The Onset of Ethnic War

Abstract

This article develops a general theory regarding the onset of ethnic war, starting with two analytic innovations: a mechanism-based approach toward social facts and an emphasis on dynamic interactions. I deploy two meta-mechanisms – the security dilemma/spiral model and intergroup-intragroup interactions – as meta-synthesizers. I then bring together the numerous factors and mechanisms scattered in the literature into a more integrative and dynamic theory of ethnic war by linking factors with immediate drivers of conflictual behavior via the two meta-mechanisms. The resulting theory not only integrates numerous factors and mechanisms identified within the existing literature, but it also reveals previously hidden or neglected factors, interactions, and mechanisms that point to fruitful directions for future inquires.

Keywords

ethnic war general theory factor driver mechanism interaction mobilization

Since 1945, intrastate wars have been far more frequent and more destructive than interstate wars, and more than half the intrastate wars have been ethnic based rather than related to class or ideology. Not surprisingly, ethnic civil war (hereafter ethnic war), as a specific form of civil war, has become one of the most visible fields of inquiry in the social sciences, producing an already voluminous and still growing literature.¹

The study of ethnic war, however, is not without problems (for earlier critical reviews, see Blattman and Miguel 2010; Brubaker and Laitin 1998; Fearon and Laitin 2000; Horowitz 2008; Kalyvas 2008; Sambanis 2001, 2004a; Tarrow 2007; Williams 1994; Wimmer, Cederman, and Min 2009). One fact stands out: the field has become extremely fragmented as it has expanded. From quantitative studies, we have many correlations that seem to link some factors with ethnic war or peace, often without causal mechanisms (Wimmer et al. 2009).² From qualitative studies (i.e., comparative case studies), we have many specific theories of ethnic war that look to particular factors and causal mechanisms. Not only are syntheses lacking, but students of ethnic war tend to pit some factors and mechanisms against others, as if their favored factors and mechanisms alone can adequately explain the complexities of ethnic war, and these different factors and mechanisms do not interact (Grigoryan 2007). In short, there is no general theory of ethnic war. Indeed, Brubaker and Laitin (1998) implicitly dismissed the possibility of a general theory of ethnic war (cf. Kaufman 2001; Sambanis 2001).

I conjecture that a key cause behind this state of the field is that we have yet to find appropriate tools that can bring most of the factors and mechanisms under the same roof. In this article, I advance a possible solution to this challenge. Critically building on earlier attempts at synthesizing (e.g., Fearon and Laitin 2000; Horowitz 1985, 2001; Kaufman 2001; Petersen 2002; Sambanis 2004a), I take several key steps toward a general theory of ethnic war, with two analytic innovations. First, I start with mechanisms rather than factors, in contrast to almost all existing studies. Specifically, I deploy two meta-mechanisms that drive intergroup conflict—the security dilemma/spiral and intergroup-intragroup interactions—as synthesizers. I then comb through the many factors that have been identified as contributing to ethnic war and ask, Can a factor be reasonably linked with ethnic war through the two meta-mechanisms and the more immediate drivers of conflictual behavior? In doing so, I am able to link most of the relevant factors with ethnic war.³ Second, I identify potential interactions of the various factors, immediate drivers, and mechanisms to bring factors and mechanisms together. The result is a more integrated theory of ethnic war.

What is a general theory of ethnic war? I consider a theory of ethnic war to be a general theory if it does two things. First, the theory states the general dynamics of ethnic war and identifies the essential ingredients for constructing theories or explanations for specific ethnic wars. Second, the theory subsumes or integrates several or many specific theories, thus achieving some degree of vertical and horizontal theoretical unification.⁴ More concretely, the general theory must integrate underlying factors and immediate drivers with ethnic war via well-defined mechanisms.

A well-constructed general theory or an integrated analytic framework serves at least three valuable purposes. First, an integrated theory helps us organize empirical data surrounding a complex social fact: an integrated theory brings some order to the field. Second, an integrated theory reveals neglected and hidden interactions of factors, mechanisms, and pathways, thus pointing to gaps within our existing knowledge and suggesting directions for further inquiries. Third, an integrated theory helps us measure both empirical and theoretical progress. Several more specific caveats are now in order.

