Dangerous neighborhoods: State behavior and the spread of ethnic conflict

Abstract

This article argues that the spread of ethnic conflict across international borders is a result of purposive repressive state action combined with political opportunity which allows the ethnic group to mobilize in response. We test our argument using a bivariate probit model to simultaneously estimate the risk of repression escalation and the spread of ethnic conflict (1976–2009), and find that the likelihood of both increases in the presence of relatively large, territorially concentrated transnational ethnic groups, and that nationalist governments are more likely to escalate repression amidst nearby ethnic violence. Political crises also exacerbate the risk of ethnic conflict spread.

Keywords

Civil war contagion ethnic conflict state repression transnational ethnic groups

The potential for intrastate conflict to spread from one state to its neighbors is a concern that looms large in the minds of scholars and policymakers.¹ Scholars point to evidence that a civil war wreaking havoc in one country can undermine stability and peace in neighboring states (Gleditsch, 2007; Hegre and Sambanis, 2006), and cause regional economic stagnation and decline (Collier et al., 2003). Buhaug and Gleditsch (2008) demonstrate that the pernicious effect of having a neighbor at war remains substantial even when controlling for other regionally clustered factors known to increase the chances of conflict, such as regime type and GDP per capita. Ethnicity plays a crucial role in conditioning the spread of civil wars (Buhaug and Gleditsch, 2008; Lake and Rothchild, 1998; Saideman and Ayers, 2000; Sambanis, 2001).² In fact, a recent study by Bosker and de Ree (2014: 207) finds that “Only ethnic wars tend to spill over, and they are more likely to spread along ethnic lines.” Fear of conflicts spreading via cross-border ethnic groups has sparked international interventions in places as disparate as Kosovo and Libya.³ Most recently, conflict involving ethnic Kurds in Syria has spread to neighboring Turkey, stymieing a historic peace process.

In this study, we ask: why do some ethnic conflicts seem to spread across international boundaries while others do not?⁴ Specifically, we are interested in the process by which ethnic conflict spreads from one state to its neighbours. It is not uncommon in research on the causes and dynamics of ethnic conflict contagion to borrow terminology from epidemiological studies, and even allude to shared theoretical underpinnings (for an overview of the “conflict–disease analogy,” see Braithwaite, 2010: 313). To understand contagion—defined as the process whereby armed conflict in one state alters the probability of new armed conflict onset in another state at a later point in time (Forsberg, 2014b: 144)—researchers identify a specific patient as the starting point (state A in a conflict), delineate the risk set, which comprises individuals most likely to be infected (proximate states B), and then examine the sources of variation in individual propensity for infection.⁵ The likelihood of “catching” conflict is said to be largely determined by factors such as the degree of exposure to the patient and to others at risk, the strength of one’s immune system, family history, vaccinations, etc. (contiguous borders or proximity to the conflict zone, state capacity, regime type, ethnic kinship ties, etc.).

Two lines of argument about the process of conflict contagion follow from the conflict–disease analogy. The first suggests that weak states are more likely to see conflict spread because they cannot effectively counter nascent insurgencies or credibly commit to meeting opposition demands, or are simply unable to block the negative externalities emanating from nearby war zones. The second posits that if an ethnic group spans the border between states A and B, once a conflict between the group and state A erupts, state B becomes more likely to experience a conflict with that group as well. Put somewhat differently, ethnic kin groups are deemed to be “conflict transmitters” (Brown, 1996: 595) that may serve as “conduits of contagion processes” (Forsberg, 2014b).

While seemingly intuitive, the main flaw of the conflict–disease analogy is that states and ethnic groups are not passive receptors and vectors in the process of conflict contagion. We argue that armed conflict spreads not simply because of an inherent susceptibility to conflict contagion brought on by either state fragility or the presence of a transnational ethnic group. Rather, it may be the result of purposive repressive actions taken by the state combined with political opportunity that allows the transnational ethnic group to mobilize in response. Of course, in some situations it may be the ethnic group that instigates new conflict. The state, however, is also an agent in this process and as such, merits specific attention in the study of conflict contagion.

We propose that ongoing ethnic conflict in state A may shape the threat perception of political leaders in state B with respect to their own ethnic populations. In other words, governments draw information regarding the potential threat a domestic ethnic group poses from the behavior of their kin nearby. Two factors combine to cause leaders in state B to be especially apprehensive about the possibility of an uprising inspired by co-ethnics in conflict, and in turn, to overreact by instituting repressive policies: regime ideology—specifically, nationalism—and the relative size of the domestic ethnic group.

When it occurs, increased repression urges members of the transnational ethnic group to reassess their own security situation. Under opportune political circumstances that reassessment can trigger or embolden ongoing dissent that may escalate toward violent rebellion. Instances of intra-regime instability, political transitions and other times where the opening of the political space allows the ethnic group to mobilize and challenge the state, increase the likelihood that a new conflict will erupt. At the same time, when such opportunity is absent, repression can appear to have had the desired effect of averting a potential armed uprising. Such was the case when Syria repressed its Kurdish population during the early 1960s, seemingly forestalling the spread of conflict from Iraq, Iran, and Turkey at that time (Brathwaite, 2014). Whether a state is able to thwart a potential rebellion with force and head off the spillover of nearby conflict into its own territory is therefore inextricably linked to political opportunity and timing.

This study contributes to a growing body of research that draws together insights from the literature on state repression and the literature on civil conflict to better understand the process of conflict spread. Most existing studies of state repression focus on domestic sources of state threat perception, such as the nature of the political regime (Davenport, 1995, 1999, 2000) or violent domestic opposition movements (Regan and Henderson, 2002; Valentino et al., 2004), often neglecting to consider international influences on threat perception. The civil conflict literature, in turn, has predominantly focused on how nearby conflict affects transnational ethnic groups, generally overlooking that conflict can also influence state behavior. Drawing the two together, Forsberg (2014b) suggests that a cross-border ethnic group in conflict in one state may lead neighboring states to be suspicious of members of that ethnic group in their own territory. Yet this proposition is not tested empirically in her study. In a similar vein, Cederman et al. (2009a, b, 2013) argue that states may consider the demographic size of a domestic ethnic group’s cross-border kin as indicative of the group’s potential strength or the possibility of external intervention, and propose that such calculations influence the risk of armed conflict onset. Their research, however, does not empirically consider the role of state repression in the process.

Paying closer attention to the interplay between state repression and civil conflict, Danneman and Ritter (2014) find evidence for conflict in one state leading to preemptive repression in nearby states, while Hultquist (2015), who directly models the endogeneity between repression and conflict, finds that greater levels of repression lead to greater levels of conflict, and concludes that widespread repression is counterproductive in fighting an insurgency. These studies, however, do not explicitly consider the impact of ethnic kinship ties on preemptive repression or the spread of armed conflict. In a timely examination of violent conflict involving the Kurds in Iraq, Iran, Turkey, and Syria, Brathwaite (2014) argues that an understanding of conflict spread requires an in-depth analysis of the dynamic interaction between state perceptions of security threats, repression, and ethnic group behavior.

Our central argument is that when state B has a neighbor in conflict (e.g. state A), the interaction between two processes—political repression ensuing from state B’s re-estimation of security threats (namely, an overreaction to the potential of a violent uprising by its own ethnic group), and an opening in the political space allowing the ethnic group to react to increasingly repressive conditions—can trigger new conflict onset. While preemptive repression escalation by the state is not the only pathway by which conflict can spread, it is one possible pathway that has so far gone largely untested.

