refugee militarization. International Migration 51(5): 1–16.

38.

Lischer

Sarah Kenyon

(2005) Dangerous Sanctuaries: Refugee Camps, Civil War, and the Dilemmas of Humanitarian Aid. Ithaca, NY: Cornell University Press.

39.

Lischer

Sarah Kenyon

(2007) Causes and consequences of conflict-induced displacement. Civil Wars 9(2): 142–155.

40.

Loescher

Gil

Milner

James

(2005) The long road home: Protracted refugee situations in Africa. Survival: Global Politics and Strategy 42(2): 153–174.

41.

Martin

Adrian

(2005) Environmental conflict between refugee and host communities. Journal of Peace Research 42(3): 329–346.

42.

McCommon

Carolyn S

(1989) Refugees in Belize: A cauldron of ethnic tensions. In: Gonzalez

Nancie L

McCommon

Carolyn S

(eds) Conflict, Migration, and the Expression of Ethnicity. Boulder, CO: Westview, 91–102.

43.

Milner

James

(2000) Sharing the security burden: Towards the convergence of refugee protection and state security. Refugee Studies Centre working paper (4).

44.

Milton

Daniel

Spencer

Megan

Findley

Michael

(2013) Radicalism of the hopeless: Refugee flows and transnational terrorism. International Interactions 39(5): 621–645.

45.

Mogire

Edward

(2004) A Preliminary Exploration of the Linkages between Refugees and Small Arms. Bonn: BICC.

46.

Moore

Will H

Shellman

Stephen M

(2007) Whither will they go? A global study of refugees’ destinations, 1965–1995. International Studies Quarterly 51(4): 811–834.

47.

Moorthy

Shweta

Brathwaite

Robert

(2016) Refugees and rivals: The international dynamics of refugee flows. Conflict Management and Peace Science. https://doi.org/10.1177/0738894216657047.

48.

Muggah

Robert

Mogire

Edward

(2006) Arms availability and refugee militarization in Africa: Conceptualizing issues. In: Muggah

Robert

(ed.) No Refuge: The Crisis of Refugee Militarization in Africa. London: Zed, 1–48.

49.

Newland

Kathrin

(1993) Ethnic conflict and refugees. Survival 35(1): 85–101.

50.

Petersen

Roger D

(2002) Understanding Ethnic Violence: Fear, Hatred, and Resentment in Twentieth-Century Eastern Europe. New York: Cambridge University Press.

51.

Refugee Council (1999) Sri Lankan refugees in the Indian state of Tamil Nadu (repository.forcedmigration.org/pdf/?pid=fmo:1940).

52.

Rüegger

Seraina

Bohnet

Heidrun

(2018) The Ethnicity of Refugees (ER): A new dataset for understanding flight patterns. Conflict Management and Peace Science 35(1): 65–88.

53.

Saideman

Stephen M

(2002) Discrimination in international relations: Analyzing external support for ethnic groups. Journal of Peace Research 39(1): 27–50.

54.

Salehyan

Idean

(2007) Transnational rebels, neighboring states as sanctuary for rebel groups. World Politics 59(1): 217–242.

55.

Salehyan

Idean

Gleditsch

Kristian Skrede

(2006) Refugees and the spread of civil war. International Organization 60(2): 335–366.

56.

Sambanis

Nicholas

(2001) Do ethnic and nonethnic civil wars have the same causes? Journal of Conflict Resolution 45(3): 259–282.

57.

Savun

Burcu

Gineste

Christian

(2019) From protection to persecution: Threat environment and refugee scapegoating. Journal of Peace Research 56(1): 88–102.

58.

Schmeidl

Susanne

(1997) Exploring the causes of forced migration: A pooled time series analysis, 1971–990. Social Science Quarterly 78(2): 284–308.

59.

Stedman

Stephen John

Tanner

Fred

(2003) Refugee Manipulation: War, Politics, and the Abuse of Human Suffering. Washington, DC: Brookings Institution Press.

60.

Stein

Barry N

(1979) The Geneva conferences and the Indochinese refugee crisis. International Migration Review 13(4): 716–723.

61.

Stein

Barry N

(1981) The refugee experience: Defining the parameters of a field of study. International Migration Review 15(1): 320–330.

62.

Steputat

Finn

(2004) Refugees, security and development: Current experience and strategies of protection and assistance in the ‘region of origin’. Danish Institute for International Studies, working paper no. 11 (https://www.diis.dk/files/media/publications/import/extra/fst_refugees.pdf).

63.

Teitelbaum

Michael S

(1984) Immigration, refugees, and foreign policy. International Organization 38(3): 429–450.

64.

Terry

Fiona

(2002) Condemned to Repeat? The Paradox of Humanitarian Action. Ithaca, NY: Cornell University Press.

65.

Thompson

William R

Dreyer

David R

(2011) Handbook of International Rivalries, 1494–2010. Washington, DC: Congressional Quarterly Press.

66.

UNHCR (2007) Convention and protocol relating to the status of refugees (http://www.unhcr.org/protect/PROTECTION/3b66c2aa10.pdf).

67.

UNHCR (2015) Global Trends 2015 (http://www.unhcr.org/statistics/unhcrstats/576408cd7/unhcr-global-trends-2015.html).

68.

US Committee for Refugees and Immigrants (2007) World Refugee Survey 2006 (http://reliefweb.int/report/world/world-refugee-survey-2006).

69.

Vogt

Manuel

Bormann

Nils-Christian

Rüegger

Seraina

Cederman

Lars-Erik

Hunziker

Philipp

Girardin

Luc

(2015) Integrating data on ethnicity, geography, and conflict: The Ethnic Power Relations dataset family. Journal of Conflict Resolution 59(7): 1327–1342.

70.

Weiner

Myron

(1992) Security, stability, and international migration. International Security 17(3): 91–126.

71.

Whitaker

Beth Elise

(2003) Refugees and the spread of conflict: Contrasting cases in Central Africa. Journal of Asian and African Studies 38(2–3): 211–231.

72.

Whitaker

Reg

(1998) Refugees: The security dimension. Citizenship Studies 2(3): 413–434.

73.

Wucherpfennig

Julian

Metternich

Nils W

Cederman

Lars-Erik

Gleditsch

Kristian Skrede

(2012) Ethnicity, the state, and the duration of civil wars. World Politics 64(1): 79–115.

74.

Zolberg

Aristide R

Suhrke

Astri

Aguayo

Sergio

(1989) Escape from Violence, Conflict and the Refugee Crisis in the Developing World. Oxford, New York: Oxford University Press.