First, I explicitly start with the position that ethnicity does matter (for recent discussions of ethnicity, see Chandra 2006; Chandra and Wilkinson 2008; Wimmer 2008, 2013). As such, I reject the position held by some rational choice theorists that there is no meaningful difference between ethnic war and nonethnic war, or as Walter (2001:89) put it, “all combatants [are] driven by the same cost calculations regardless of ethnic affiliation or identity [italics added]” (see also Fearon and Laitin 1996; Lake and Rothchild 1996; for critiques, see Kaufmann 2005; Wimmer 2008, 2013; Wimmer et al. 2009).

Second, I assume that some ethnic identities are already in place and can be manipulated. In other words, I start with some ethnic fractionalization, usually underpinned by linguistic, religious, or racial differences. By definition, ethnic wars are possible only within multiethnic countries (Sambanis 2001, 2004b). Although this reifies ethnicity somewhat, I avoid the muddy water of debating different approaches toward the construction of ethnic identities and nationalism (e.g., primordialism vs. instrumentalism vs. constructivism; for recent general surveys, see Fearon and Laitin 2000; Hall 1998; Smith 1998). Like most students of ethnic war, I also insist that ethnic identity (or categorization) is necessary but insufficient for ethnic war. I thus start with a modified, modernism-informed instrumentalist and constructivist stand, holding that (hard and soft) instrumentalism and (cultural and psychological) constructivism are inherently compatible (Brubaker 1996; Brubaker and Laitin 1998; Fearon and Laitin 2000; McKay 1982; Scott 1990; Wimmer 2002, 2008, 2013; cf. Arfi 1998 vs. Gagnon 1994–1995; Kaufman 2006 vs. Grigoryan 2007).

Third, I take some deep-rooted causes of nationalism and ethnic war—such as modernization, decolonialization, empire collapse, state nationalization, and exclusion/closure during the making of modern nation-states—as given. These major trends constitute key properties of the social system (Tang 2014). Not surprisingly, these trends profoundly shape states’ policies toward minority ethnic groups within their boundaries and, in turn, ethnic groups’ calculations of submission, resistance, and rebellion (Mylonas 2013; Wimmer 2002, 2013). These deeper causes have been thoroughly examined by others (e.g., Brubaker 1996; Gellner 1983; Wimmer 2002, 2013), so here I focus on the more immediate causes of ethnic war.

Fourth, because of the sheer volume of the literature on ethnic war, not to mention the broader literature on ethnic politics and civil war, a comprehensive survey of the literature is beyond the reach of any one article. I thus provide only a brief critical survey here, paying more attention to contributions that are of special interest to my goal of constructing a general theory of ethnic war.

Fifth, as suggested by the title, I concur with the notion that onset of an ethnic war and the duration/cessation of an ethnic war may have quite different dynamics. Moreover, peace building after a conflict may have very different dynamics from both the onset and the duration/cessation of a conflict. In this article I deal with onset only, leaving duration/cessation and postconflict peace building to subsequent works.

Sixth, because several mechanisms interact with numerous factors to drive ethnic war, enormous causal complexity undergirds the topic: many tragic roads can lead to ethnic war. As a result, although I identify several mechanisms and numerous factors, I cannot confidently talk about specific combinations of factors plus mechanisms as sufficient conditions or even nearly sufficient conditions for ethnic war. What I can talk about with some degree of confidence is that the immediate drivers and the two meta-mechanisms are necessary for ethnic war to break out, but some specific combinations may constitute the sufficient condition for a specific ethnic war or a subset of ethnic wars (Goertz and Mahoney 2012; Ragin 2000).