This study thus furthers our understanding of the process whereby ethnic conflict spreads in three ways. First, by demonstrating that states base their threat perception on international as well as domestic factors, this study contributes to our knowledge about how and when states decide to repress domestic ethnic groups, furthering our understanding of repression in general. Concerned by the possibility that a nearby conflict can spread across their borders via transnational ethnic kinship ties, states overreact to a potential threat by resorting to increasingly repressive measures; under certain political conditions, this overreaction can culminate in a new armed conflict. This clarifies one important process by which civil conflict can spread.

Second, rejecting the conflict–disease analogy, this study argues that states are not passive receptacles of contagion, but are precipitators of, and active participants in the conflict contagion process. By focusing on state behavior, this study directly tackles the role of state agency in driving or blocking conflict spread—a key aspect of the contagion process that, as both Black (2012) and Forsberg (2014a) point out, is consistently overlooked.

Third, instead of the catchall category of state capacity as a proxy for susceptibility to contagion, we focus on “political opportunity”—a set of institutional indicators that reflect a dynamic and highly contingent political environment—that allows the ethnic group to respond to increasing repression. Political opportunity therefore may be the condition that helps to determine whether state repression is “successful” in preventing conflict spread or contributes to spreading the conflict.

Utilizing a set of spatial and temporal restrictions, we identify states that are at risk of conflict contagion following the onset of intrastate ethnic conflicts in a nearby state between 1976 and 2009. In other words, having a neighbor in conflict is a scope condition for our analysis (Forsberg, 2014b: 144). We test our argument using a seemingly unrelated bivariate probit model that simultaneously predicts repression escalation and ethnic conflict onset, taking into account the correlation between the residuals from the equations (Greene, 2003). This statistical technique explicitly accounts for the potential endogeneity between deteriorating human rights conditions and the outbreak of civil conflict, as well as for the possibility that there are omitted explanatory factors which affect both the likelihood of repression escalation and ethnic conflict onset.⁶

Having identified the set of countries—states B—which are at risk of experiencing conflict spreading from neighboring state A, we examine which factors best account for an increase in their respective levels of political repression. Keeping the scope condition of conflict in state A in mind, we pay close attention to regime ideology and group-specific attributes that shape state threat perception and inform state B’s overreaction in the form of repression escalation. We then consider the role of increasing repression and political opportunity in the form of intra-regime instability and political transitions to determine their respective influence on the likelihood of new ethnic conflict onset. Based on the results of this analysis, we argue that ethnic conflict can spread to new states as a result of overreaction by the political authorities of state B and political opportunity, allowing the ethnic group to mobilize in rebellion.

The remainder of the article proceeds as follows. We begin with an overview of extant explanations for the spread of civil conflicts. The following section lays out our argument which integrates international factors, state and ethnic group behavior, and domestic political and ideological variables into a unified theoretical framework. Next, we discuss the data and the statistical technique, and then conduct several quantitative tests of the proposed hypotheses. After presenting the findings, we conclude with a discussion of the study’s implications for scholarship and policy.

Dangerous neighborhoods

Why do some ethnic conflicts seem to spread across international boundaries while others do not? Two sets of explanations have been suggested. The first set of arguments considers the influence of having ethnic kin in conflict in state A on the members of that group in state B. Focusing on changes to the ethnic balance of power, several authors have argued that the risk of conflict in nearby states increases as the ethnic group grows stronger in relation to the state, either as the result of easily attainable weapons (Gleditsch, 2007), or owing to refugee flows inducing a shift in the demographic balance between competing ethnic groups (Fisk, 2014; Lake and Rothchild, 1998; Salehyan and Gleditsch, 2006; Salehyan, 2007). Cross-border ethnic kinship ties can also increase the likelihood of conflict spread because the group in conflict may provide a safe haven for its kin as they plan, mobilize, and undertake a rebellion against their own government (Salehyan, 2007). Moreover, ethnic groups in conflict may supply trained and experienced fighters, directly aiding their kin in fighting the new war (Krcmaric, 2014). Receiving states sometimes view refugees as a potential tool against rivals, arming or supplying them and pointing them toward state enemies (Lischer, 2005). Alternatively, cross-border groups may share the goal of obtaining their own united ethnic state. Conflict in one state may convince nearby kin to join the fight for this common goal (Saideman and Ayers, 2000).

Shifting focus from collaboration between cross-border co-ethnics to demonstration effects, other scholars have suggested that when an ethnic group experiences violence at the hands of its government, ethnic kin nearby may be compelled to reevaluate their own security situation—fearing that “if it can happen there, it could happen here” (Kuran, 1998). Similarly, if the ethnic group succeeds in gaining the desired policy concessions by violent means, or even defeats the government and takes power, their kin nearby may be inspired to take up arms in pursuit of similar goals—or as Fearon (1998: 112) puts it, people may start believing that “if it can happen there, why couldn’t it happen here?”

The second set of explanations for conflict spread focuses on the state—especially, state capacity—rather than the ethnic group. Here, the argument is that weak states are more likely to see conflict spread because they are unable to accommodate opposition demands (Sobek, 2010) or simply incapable of insulating themselves from the negative externalities coming over the border from the nearby conflict-affected state (Braithwaite, 2010).

Both sets of explanations focus predominantly on a single actor in the conflict, either the ethnic group or the state. We suggest that this leaves half of the story unexplained. The outbreak of armed conflict is necessarily the product of an interaction between the government in power and the opposition, as well as the political conditions under which both operate.

Theory

We argue that when conflict involving a transnational ethnic group erupts in state A, leaders in state B take that conflict as an indicator of the potential threat posed by kin members of that group in their own territory. This threat perception is magnified when the state is led by a nationalist government and when the ethnic group is relatively large. State B overreacts to the potential threat, instigating or increasing repression. Members of the ethnic group also reassess their own security situation, based in part on this repression. When the regime is stable and the ethnic group is unable to mobilize to challenge it, repression may seem to have successfully prevented the conflict’s spread. However, in moments of regime instability or political transition, the opening of political space presents an opportunity for the ethnic group to effectively challenge the state. In these circumstances, increasingly repressive state policies may lead the ethnic group to rebel and a new conflict to begin. The following two sections lay out this logic in more detail.

Re-estimation of security threats, overreaction, and repression escalation

State repression is traditionally defined as a form of coercive sociopolitical control utilized by political authorities against individuals or groups within their territorial jurisdiction, for the purpose of imposing a cost on the target as well as deterring specific activities and/or beliefs perceived to be challenging to government personnel, practices or institutions (Goldstein, 1978). Forms of state repression vary extensively, from restricting civil liberties, limiting the freedom of movement, organization, and access to resources necessary for popular collective action, to political arrests and state terror tactics such as torture and extrajudicial killings (Davenport, 2007a). Research on the causes of state repression shows that the application of coercive power is influenced by the nature of the domestic threat, the character of the military, ideology, and regime type (Davenport, 2000). However, it is also possible that international factors shape threat perception and motivate repressive action.