Finally, I do not systematically test the hypotheses presented here, although I do briefly mention several studies that are consistent with our proposed theoretical framework.⁵

The Study of Ethnic War Onset: State of the Field

General Critique

Besides typical methodological drawbacks such as a questionable universe of cases, the quality of data (i.e., indicator, coding, and measurement error), model specification, comparison of cases, level of aggregation,⁶ and generalizations (for earlier critiques, see Blattman and Miguel 2010; Brubaker and Laitin 1998; Hegre and Sambanis 2006; Horowitz 2008; Kalyvas 2008; Keen 2012; Sambanis 2004a, 2004b; Tarrow 2007; Van Houten et al. 2008; Wimmer et al. 2009), existing studies of ethnic war suffer from some more foundational defects.

Theories and empirical hypotheses are fundamentally different,⁷ but many studies, especially quantitative ones, implicitly equate deriving empirical hypotheses with theorizing. As a result, many quantitative studies on ethnic war are only marginally theoretical, if not atheoretical (Horowitz 2008; Kalyvas 2008; Sambanis 2004a; Tarrow 2007; Wimmer 2013), despite some fine exceptions. Most notably, although many quantitative studies link a host of factors with ethnic war, most of their results are no more than correlations, because they do not identify and test mechanisms that can potentially link these factors with ethnic war.

Partly because of the error of confusing hypotheses with theories, empirical results, even if robustly established, are correlational descriptions that cannot be meaningfully and consistently interpreted (Horowitz 2008). For instance, students of social movements, rebellion, and civil war have long believed that grievance is a key causal factor of these contentious politics. Yet Fearon and Laitin (2003), Collier and Hoeffler (2004), and Collier et al. (2009) all found that most indicators of (ethnic) grievance are not significantly associated with the onset of civil war. These authors took their results at face value and did not offer a coherent explanation for this “unexpected” finding. As Kalyvas (2006) pointed out, however, grievance rarely translates automatically into action (i.e., rebellion), because humans are strategic animals who can act contrary to their grievances under duress. Certainly, without active mobilization by the ethnic elite, few ethnic rebellions will ever get started. Thus, grievance is only a necessary but never a sufficient causal factor for the onset of ethnic war. Because regression implicitly assumes that each independent variable causes the outcome independently, and thus somehow sufficiently, regression is often hard pressed to uncover necessary but insufficient causal factors, even with interactive terms.

Some quantitative studies are conceptually misleading. Most prominently, Collier and Hoeffler (1998, 2004) and Collier et al. (2009) lumped together many factors under the labels of opportunity, greed, and grievance. However, their proxies for opportunity, such as extortion of natural resources, donations from diasporas, and subversion from hostile foreign governments, really tap into factors that underpin the capabilities of potential rebels. Yet opportunity and capability are not the same, even if we admit that (absolute and relative) capability partly underpins opportunities for rebellion.⁸

Existing studies, especially quantitative ones, neglect the critical role of political processes as mechanisms. Yet manipulation by the elite of ethnic identity, fear, and hatred is the most crucial process driving ethnic politics toward ethnic war. Existing studies also fail to note the possibility that intermediating political processes can change the effect of factors, and these changes can then loop back to shape the contour of ethnic war. For instance, earlier episodes of violence not only generate hatred but also produce situations such as weakened state capacity, ravaged economies, and political mobilizations along ethnic boundaries, which may fuel and facilitate further violence later on (Kalyvas 2008; Sambanis 2004a; Wimmer 2002, 2013). Likewise, outbidding and escalation by ethnic elites can come back to limit the options for accommodation later on. Without accounting for these political processes, merely linking some factors with ethnic wars operates in the black-box mode, which is increasingly rejected by scientific realism’s demand to identify mechanisms (Bunge 1997; Mahoney 2010).

Existing studies of ethnic war have failed to combine factors into an integrated theory via mechanisms (Horowitz 2008; Sambanis 2004a). Indeed, most qualitative studies pit their favored factors or mechanisms against others. Thus, Gagnon (1994–1995) emphasized the political side of elite manipulation of ethnic identities and fear, and Arfi (1998) focused on the social and symbolic side, as if these two sides are not part of the same story, even though they examined the same case of Yugoslavia. Likewise, both Arfi (1998) and Kaufman (2001, 2006, 2009, 2011) pitted factors that fall within the rational choice realm against more symbolic factors, as if material and symbolic interests cannot coincide. Most quantitative studies, heeding the underlying logic of statistical inferences, also pit factors that may affect different motives and constraints against each other, thus implicitly assuming that explanatory factors, as independent variables, are actually independent from each other in the real world. Existing quantitative techniques are hard pressed to handle interactions of more than three variables, especially when the effect of one or more factors can be nonlinear and nonmonotonic.