One way in which nearby unrest involving a transnational ethnic group in state A may factor into the thinking of political leaders in state B is by raising concerns about a copy-cat rebellion. For example, Baltic states like Latvia are becoming increasingly alarmed that Ukraine’s conflict with its Russian minority may lead to volatile interethnic tensions and possible conflict with their own Russian-speaking citizens (Los Angeles Times, 2015). The government may also worry that close contacts between the domestic ethnic group and its kin in conflict will lead to radicalization and provide material support for rebellion, a fear Turkey has repeatedly expressed in regards to its Kurdish minority. In either case, the government draws information about the threat posed by the domestic ethnic group from the behavior of their kin in nearby states. Once the threat is perceived, the government may overreact by repressing the domestic ethnic group to prevent a potential ethnic uprising within their own borders. In contrast, concerns about potential challenges to the state inspired by nearby hostilities should be less severe in the absence of transnational ethnic kinship ties. The first hypothesis we consider is:

H1: Given ethnic conflict in state A, the probability of repression escalation in state B increases in the presence of transnational ethnic kinship ties.

How can we know, however, that the government of state B decides to increase repression based on the behavior of the transnational ethnic group’s kin in state A and not in direct response to violent domestic opposition activity? And what precisely does an “overreaction” entail? Assessment of threat normally involves some ambiguity, yet it is clear that state authorities do not perceive all forms of threats with equal urgency. Specifically, state leaders carefully assess the magnitude of the threat based on the frequency and severity of dissident activities, the complexity and diversity of protest tactics, and the gap between the challengers’ behavior and culturally accepted levels of dissent (Davenport, 1995).⁷

Mass, organized, and especially violent opposition threats are particularly likely to be met with high levels of repression (Carey, 2010; Regan and Henderson, 2002). Considering that states do not tend to respond to different forms of actual and perceived threats with equal force, the use of increasingly more violent repression in the face of predominantly non-violent protest would qualify as an “overreaction.” Thus, if we find that leaders who observe conflict in a nearby state resort to severe and systematic rights violations to counter nonviolent protest at home, we can have greater confidence in our claim that international factors shape threat perception and precipitate repressive action.

H2: Given ethnic conflict in state A, the probability of repression escalation in state B increases in the face of largely nonviolent political dissent.

If political authorities in state B worry that ethnic conflict in state A can prompt domestic ethnic groups to mobilize against the regime, they will be more likely to overreact to this potential threat by increasing repression when a nationalist government is in power. In multiethnic states, portraying competing ethnic groups in threatening terms can bolster the incumbent’s popularity and help him consolidate political power (Brown and Farhad, 1995; Roessler, 2011). This is in part because the inflation of threat from other groups serves to sharpen the in-group/out-group distinction and to enhance in-group solidarity and cohesion. Especially when the nationalist government controls the media, as Brown (1996: 20) puts it, “the relentless drumbeat of ethnic propaganda can distort political discourse quickly and dramatically,” and precipitate conflict by “dividing and radicalizing groups along ethnic fault lines.” Indeed, both historical and experimental evidence suggests that the general public tends to be most responsive to threat inflation efforts when leaders use rhetoric that emphasizes cultural and national differences (Kaufman, 2001; Rousseau and Garcia-Retamero, 2007). Examples of virulent forms of ethnic nationalism and politicians playing the “ethnic card” to justify the repression of ethnic minorities include but are certainly not limited to Bosnia, Burundi, Croatia, Kenya, Rwanda, and Serbia.

Our next hypothesis therefore concerns regime ideology:

H3: Given ethnic conflict in state A, the probability of repression escalation in state B increases if the government is led by a nationalist party

We also posit that leaders in state B will be especially apprehensive about the possibility of a violent rebellion stirred by transborder co-ethnics in conflict in state A, and therefore more likely to increase repression if the domestic ethnic minority group constitutes a relatively large portion of the national population. Studies of civil conflict suggest that group-specific factors such as relative population size, territorial settlement patterns, access to political power, and regional autonomy are related to the group’s willingness and capacity to mobilize for rebellion, and in turn, to conflict spread (Cederman et al., 2009a, 2013; Forsberg, 2014b). We do not take issue with these assertions. Instead, we focus on how such group-specific attributes combine with and are amplified by group behavior abroad to shape state threat perception, and in turn, state behavior.

Insofar as demographic factors influence “geopolitics, fiscal politics, ethnic and religious conflicts, and voting patterns,”Goldstone et al. (2012: 4) are correct to argue that political scientists studying conflict cannot afford to ignore demography. In multiethnic states, the imperative to act quickly to diffuse the potential threat of an uprising galvanized by ethnic kin involvement in nearby conflict is especially high where the group in question constitutes a sizable portion of state B’s population. The reason for this is straightforward: relatively large ethnic groups pose the most serious threat to states. Larger ethnic populations can raise larger political opposition movements and field larger rebel armies (Buhaug et al., 2008; Cederman et al., 2009a, 2013; Horowitz, 1985). State leaders are therefore acutely aware that rising discontent and opposition activity by relatively large ethnic minority populations should be contained and countered sooner rather than later.

Moreover, for the incumbent, relatively large ethnic minority groups represent not only a potentially formidable military threat but a political one as well (Konaev, 2016). Size connotes political legitimacy and the right to make ethnic demands in an organized fashion (Horowitz, 1985). Enhanced representation and participation in politics on the national stage translate into decision-making power that may conflict with the interests of the current leadership (Gurr, 2000). Thus, in multiethnic states, the ethnic group in power may perceive accommodation and co-optation of opposition groups representing the ethnic population whose kin are involved in conflict in a nearby state as too costly a strategy since such actions can effectively undermine the regime from within.

Recent research shows that leaders’ expectations of political survival, e.g. how secure they feel in their position of power, shape their behavior amid domestic challenges (Ritter, 2014; Roessler, 2011; Young, 2009). Here we argue that political leaders are aware that their hold on power may be weakened if not altogether compromised as a result of a series of reforms that increase the participation and representation of an excluded or underrepresented ethnic group. Cederman et al. (2013: 398) posit that “elites defending minority rule will be inclined to rely heavily on coercion to stifle challenges to their supremacy,” and will be “unlikely to compromise with challengers because their fear that concessions will threaten their long-term survival and the benefits of rule.” This uncompromising attitude, however, is not limited to cases of minority rule such as the apartheid government in South Africa, and has in fact played a role in fueling repression and ethnic violence in Sri-Lanka, Northern Ireland, Croatia, Iraq, and Turkey. In other words, when the excluded group is relatively large, state authorities are more likely to select coercive measures to block a potential uprising triggered by ethnic kin involvement in armed conflict nearby than accommodative measures that may endanger the survival of the current regime.

None of this is to say that repression is the government’s only means for keeping power amidst external and internal threats. Even authoritarian leaders tend to balance between repression and co-optation to maximize their support base, by supplying private goods, patronage, and policy concessions to the political and economic elites, the armed forces, and at times, to the citizens as well (Linz, 2000). However, given ongoing conflict in state A, and based on the probability of a successful campaign against the opposition, the cost of repression outweighing the benefits of accommodation, and the absence of preferable alternatives (Davenport 1995, 2007a; Shellman, 2006), the incentives to increase political repression in state B are greater in the presence of a relatively large transnational ethnic group.

H4: Given ethnic conflict in state A, the probability of repression escalation in state B increases if the transnational ethnic group constitutes a relatively large portion of the national population in state B.

Repression escalation, political opportunity structure, and new ethnic conflict onset

We argue that states with transnational ethnic kin in conflict in state A are more likely to experience not only increasing levels of repression, but conflict of their own. When the government of state B increases repression based in part on the behavior of the transnational ethnic group’s kin in state A, the group in state B must reassess its own security situation and respond. There are three possible outcomes in this situation. First, the group could decide that responding to repression with dissent is too costly. Second, the group could choose to respond to repression with dissent; the government, in turn, may further escalate its repressive campaign and suppress the nascent uprising. In both scenarios, the state’s repressive policy would then appear to have succeeded in preventing an ethnic rebellion. The third outcome entails the group responding to state repression with dissent, and the confrontation between state and opposition forces escalating to a full-scale armed conflict.

The fact that researchers find support for repression leading to dissent as well as repression preventing or ending dissent suggests that there are conditions that influence when repression leads groups to escalate dissent and when it does not (Davenport, 2007a: 622). Political opportunity within the state is one such important condition. The civil conflict literature tends to focus on state capacity (especially, extractive capacity or effective territorial control) as an indicator of political opportunity allowing for rebellion (Braithwaite, 2010; Fearon and Laitin, 2003; Hendrix, 2010). Yet measures such as GDP, primary commodity dependence, and mountainous terrain are often slow moving or even non-varying, and thus unable to accurately predict specific moments of increased state vulnerability.

We argue that political crises within the regime are a more precise indicator of change in the political opportunity structure. Government crises, defined as “a rapidly developing situation that threatens to bring the downfall of the current regime” (Banks, 2011) and regime transitions which reflect a normative and substantive change in political authority (Marshall et al., 2013), better capture the types of political opportunity that might lead repressed and disenfranchised groups to confront the government. Moments in which the regime faces intense crises or periods of political uncertainty and change are moments in which the state is weakened and may not be able to withstand a challenge from various opposition groups (Gleditsch and Ruggeri, 2010; Schock, 1996). The groups most likely to capitalize on such political opportunity are those that face increased repression from the state.

The last two hypotheses to test are then:

H5: Given ethnic conflict in state A, escalating state repression increases the probability of new conflict onset in state B.

H6: Given ethnic conflict in state A, political crises and transitions increase the probability of new conflict onset in state B.

Research design

Although our theory argues that repression escalation in state B is a significant predictor of the spread of civil conflict from state A to encompass state B, there is a mutual interdependence between the outbreak of armed conflict and increased repression in state B. Failing to model this endogenous relationship as well as to account for the possibility that there are explanatory factors which are difficult to isolate and measure that affect both the likelihood of repression escalation and ethnic conflict onset can bias our results. Therefore, we estimate a seemingly unrelated bivariate probit model, which is a two-equation econometric model that allows for the simultaneous estimation of the determinants of repression escalation and onset of ethnic conflict, taking into account the correlation between the residuals from the equations (Greene, 2003; Hultquist, 2015).

Dataset

To understand why ethnic conflicts spread, we must first identify those states that are at risk of contagion (Buhaug and Gleditsch, 2008; Forsberg, 2014b). Following Forsberg’s dyadic conceptualization of conflict contagion, we focus on the risk of violence spillover between proximate states. Each dyad in the dataset is composed of state A, where an armed ethnic conflict has begun and is ongoing at any point in time from 1976 to 2009, and state B, e.g. each one of state A’s proximate states.⁸ According to Gleditsch and Ward (2001), “proximity” is defined as a distance of 950 km or less between the two states.⁹ In alternative specifications, we restrict our sample to states B that are within 100 km of state A and (separately), those states B that share a border with state A, e.g. direct neighbors. The potential impact of ethnic conflict in state A on the behavior of the government and the ethnic group in state B is temporally constrained. Thus, each dyad remains active for every year of the ongoing armed conflict in state A, and for five years after the conflict is terminated (Forsberg, 2014b: 153). The dataset encompasses 897 dyads.

Dependent variables

In the first equation of the simultaneous model (repression escalation equation), the dependent variable is a binary indicator for an increase in the level of repression in state B from one year to the next. In the second equation of the simultaneous model (ethnic conflict onset equation), the dependent variable is the onset of ethnic conflict in state B.

For our measure of repression escalation, we rely on the Political Terror Scale (PTS). Based on information derived from reports published by Amnesty International and the US State Department, the PTS documents the intensity, scope, and range of violent measures employed directly by the state, namely, the police, military, secret service, and the national guard, as well as state-sponsored paramilitaries and state-affiliated death squads against those within its borders (Gibney et al., 2016). The PTS score is an ordinal measure that places state repression on a five-level continuum, ranging from limited (1) to most severe (5) repression. For our purposes, PTS levels 4 and 5 are most pertinent. According to Gibney et al. (2016), the substantive interpretation for PTS levels 4 and 5 is as follows:

Level 4—civil and political rights violations have expanded to large numbers of the population. Murders, disappearances, and torture are a common part of life. In spite of its generality, on this level terror affects those who interest themselves in politics or ideas.

Level 5—terror has expanded to the whole population. The leaders of these societies place no limits on the means or thoroughness with which they pursue personal or ideological goals.

In the repression escalation equation, the dependent variable Repression Escalation is coded 1 if the annual PTS score reflects an increase in the severity of state violence, specifically, a rise from lower levels of political repression to levels 4 or 5.¹⁰ When defining the baseline category for this variable it is crucial to differentiate between situations where there has been no change in the level of repression because a given state remains under a secure rule of law and is steadfast in its respect for human rights, and situations where human rights conditions remain deplorable as the regime relentlessly persecutes and victimizes civilians from one year to the next. Cases where the repression remains static at levels 4 or 5 are therefore excluded from the analysis. The reference category for Repression Escalation thus encompasses situations of where repression has remained relatively low (levels 1–3) and situations where human rights conditions in a given country have deteriorated from the previous year, but have not reached the state terror levels of 4 and 5. Because duration dependency is likely when working with this type of data, we include one-year lags of the dependent variable as an independent variable in the repression escalation equation. Finally, in order to better understand how the patterns of repression escalation from one year to the next affect the likelihood of conflict spreading from state A to state B, and to directly test hypothesis H5, we include Repression Escalation as an independent variable in the ethnic conflict onset equation.

In the ethnic conflict onset equation, we shift attention to the determinants of ethnic conflict spread. Following the standard Uppsala Conflict Data Program (UCDP) definition, an armed conflict is defined as “a contested incompatibility that concerns government and/or territory where the use of armed force between two parties, of which at least one is the government of a state, results in at least 25 battle-related deaths in one calendar year” (UCDP, http://www.pcr.uu.se/research/ucdp/definitions/, 2015). Ethnic conflicts are defined as “wars among communities that are in conflict over the power relationship that exists between those communities and the state” (Sambanis, 2001: 261). Ethnicity is the principal cause of this subset of intrastate armed conflicts, and the warring parties are mobilized along ethnic lines.