Existing studies, especially quantitative ones, tend to overemphasize the more easily measurable, and hence material, factors (e.g., gross domestic product per capita) or tangible interests (e.g., oil, gas, diamonds, territory). As such, they implicitly adopt a largely materialistic instrumentalism approach to ethnicity and ethnic war (e.g., Collier and Hoeffler 2004; Fearon and Laitin 1996). Such an approach, however, is misleading, as Connor ([1984] 1994) pointed out long ago (see also Cramer 2002). At least in the short run, much grievance can be based on subjective constructions: with emotions running high in conflictual situations, depictions of other groups’ (unwarranted) status and one’s own unjustly denied benefits can easily be created to mobilize for conflict (Horowitz 1985; Kaufman 2001; Petersen 2001, 2002; Rothschild 1981). Neither a purely materialistic instrumental approach nor a purely emotional constructivist approach is adequate for understanding ethnic war. The key challenge is to integrate emotional and material factors into an integrated framework.

Existing studies, especially quantitative ones, tend to overemphasize structural (i.e., macro-social, economic, and political) factors that are easier to measure. As such, they implicitly adopt a largely structuralist approach toward ethnic war that almost inevitably marginalizes the role of agents, especially that of ethnic entrepreneurs. A mostly structuralist approach, however, is incomplete at best: ethnic war cannot result without agents. Indeed, because ethnic entrepreneurs can construct grievance by manipulating emotions and perceptions, structural factors may be less potent in determining the outbreak of ethnic war than most quantitative studies have implied. Again, the challenge is to bring together agent and structure with the understanding that agents operate under structural constraints and opportunities (Tang 2014).

Earlier Attempts at Synthesizing: Unfulfilled Promises

Some synthesizing attempts have made important strides toward a more integrated theory of ethnic war. Unfortunately, none has come sufficiently close to providing us with a unifying analytic framework.

Horowitz (1985), for example, identified several immediate (master) drivers of ethnic conflict (e.g., interest and resentment) and numerous underlying factors that can be linked with the master drivers. Yet Horowitz generally failed to link the underlying factors with the master drivers, at least not systematically and rigorously.

Gurr and Harff (1994) made another earlier synthesizing attempt. They not only noted that ethnic war is impossible without ethnic mobilization, but they identified five to seven key sets of factors that affect mobilization. Their discussion has at least three key limits, however. Foremost, their elaboration of mobilization is inadequate. Second, although they examined several sets of factor that may affect mobilization, the theoretical rationales for singling out these factors and then linking them with ethnic mobilization are underdeveloped. Third, their case studies are simply too thin.

More recently, Collins (2008, 2012) presented an integrated theory of violence that emphasizes emotions (e.g., fear and anger) as immediate drivers of conflictual behavior. Yet this is mostly a micro-level theory, without much of a role for elites. As such, the theory is more applicable to individual or local mass-led violence but less so to elite-led violence, although it does provide important insights for understanding violence in general (see also Hale 2008).

In summary, most existing attempts at synthesizing are lacking. Next, I examine two synthesizing attempts, those of Kaufman (2001) and Petersen (2002), in more detail. Despite their numerous shortcomings, these two authors take some key steps toward a more integrated theory of ethnic war.

Kaufman (2001)

Very early on, Kaufman (1994–1995, 1996a, 1996b) identified the two-level game (i.e., the interaction of intra- and intergroup politics, especially leaders’ manipulation of the relations between ethnic groups to shift politics within their groups to their favor) and the security dilemma as two key links for understanding ethnic war. Kaufman also noted that for mass insurgency to occur, the fear of ethnic extinction must be very strong (e.g., Armenians vs. Azerbaijanis in the Nagorno-Karabakh conflict; Kaufman 1996a, 1996b). In addition, Kaufman recognized that elites can engage in ethnic outbidding, because they truly believe in the ethnonationalist cause (e.g., Franjo Tuđman of Croatia), for instrumental reasons, or because of a combination of both factors (e.g., Slobodan Milošević of Serbia [Kaufman 1996b]).