The dependent variable New Ethnic Conflict Onset is coded 1 for the year hostilities between the warring parties first reach 25 battle related deaths or if a previous conflict reignites after at least three years of inactivity. The variable is then coded 0 for the remaining years of activity after the initial or repeated onset. During the period under observation, or the time each proximate state B is considered at risk of conflict contagion, a state can experience more than one onset of ethnic conflict. A recent history of ethnic strife increases the likelihood of subsequent violence, and countries that have experienced ethnic wars are more likely to see these conflicts recur as well as experience new conflict onset. To control for temporal correlation, we follow a procedure recommended by Raknerud and Hegre (1997) and include a decay function of time passed since the last conflict onset in the analysis.¹¹

Independent variables

Ethnic kinship ties

To identify the presence of transnational ethnic kinship ties, we rely on Forsberg’s (2014b) Ethnic Kin variable. Ethnic Kin is coded 1 if at least one of the rebel organizations fighting in state A is linked to a specific ethnic constituency that also constitutes a politically relevant ethnic group in state B. The emphasis on politically relevant ethnic groups follows the definitions and coding rules of the Ethnic Power Relations dataset (Cederman et al., 2010). Where kinship ties exist, Ethnic Kin is coded 1 for every year after the conflict first becomes active in state A and for five years after the conflict is terminated, at which point, unless the conflict recurs, the dyad is censored.

Dissent

To estimate the influence of largely nonviolent political dissent on the likelihood of repression escalation, we use Banks’s measure of Demonstrations_{(state B)}, defined as peaceful, public protests of at least 100 people gathered to voice their opposition to government policies or government authority (Banks, 2011).

Regime ideology

To account for regime ideology, we include the variable Nationalist State_{(state B)} which takes the value of 1 if the executive party’s platform advocates for the defense of a national or ethnic identity, especially at the expense of other politically relevant groups, or if the ruling party has fought for independence from colonial power, either militarily or diplomatically (Beck et al., 2001).

Group-specific attributes

We employ a binary indicator for Large Ethnic Group_{(state B)} to identify those ethnic minorities that constitute at least 10% of the national population in state B. Data on the relative population size of the different ethnic groups is derived from the Ethnic Power Relations dataset (Cederman et al., 2010). Because geographically concentrated groups have better political, socio-cultural, and economic networks they can draw on to both commence and sustain an armed struggle against the state (Lichbach, 1995; Toft, 2003), territorial settlement patterns are likely to influence the extent to which the state sees the group as a potential threat as well as the likelihood of new armed conflict onset. The variable Territorially Concentrated Group_{(state B)} records whether or not the group has a set territorial area wherein the majority of its members are concentrated (Wucherpfennig et al, 2011).

Political opportunity

We use two measures that reflect a dynamic and highly contingent political environment: Government Crises_{(state B)} is defined as any rapidly developing situation that threatens to bring the downfall of the present regime. This definition excludes situations of mass revolt aimed at the overthrow of the current state, and is therefore an appropriate measure of intra-regime political instability (Banks, 2011). The variable Regime Transition_{(state B)} is coded 1 for the year when a normative and substantive change in political authority has taken place.¹² Regime transitions are defined by “the ending of an established polity and the beginning of a new polity,” thereby providing “a measure of the vulnerability and durability (i.e. persistence) of a particular regime and its authority patterns” (Marshall et al., 2013: 30). We use Cheibub’s et al. (2009) coding of transitions based on their six-fold classification of political regimes according to the institutions capable of removing the government from power, distinguishing between parliamentary, presidential, and mixed democracies, and monarchic, military, and civilian dictatorships.

Controls

Whether the ethnic conflict in state A is being fought for control of the government or over territory may affect the likelihood of conflict spread. Secessionist struggles are presumably more likely to spread, especially if the rebels’ vision for an independent homeland incorporates co-ethnics who currently reside beyond the borders of the warring state (Saideman and Ayers, 2000). To test this proposition, we include the variable Territorial Conflict_{(state A)} to account for armed conflicts fueled by demands for secession or autonomy (Gleditsch, et al., 2002). We also include a measure of conflict severity: a binary variable Civil War_{(state A)} coded 1 for the years in which the armed conflict reached the 1000 battle deaths threshold.

The “democratic domestic peace” thesis holds that for normative and institutional reasons, severe political repression is relatively rare in democracies (Davenport, 1995, 1999, 2007b; Poe and Tate, 1994). There is however no consensus in the repression literature on whether the impact of democracy on repression is linear, or as the “More Murder in the Middle” thesis holds, takes the form of an inverted U-shape (Regan and Henderson, 2002). To tests these different propositions, the models include two measures of regime type: Polity and its squared term, Polity squared. The variable is drawn from the Polity IV database, which ranks countries on a 21-point scale, ranging from −10 (hereditary monarchy) to +10 (consolidated democracy).

Population size and low levels of economic development often correlate with internal conflict and widespread repression, as poor and populous countries tend to be more volatile and violent (Collier and Hoeffler, 2004; Fearon and Laitin, 2003). We therefore include lagged measures of Population (log) and GDP per capita (log) in state B in our analysis, using data from Gleditsch (2002).

Analysis

We begin our analysis by considering all those proximate states B that are at risk of contagion amidst ongoing ethnic conflict in state A. This allows us to demonstrate that conflict in one state can lead to both preemptive repression escalation and the outbreak of new ethnic conflict in nearby states, as well as to assess the impact of transnational kinship ties on these two outcomes of interest. The results from the seemingly unrelated bivariate probit models are reported in Table 1. The bottom half of each model reports estimates from the ethnic conflict onset equation. The significant rho coefficients at the bottom of the table for each model indicate that there is a strong correlation between the residuals in two equations. The Wald test of independent equations in different models confirms that the equations are dependent by reporting χ² values ranging from 82.2 to 105.4 which are all significant at the 99% level. In other words, repression escalation and the risk of new armed conflict onset are endogenous as expected.

Table 1.

Bivariate probit estimates of repression escalation and ethnic conflict onset

	(1)	(2)	(3)	(4)	(5)
Repression escalation equation
Repression Escalation_(t−1)	0.636***	0.66***	0.641***	0.629***	0.636***
	(0.064)	(0.061)	(0.065)	(0.063)	(0.064)
Ethnic Kin	0.224**			0.216**	0.224**
	(0.102)			(0.102)	(0.102)
Large Ethnic Group_{(state B)}		0.274
		(0.212)
Territorially Concentrated Group_{(state B)}			0.542***
			(0.149)
GDP per capita_(log)	−0.127***	−0.12***	−0.117***	−0.134***	−0.127***
	(0.029)	(0.03)	(0.03)	(0.03)	(0.029)
Population_(log)	0.237***	0.246***	0.24***	0.236***	0.237***
	(0.024)	(0.025)	(0.023)	(0.025)	(0.024)
Polity	−0.019***	−0.02***	−0.02***	−0.019***	−0.019***
	(0.005)	(0.004)	(0.004)	(0.005)	(0.005)
Polity squared	−0.007***	−0.007***	−0.007***	−0.007***	−0.007***
	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
Demonstrations_{(state B)}	0.045***	0.041***	0.034***	0.047***	0.045***
	(0.011)	(0.011)	(0.011)	(0.012)	(0.011)
Nationalist State_{(state B)}	0.201***	0.253***	0.229***	0.204***	0.201***
	(0.055)	(0.061)	(0.063)	(0.055)	(0.055)
Constant	−2.184***	−2.31***	−2.275***	−2.126***	−2.184***
	(0.301)	(0.326)	(0.314)	(0.308)	(0.301)
Ethnic conflict onset equation
Repression Escalation	0.322***	0.328***	0.313***	0.348***	0.322***
	(0.109)	(0.104)	(0.108)	(0.106)	(0.108)
Ethnic Kin	0.307***			0.322***	0.302***
	(0.091)			(0.096)	(0.091)
Large Ethnic Group_{(state B)}		0.368***
		(0.125)
Territorially Concentrated Group_{(state B)}			0.395***
			(0.148)
GDP per capita_(log)	−0.302***	−0.294***	−0.291***	−0.266***	−0.309***
	(0.041)	(0.04)	(0.04)	(0.041)	(0.041)
Population_(log)	0.079***	0.088***	0.082***	0.087***	0.075***
	(0.025)	(0.026)	(0.026)	(0.027)	(0.024)
Polity	0.011**	0.009**	0.009**	0.012**	0.011**
	(0.004)	(0.004)	(0.004)	(0.005)	(0.004)
Polity squared	−0.001	−0.001	−0.001	−0.002	−0.001
	(0.002)	(0.001)	(0.002)	(0.002)	(0.002)
Government Crises_{(state B)}	0.215***	0.204***	0.211***		0.211***
	(0.051)	(0.053)	(0.05)		(0.05)
Regime Transition_{(state B)}				0.436***
				(0.15)
Territorial Conflict_{(state A)}					0.082
					(0.085)
Decay Function	0.594***	0.59***	0.583***	0.684***	0.598***
	(0.101)	(0.102)	(0.102)	(0.094)	(0.099)
Constant	−0.55	−0.65*	−0.603	−0.897**	−0.486
	(0.376)	(0.378)	(0.374)	(0.376)	(0.372)
N	7,668	7,814	7,814	7,548	7,668
Rho	0.423***	0.418***	0.414***	0.455***	0.426***
	(0.046)	(0.045)	(0.046)	(0.044)	(0.046)
χ² (1)	84.9***	86.6***	82.2***	105.4***	87.4***