Building on his earlier work, Kaufman (2001) moved toward a more integrated theory. He began by noting that each of the four stories about ethnic war—ancient hatred, manipulative leaders, economic rivalry, and insecurity spirals—alone cannot possibly account for ethnic war. Likewise, although both rational choice theories and psychological theories provide important insights, each alone is inadequate for explaining ethnic war. As such, the correct way forward is to combine these stories (Kaufman 2001: 10-12). Moreover, Kaufman recognized that synthesis requires both mechanisms and factors, and he identified elites’ ethnic outbidding (as a key component of ethnic mobilization) as a key process of manipulating the fear of ethnic extinction or domination in driving ethnic war. He also singled out manipulation of ethnic symbols (or symbolic politics) as a key for understanding ethnic war. He stressed that symbolic politics requires existing myths and symbols and manipulative elites, and symbolic politics and a spiral of insecurity or fear can feed off each other (pp. 12, 36). Kaufman then summarized his (mature) theory as follows:

The necessary preconditions for ethnic war are ethnic myths and fears and the opportunity to act on them politically. Ethnic war occurs when the politics of ethnic symbolism goes to extremes, provoking hostile actions and leading to a security dilemma. In some cases, the turn toward extremism is mass-led; in other cases, it is elite-lead. Either way, war results from a process in which extremist politics and insecurity mutually reinforce each other in an escalatory spiral. (p. 12)

Unfortunately, Kaufman’s (2001) synthesizing enterprise suffers from serious flaws. First, in terms of its logic, it is unclear whether Kaufman’s three necessary conditions (i.e., myths justifying ethnic hostility, ethnic fears, and opportunity to mobilize and fight) together constitute a sufficient condition for ethnic war or whether additional necessary conditions have been left out (pp. 12, 32–34). Indeed, his subsequent discussion (pp. 34–36, 39–40) implies that his “three necessary conditions” require three more processes: “rising mass hostility, chauvinistic mobilization by leaders making extreme symbolic appeals, and a security dilemma between groups.” Without this logic constraint, Kaufman can easily smuggle in new factors and mechanisms in an ad hoc way, thus committing the “sin of commission” (Grigoryan 2007:184–86). And the fact that Kaufman (1994–1995) singled out different conditions earlier does not help our confidence in his new theory.

Second, although Kaufman (2001:9–10, 31–32, 34–36) recognized some utilities of the security dilemma/spiral model (SD/SM), he failed to properly grasp the nature and power of SD/SM. Kaufman (2001:37–38) argued that a security dilemma takes off only when one group responds in kind after the other group has already attacked. Because Kaufman relied on a scenario in which at least one side has already attacked, his explanation of ethnic war using a deep or imperialistic security dilemma is superficial: if one side has already attacked, war is almost inevitable with or without the security dilemma. Moreover, when one side has already attacked, the security dilemma ceases to operate (for a more detailed critique, see Tang 2011b; see also Grigoryan 2007:184–86).

Worse, in his more recent work, Kaufman (2006, 2009, 2011) equated the security dilemma with a standard rational choice approach (e.g., Fearon 1995; Lake and Rothchild 1996), simply because some rational choice theorists have emphasized insecurity as entailed by (emergent) anarchy. Yet those rational choice theorists have committed the sin of stripping the security dilemma of all its psychological components (for a more detailed critique, see Tang 2011b; see also Grigoryan 2007). Kaufman thus failed to grasp that SD/SM provides a key platform for integrating the numerous factors and mechanisms that have been singled out in the literature, including symbolic politics.

Third, Kaufman implicitly equated his ideal types of ethnic violence with real-world cases. Indeed, Kaufman (1996b:116, 2001:34–38) and his subsequent case studies (Kaufman 2006, 2009, 2011) mostly fit cases into his four ideal types of ethnic violence (i.e., popular chauvinism and mass insurgency as two forms of mass-led violence, government jingoism and elite conspiracy as two forms of elite-led violence) rather than offering genuinely new theoretical insights. As a result, Kaufman failed to grasp that mass-led violence cannot become a full-scale ethnic war without elite manipulation and mobilization of mass media (Brass 1997; Horowitz 2001; Mann 2005; cf. Kaufman 2001, chap. 3). Moreover, elites can engineer ethnic violence to drive ethnic war without mass-led violence in the first place. For instance, elites can engineer ethnic fear within their own groups by letting thugs inflict violence against the other group. This prompts reprisals, which then instill fear in the formerly moderate masses of the first group, leading to their willing participation in ethnic war (De Figueiredo and Weingast 1999; Fearon and Laitin 2000; Mueller 2000). Indeed, because elites can choose to clamp down mass-led violence if they are determined and decisive enough to intervene in the early stage of mass mobilization, as Kaufman’s (2001, chaps. 3 and 4) examination of the Karabakh conflict between Armenians and Azerbaijanis and the conflict between Georgians and Abkhaz/South Ossetians shows, for mass-led violence to become a full-scale war, elites must acquiesce, at the very least, if not tacitly encourage ethnic violence.⁹

Fourth, Kaufman (2001:32–34) conflated indicators of (internal) capabilities with indicators of (external) constraints or opportunity under the heading of “opportunity to mobilize and fight.” He thus failed to grasp that these two sets of factors, although interacting, are distinct and affect the contour of ethnic war differently. As will become clear, lumping capability with constraint clouds our understanding.

Finally, despite his claim of a synthetic story, Kaufman too fell into the bad habit of pitting things against each other. Most prominently, he consistently pitted instrumental calculation (i.e., rational choice) against symbolic politics (Kaufman 2001:17–29, 2006, 2009, 2011). Yet symbolic politics (i.e., manipulation of ethnic symbols) are processes, whereas myths and economic factors are facilitating or constraining factors. Likewise, Kaufman singled out ethnic outbidding as being totally detached from other factors, except for fear of ethnic distinction as part of the security dilemma. He failed to see not only that outbidding can be regulated by many other factors, such as preexisting mass hostility (Horowitz 1985) but also that outbidding is innately compatible with rational calculation. To forge ahead, we must synthesize various factors via processes rather than pitting different factors and processes against each other or pitting processes against factors.

Petersen (2002)

Building on earlier work that emphasized emotions in driving ethnic war (e.g., Horowitz 1985, 2001; Kaufman 2001; Scheff 1994; see also Collins 2008, 2012; Lebow 2008),¹⁰ Petersen’s (2002) work was an innovative attempt to synthesize several key (emotional) drivers of ethnic war into a coherent theory. Petersen (2002:32–33) began with a critique of the rational choice approach, insisting that ethnic war is as much about material stuff as about symbolic values. He then stressed that emotion is a mechanism (or switch) that triggers action to satisfy a pressing concern (pp. 3, 17–21, 36–37). As such, all emotions, except rage, are instrumental: instrumental calculation is perfectly compatible with emotions, contrary to the conventional wisdom propagated by rational choice theorists (Petersen 2002:17–19; for in-depth discussions of emotion and rationality, see Mercer 2005, 2010). Moreover, more often than not, a multiplicity of motivations, rather than a single motivation, drives individuals to participate in ethnic violence (Petersen 2002:3, 20, 25). Hence, Petersen’s framework contains at least honor (i.e., grievance and resentment), hatred, rage, and fear. Petersen (2002:19) also differentiated hatred from resentment, noting that hatred is underpinned by historical grievance or earlier episodes of conflict, whereas resentment is underpinned by status and self-esteem discrepancies (see also Lebow 2008). Equally important, Petersen (2002:19, 22, 75–84) noted that rage must be treated separately.

Unfortunately, Petersen’s framework also suffers from several shortcomings. First and most critically, although Petersen depicted emotions as immediate drivers of actions, he did not attempt to link emotions with the numerous underlying factors identified in existing literature on ethnic war. Indeed, banking on emotions, Petersen (2002) left many physical factors, including tangible interests, out of the picture, other than “structural changes (in status).” As a result, Petersen’s framework is so psychological that it cannot synthesize the numerous factors in the existing literature into an integrated framework.