Robust standard errors clustered on state A in parentheses. *p < 0.1; **p < 0.05; ***p < 0.01.

χ² (1) reports the Wald tests of independent equations for bivariate probit models.

The results from Models 1–5 offer strong support for our key claims. The Ethnic Kin variable is consistently statistically significant in the expected direction; as proposed in hypothesis H1, transnational ethnic kinship ties are associated with more frequent and severe human rights violations. The positive and significant coefficient on the Demonstrations_{(state B)} variable affirms the claim put forth in hypothesis H2, and suggests that even largely non-violent political protest is likely to be met with a highly violent and repressive response by the state. This finding lends credibility to our broader argument that international factors play an important role in shaping state threat perception and subsequent behavior. As stipulated by hypothesis H3, amidst nearby conflict, states led by nationalist leaders are more likely to resort to extreme methods of coercive control. In line with hypothesis H4, the presence of a relatively large transnational ethnic group increases the likelihood of more severe repression, albeit the coefficient is not statistically significant. The likelihood of repression escalation is also higher where the transnational ethnic kin group is territorially concentrated.

Shifting attention to the bottom half of the table, to the new ethnic conflict onset equation, the significant and positive coefficient on the Repression Escalation variable offers support for hypothesis H5, suggesting that deteriorating human rights conditions are a viable predictor of conflict spreading from state A to state B. According to Models 1, 4, and 5, the Ethnic Kin variable is statistically significant in the expected direction: amidst nearby instability and violence, the risk of armed conflict onset is higher in the presence of ethnic kinship ties. The likelihood of new conflict onset is higher where the transnational ethnic group is relatively large and territorially concentrated. The positive and significant coefficients for Government Crises_{(state B)} and Regime Transition_{(state B)} offer support for hypothesis H6 regarding how the opening of political space allows potentially violent opposition groups to coalesce, and that their ensuing confrontations with the government can escalate into a full-scale armed conflict.

The results on the control variables in Models 1–5 largely correspond to our expectations and previous research. Poor, populous states are more likely to experience both repression escalation and the outbreak of armed conflict within their own borders when a nearby state is plagued by ethnic violence. There appears to be a curvilinear relationship between democracy and repression, whereas the likelihood of new ethnic conflict onset is lower in more democratic states.

Overall, the above analysis supports the central claims of this study: international factors, regime ideology and group-specific attributes shape state threat perception, and subsequently, the decision to escalate repression to counter potential popular mobilization inspired by political developments beyond the state’s borders. At the same time, increasing state repression, transnational ethnic kinship ties, and political crises and transitions can precipitate the spread of ethnic conflict across international borders.

Our contention is that preemptive repression escalation is one important pathway by which ethnic conflict spreads in the presence of transnational ethnic kinship ties. Whereas the findings from the bivariate probit models demonstrate that transnational ethnic kinship ties increase the risk of both preemptive repression escalation and new ethnic conflict onset, a more direct test of the aforementioned pathway is reported below. Table 2 presents the results from a logit analysis estimating the likelihood of new ethnic conflict onset only in those states B that share ethnic kinship ties with the conflict-affected state A.¹³ The analysis in Models 7 and 8 encompasses all proximate states B with ethnic kinship ties to state A, while Models 9 and 10 focus exclusively on those states B that share a border with state A.

Table 2.

Repression escalation, transnational ethnic kinship ties, and new ethnic conflict onset

Variables	(7)	(8)	(9)Direct Neighbor	(10)Direct Neighbor
Repression Escalation	0.872**	0.948***	0.865**	0.957***
	(0.347)	(0.353)	(0.337)	(0.324)
Large Ethnic Group_{(state B)}	−0.569**	−0.505**	−0.815***	−0.742***
	(0.259)	(0.249)	(0.279)	(0.252)
GDP per capita_(log)	−0.883***	−0.804***	−0.858***	−0.777***
	(0.16)	(0.157)	(0.172)	(0.177)
Population_(log)	0.186**	0.212**	0.145	0.169
	(0.09)	(0.089)	(0.111)	(0.105)
Polity	0.033	0.029	0.039	0.033
	(0.022)	(0.024)	(0.025)	(0.028)
Polity squared	−0.004	−0.006	0.000	0.000
	(0.007)	(0.007)	(0.007)	(0.007)
Territorial Conflict_{(state A)}	0.421	0.434	0.298	0.285
	(0.327)	(0.341)	(0.31)	(0.332)
Civil War_{(state A)}	−0.406	−0.516	−0.526*	−0.69*
	(0.286)	(0.333)	(0.304)	(0.366)
Nationalist State × Large Ethnic Group _{(state B)}	2.97**	2.788**	3.249***	2.962**
	(1.201)	(1.204)	(1.249)	(1.289)
Nationalist State_{(state B)}	−2.627**	−2.596**	−2.634**	−2.6**
	(1.123)	(1.128)	(1.126)	(1.129)
Government Crises_{(state B)}	0.347*		0.312
	(0.194)		(0.202)
Regime Transition_{(state B)}		1.233*		1.43*
		(0.728)		(0.794)
Decay Function	0.421	0.625	0.212	0.42
	(0.482)	(0.486)	(0.534)	(0.539)
Constant	2.206	1.289	2.521*	1.558
	(1.413)	(1.480)	(1.499)	(1.542)
Pseudo R²	0.1796	0.1884	0.1747	0.1895
Observations	1,284	1,284	922	920

Entries are logit coefficients, with robust standard errors clustered on state A reported in parentheses. *p < 0.1; **p < 0.05; ***p < 0.01 (two-tailed).

Figure 1 illustrates the average marginal effects of the main independent variables on the probability of new ethnic conflict onset in proximate (Model 8) and direct neighbor states B (Model 10). While the results are largely consistent across the different spatial delimitations, the effect is somewhat stronger for direct neighbors.

Figure 1.