Second, despite recognizing that individuals may be compelled to participate in ethnic violence for multiple reasons, Petersen consistently pitted different emotions against each other and eventually singled out resentment as the most potent driver of ethnic violence.

Third, Petersen (2002:34–36, 55) questioned the critical role of elite manipulation in driving ethnic war, thus contradicting his earlier discussion (Petersen 2001). As a result, he failed to see not only that mobilization and manipulation bank on emotions but also that manipulation and mobilization can exacerbate hostile emotions, and it is this vicious feedback loop between hostile emotion and mobilization that ultimately drives ethnic war.

Fourth, despite briefly mentioning the security dilemma, Petersen (2002:70–73) did not fully appreciate its power. As such, Petersen did not see the possibility that SD/SM and intergroup-intragroup interactions may be key tools for linking the numerous factors identified in the literature with ethnic war via the emotional drivers he identified.

Finally, Petersen’s treatment of the directions of emotions tended to be one sided. Although Petersen (2002:256) noted that status reversal is usually a potent factor in causing resentment, he implied that resentment goes only one way: only subordinate groups resent dominant groups. In reality, dominant groups may resent any distribution of things (real or perceived) to subordinate groups. Likewise, Petersen (2002:84) implicitly assumed that only beaten-down groups can develop rage, thus foreclosing the possibility of a dominant group challenged by a subordinate group getting into a rage and resorting to violence to reestablish dominance (“let them know who the real master is here”). The same critique applies to his treatment of hatred.

To summarize, both Kaufman and Petersen failed to see that (immediate) emotional drivers alone are insufficient drivers of ethnic war and are thus inadequate for understanding ethnic war: we need to bring other factors into the picture. As a result, both are far from an integrative and dynamic theory of ethnic war.

Two Building Blocks

The SD/SM

The security dilemma, first posited by Herz (1951) and Butterfield (1951) and crucially developed by Jervis (1976, 1978), is a key concept and theory of conflict in international relations. Prompted by the outbreak of so many conflicts after the collapse of the Soviet Union, Posen (1993) introduced the security dilemma to the study of ethnic war.¹¹ Unfortunately, much existing discussion of SD/SM in ethnic war is muddled, partially because SD/SM is not well defined and understood (Tang 2009, 2010, 2011b).

SD/SM captures some key dynamics of how interactions between two agents (states, ethnic groups, or even individuals) can drive them toward conflict (see Figure 1). Equally important, SD/SM is a potent tool for integrating material and psychological factors that have been identified as facilitating or constraining ethnic war, because SD/SM explicitly contends that physical and psychological factors can act as regulators of SD/SM, and final outcomes depend on the specific mixtures of these regulators (Jervis 1978; Snyder and Jervis 1999; Tang 2009, 2010, 2011b).¹² As such, SD/SM can serve as part of the foundation for constructing a general theory of ethnic war.

Figure 1.

The general dynamics of ethnic war.

For instance, the four psychological factors (fear, hatred, resentment, and rage) identified by Petersen (2002) can easily be integrated into SD/SM as psychological regulators of the spiral dynamics. Foremost, fear (of survival) is a central component of the security dilemma model. As such, fear of ethnic extinction, as emphasized by Kaufman (1996b:111, 115–16, 2001:25–27, 31–32), should be no different: fear of ethnic extinction must be a central component of the security dilemma in ethnic war. The same applies to hatred, whether it is “ancient hatred” (Kaplan 1994) or “modern hatred” (e.g., myths of past atrocity) (Kaufman 2001:25, 30–32, 2006). The same goes for resentment. Simply put, these psychological drivers color individuals’ understanding of the conflict of interest and affect their determination to enact violent behaviors (cf. Hale 2008). These psychological factors also make elite manipulation possible: without emotional ingredients, ethnic mobilization would not occur (Kaufman 2001; Petersen 2002).