Average marginal effects of key variables on probability of new ethnic conflict onset in state B.

We find that in the presence of transnational ethnic kinship ties, escalating state repression can be seen as a precursor to the spread of armed conflict. Substantively, the probability of new ethnic conflict onset increases by nearly 150% in response to an adverse change in the level of state repression compared with when widespread and systematic rights violations are largely absent.

Our previous analysis demonstrates that nationalist regime ideology and the relative size of the domestic ethnic group influence the state’s decision to increase repression in response to nearby ethnic conflict. Models 7–10 test whether these factors also shape the state’s ability to effectively counter the spread of ethnic violence from abroad by estimating their effect on the risk of new ethnic conflict. Interestingly, we find that in situations where a nationalist government is confronted with a relatively large transnational ethnic minority group whose kin are involved in conflict across the border (Model 10), the predicted probability of new ethnic conflict onset increases to nearly 47%. This state of affairs describes the process of conflict contagion throughout the Balkans in the 1990s, as well as the current escalation of violence in Turkey. In contrast, for countries where the executive is not controlled by a nationalist party, neither the presence of relatively large ethnic minority groups (the reference category for the interaction term Nationalist State × Large Ethnic Kin_{(state B)}), nor the presence of smaller ethnic kin groups appears to increase the likelihood of new ethnic conflict. Finally, the positive and significant coefficients on Government Crises_{(state B)} and Regime Transition_{(state B)} lend support to our argument about the importance of political opportunity as a catalyst for popular mobilization against the state.

Conclusion

This study aims to advance our understanding of the process whereby ethnic conflict spreads from one state to its neighbors by integrating international factors, state and ethnic group behavior, and domestic political conditions into a unified theoretical and empirical framework. We find that states are more likely to respond to ethnic conflict in their neighborhood by preemptively escalating repression against their own citizens in the presence of relatively large, territorially concentrated transnational ethnic groups, and that nationalist governments are more likely to use violent measures to stifle potential opposition challenges inspired by nearby rebellion. Overall, there is strong support for our the theory that the spread of ethnic conflict across state boundaries follows from increasingly repressive state policies combined with political opportunity that allows the transnational ethnic group to mobilize in response.

We are not the first to argue that fear of civil conflict spreading across borders can prompt leaders to repress their citizens preemptively in order to prevent a potential rebellion at home (Brathwaite, 2014; Danneman and Ritter, 2014). To our knowledge, however, this is the first study to directly test the impact of transnational ethnic kinship ties on both preemptive repression and the spread of ethnic conflict. Thus, in addition to explicitly examining how international developments shape state threat perception and subsequent behavior, this project also offers new insights about a largely overlooked mechanism through which ethnic conflicts spread.

The aggregate cross-national research design has its faults; the Political Terror Scale measures of state repression account for human rights violations on a national scale, and may not fully reflect the experience of ethnic groups linked to nearby conflicts. At the same time, analysis at the country-year level has allowed us not only to demonstrate that conflict in one state does lead to preemptive repression in nearby states, and that widespread repression can in fact escalate to armed conflict, but also to closely consider the influence of transnational ethnic kinship ties on both of these interlinked processes.

Given the number of ongoing conflicts around the world involving ethnic kin groups, a better understanding of the dynamic interaction between state perceptions of security threats, repression, and ethnic group behavior can improve policy interventions designed to avert the spread of ethnic conflict. Taking into account that repressive state measures targeting ethnic minority groups are often preceded by or coincide with restrictions or exclusion from access to political power, we also sought to evaluate the robustness of our findings by testing whether an adverse change in political status exacerbates the risk of ethnic conflict spread in the presence of transnational ethnic kinship ties. The resuts (reported in the Statistical Appendix) suggest that recent loss of political status significally increases the probability of new ethnic conflict onset in state B where the downgraded ethnic group shares kinship ties with one of the groups in the conflict-affected state A.¹⁴ If political disenfranchisement and deteriorating civil and human rights conditions signal a growing risk of ethnic conflict, then a myopic focus on economic determinants of stability may be ill-advised. Finally, considering that government crises and regime transitions play an important role in the process of conflict contagion, democratization and political reform initiatives such as those currently being undertaken in Myanmar, for example, should be navigated with close attention to ethnic politics and conflicts in neighboring countries such as India, Bangladesh, and Thailand.

Footnotes

Acknowledgements

We would like to thank Erica Forsberg for her generous help with data. We are also grateful to Jason Quinn, Tanisha Fazal, and several anonymous reviewers for their comments on earlier versions.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Supplementary material

Replication materials and the Statistical Appendix are available on SAGE’s CMPS website.

Notes

References

Banks

. (2011) Cross-national Time-series Data Archive User’s Manual. Distributed by Databanks International, Jerusalem.

Beck

Clarke

Groff

Keefer

Walsh

(2001) New tools in comparative political economy: The Database of Political Institutions. World Bank Economic Review 15(1): 165–176.

Black

(2012) The spread of violent civil conflict: Rare, state-driven, and preventable. PhD dissertation. Department of Political Science, Massachusetts Institute of Technology, Cambridge, MA.

Bosker

de Ree

(2014) Ethnicity and the spread of civil war. Journal of Development Economics 108: 206–211.

Braithwaite

(2010) Resisting infection: How state capacity conditions conflict contagion. Journal of Peace Research 47(3): 311–319.

Brathwaite

KJH

(2014) Repression and spread of ethnic conflict in Kurdistan. Studies in Conflict and Terrorism 37(6): 473–491.

Brown

Farhad

(1995) Playing the “Communal Card”: Communal Violence and Human Rights. New York: Human Rights Watch.

Brown

(1996) The causes and regional dimensions of internal conflict. In: Brown

(ed.), The International Dimensions of Internal Conflict. Cambridge, MA: MIT Press.

Buhaug

Gleditsch

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

10.

Buhaug

Cederman

L-E

Rød

(2008) Disaggregating ethno-nationalist civil wars: A dyadic test of exclusion theory. International Organization 62(3): 531–551.

11.

Carey

(2010) The use of repression as a response to domestic dissent. Political Studies 58: 167–186.

12.

Cederman

L-E

Girardin

Gleditsch

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

13.

Cederman

L-E

Girardin

Gleditsch

(2009b) Ethno-nationalist dyads and civil war: A GIS-based analysis. Journal of Conflict Resolution 53(4): 496–525.

14.

Cederman

L-E

Wimmer

Min

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

15.

Cederman

L-E

Gleditsch

Salehyan

Wucherpfennig

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

16.

Cheibub

Gandhi

Vreeland

(2009) Democracy and dictatorship revisited. Public Choice 143(2–1): 67–101.

17.

Collier

Hoeffler

(2004) Greed and grievance in civil wars. Oxford Economic Papers 56(4): 563–596.

18.

Collier

Elliott

Hegre

Hoeffler

Reynal-Querol

Sambanis

(2003) Breaking the Conflict Trap: Civil War and Development Policy. Washington, DC: World Bank and Oxford University Press.

19.

Danneman

Ritter

(2014) Contagious rebellion and preemptive repression. Journal of Conflict Resolution 58(2): 254–279.

20.

Davenport

(1995) Multi-dimensional threat perception and state repression: An inquiry into why states apply negative sanctions. American Journal of Political Science 39(3): 683–713.

21.

Davenport

(1999) Human rights and the democratic proposition. Journal of Conflict Resolution 43(1): 92–116.

22.

Davenport

(eds) (2000) Paths to State Repression: Human Rights Violations and Contentious Politics. Lanham, MD: Rowman & Littlefield.

23.

Davenport

(2007a) State repression and political order. Annual Review of Political Science 10: 1–23.

24.

Davenport

(2007b) State Repression and the Domestic Democratic Peace. Cambridge: Cambridge University Press.

25.

Fearon

(1998) Commitment problems and the spread of ethnic conflict. In: Lake

Rothchild

(eds), The International Spread of Ethnic Conflict: Fear, Diffusion, and Escalation. Princeton, NJ: Princeton University Press.

26.

Fearon

Laitin

(2003) Ethnicity, insurgency, and civil war. American Political Science Review 97(1): 80.

27.

Fisk

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

28.

Forsberg

(2014a) Diffusion in the study of civil wars: A cautionary tale. International Studies Review 16: 188–198.

29.

Forsberg

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

30.

Gibney

Cornett

Wood

Haschke

Arnon

(2016) The Political Terror Scale 1976–2015. Political Terror Scale website. Available from: http://www.politicalterrorscale.org.

31.

Gleditsch

(2002) Expanded trade and GDP data. The Journal of Conflict Resolution 46(5): 712–724.

32.

Gleditsch

(2007) Transnational dimensions of civil war. Journal of Peace Research 44(3): 293–309.

33.

Gleditsch

Ruggeri

(2010) Political opportunity structures, democracy, and civil war. Journal of Peace Research 47(3): 299–310.

34.

Gleditsch

Ward

(2001) Measuring space: A minimum-distance database and applications to international studies. Journal of Peace Research 38(6): 739–758.

35.

Gleditsch

Wallensteen

Eriksson

Sollenberg

Strand

(2002) Armed conflict, 1945–1999: A new dataset. Journal of Peace Research 39(5): 615–637.

36.

Goldstein

(1978) Political Repression in Modern America: From 1870 to the Present. Boston, MA: Schenckman/G.K. Hall.

37.

Goldstone

Kaufmann

Duffy Toft

(2012) Political Demography: How Population Changes Are Reshaping International Security and National Politics. Oxford: Oxford University Press.

38.

Greene

(2003) Econometric Analysis, 5th edn. Upper Saddle River, NJ: Prentice Hall.

39.

Gurr

(2000). Peoples versus States: Minorities at Risk in the New Century. Washington, DC: United States Institute of Peace Press.

40.

Hegre

Sambanis

(2006) Sensitivity analysis of the empirical literature on civil war onset. Journal of Conflict Resolution 50(4): 508–535.

41.

Hendrix

. (2010) Measuring state capacity: Theoretical and empirical implications of the study of civil conflict. Journal of Peace Research 47(3): 274

42.

Horowitz

(1985) Ethnic Groups in Conflict. Berkeley, CA: University of California Press.

43.

Hultquist

(2015) Is collective repression an effective counterinsurgency technique? Unpacking the cyclical relationship between repression and civil conflict. Conflict Management and Peace Science. Available from: https://dx-doi-org.web.bisu.edu.cn/10.1177/0738894215604972

44.

Kaufman

(2001) Modern Hatreds: The Symbolic Politics of Ethnic War. Ithaca, NY: Cornell University Press.

45.

Konaev

(2016) Power and purpose: The durable settlement of self-determination conflicts. PhD dissertation. Department of Political Science, University of Notre Dame, Notre Dame, IN.

46.

Krcmaric

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

47.

Kuran

(1998) Ethnic dissimilation and its international diffusion. In: Lake

Rothchild

(eds), The International Spread of Ethnic Conflict: Fear, Diffusion, and Escalation. Princeton, NJ: Princeton University Press.

48.

Lake

Rothchild

(1998) Spreading fear: The genesis of transnational ethnic conflict. In: Lake

Rothchild

(eds), The International Spread of Ethnic Conflict: Fear, Diffusion, and Escalation. Princeton, NJ: Princeton University Press.

49.

Lichbach

(1995) The Rebel’s Dilemma. Ann Arbor, MI: University of Michigan Press.

50.

Linz

(2000) Totalitarian and Authoritarian Regimes. Boulder, CO: Lynne Renner.

51.

Lischer

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

52.

Los Angeles Times (2015) Latvia, With a Large Minority of Russians, Worries about Putin’s Goals, 2 May.

53.

Marshall

Jaggers

(2009) POLITY IV Dataset [Computer file: version p4v2007]. George Mason University and Center for Systemic Peace.

54.

Marshall

Gurr

Jaggers

(2013) POLITY IV PROJECT: Dataset Users’ Manual. Center for Systemic Peace.

55.

Melander

(2009) Selected to go where murderers lurk? Conflict Management and Peace Science 26(4): 389–406.

56.

Poe

Tate

(1994) Repression of human rights to personal integrity in the 1980s—a global analysis. American Political Science Review 88(4): 853–872.

57.

Raknerud

Hegre

(1997) The hazard of war: Reassessing the evidence for the democratic peace. Journal of Peace Research 34(4): 385–404.

58.

Regan

Henderson

(2002) Democracy, threats and political repression in developing countries: Are democracies internally less violent? Third World Quarterly 23(1): 119–136.

59.

Ritter

(2014) Policy disputes, political survival, and the onset and severity of state repression. Journal of Conflict Resolution 58(1): 143–168.

60.

Roessler

(2011). The enemy within: Personal rule, coups, and civil war in Africa. World Politics 63(2): 300–346.

61.

Rousseau

Garcia-Retamero

(2007) Identity, power and threat perception: A cross-national experimental study. Journal of Conflict Resolution 51(5): 744–771.

62.

Saideman

Ayres

(2000) Determining the causes of irredentism: Logit analyses of Minorities at Risk Data from the 1980s and 1990s. Journal of Politics 62(4): 1126–1144.

63.

Salehyan

(2007) Refugees and the study of civil war. Civil Wars 9(2): 127–141.

64.

Salehyan

Gleditsch

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

65.

Sambanis

(2001) Do ethnic and non-ethnic civil wars have the same causes? A theoretical and empirical inquiry (Part I). Journal of Conflict Resolution 45(3): 259–282.

66.

Schock

(1996) A conjunctural model of political conflict: The impact of political opportunities on the relationship between economic inequality and violent political conflict. The Journal of Conflict Resolution 40(1): 98–133.

67.

Shellman

(2006) Leaders’ motivations and actions: Explaining government-dissident conflict-cooperation processes. Conflict Management and Peace Science 23(1): 73–90.

68.

Sobek

(2010) Masters of their domains: The role of state capacity in civil wars. Journal of Peace Research 47: 267.

69.

Toft

(2003) The Geography of Ethnic Violence: Identity, Interests, and the Indivisibility of Territory. Princeton, NJ: Princeton University Press.

70.

Valentino

Huth

Balch-Linsay

(2004) Draining the sea: Mass killing and guerrilla warfare. International Organization 58(2): 375–407.

71.

Wucherpfennig Weidmann

Girardin

Cederman

L-E

Wimmer

(2011) Politically relevant ethnic groups across space and time: Introducing the GeoEPR Dataset. Conflict Management and Peace Science 28: 423–437.

72.

Young

(2009) State capacity, democracy and the violation of personal integrity rights. Journal of Human Rights 8(4): 283–300.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.69 MB

0.00 MB