Similarly, many material factors identified in the literature can be easily integrated into SD/SM as physical regulators of the spiral dynamics. For instance, geography, the factor Toft (2003) singled out, has always been part of the discussion of SD/SM (Jervis 1978). Similarly, conflicts over material interests (e.g., income, territory, oil or gas) can easily be brought into the security dilemma model, because conflicts of interest have always been at the core of realism’s theory of conflict (Jervis 1978; cf. Ross 2006; Toft 2003; Woodward 1995). Finally, the possibility that the presence of allies (or foreign patrons in ethnic war) can exacerbate the security dilemma has long been recognized (Snyder 1984). The challenge now is to bring all of these factors into an integrated framework.

Intergroup-Intragroup Interactions

To understand interstate conflicts, it is often useful to first depict states as unitary actors and then open up the black box by bringing domestic politics into the picture. To understand ethnic war, however, depicting ethnic groups as unitary actors is wrong even as a first cut, because intragroup interactions between elites and the masses is key, and this intragroup interaction is constantly influenced by the dynamics of intergroup interactions. As Kaufman (1994–1995, 2001:284–87) noted, a synthesis of the “two-level game” (Putnam 1988) with the “second image reversed” approach (Gourevitch 1978) is needed to understand the dynamics of ethnic war, because these approaches essentially deal with two sides of the same coin, that is, the interaction between international and domestic politics in the context of international politics or the interaction between inter- and intragroup politics in the context of ethnic war.¹³

Indeed, only through intragroup-intergroup interactions can we grasp why intergroup politics tend to favor the more strident or chauvinistic elite in intragroup politics and why intragroup politics limit the feasibility of reaching bargains in intergroup politics (Brubaker and Laitin 1998; Fearon and Laitin 2000; Horowitz 1985; Rothschild 1981). Moreover, intergroup-intragroup interactions are intimately linked with the intergroup security dilemma or spiral. Thus, the second key building block for a general theory of ethnic war is intergroup-intragroup interactions.

At the epistemological level, the intergroup-intragroup interaction approach subsumes the rational choice approach toward ethnic war, because rational calculation centered on material or symbolic interests is only one form of the more general interaction approach. Moreover, this approach is compatible with, and can indeed subsume, a social constructivist approach toward security (e.g., Arfi 1998), including “ontological security” (Mitzen 2006). As a matter of fact, elite manipulation of ontological (in)security is equivalent to elite manipulation of fear of ethnic extinction or domination. Essentially, elite manipulation of ethnic identity, hatred, and fear can be understood as a process of mobilization through construction of a hardened ethnic identity and engineering a sense of ontological insecurity.

More concretely, within each group, there are at least two actors: the elite (including the top leadership) and the masses. And more often than not, neither the elite nor the masses will be united. Assuming that the elite within one group can adopt two positions toward the other group (i.e., moderate or hostile), and that the majority of the masses are moderate, the most common scenario will have the elite and the masses in both groups divided.¹⁴ Any general theory of ethnic war will have to build on this scenario. Most critically, the prevalence of such a scenario points to two critical lessons. First, ethnic mobilization is crucial for understanding ethnic war, and the success of ethnic mobilization is conditional rather than predetermined. Second, and equally important, an approach that emphasizes structural (i.e., macro-social, economic, and political) factors is at least incomplete, if not invalid, for understanding ethnic war: ethnic wars will not occur without malignant ethnic elites as crucial agents (Petersen 2001; Zürcher 2007).

Tying Underlying Factors With Immediate Drivers

My integrated theory of ethnic war begins by linking various factors associated with ethnic war via four immediate (master) drivers. Because all four immediate drivers can be linked with ethnic war via the two meta-mechanisms (i.e., SD/SM and intergroup-intragroup interactions), by linking factors with the four master drivers and the two meta-mechanisms, I shall have integrated many, if not most, of the factors identified in the literature on ethnic war. This paves the way to link these factors with ethnic mobilization by elites. And because escalatory dynamics (also known as “escalatory spiral”) are a component of SD/SM, my framework subsumes the notion that all ethnic wars are driven by escalatory dynamics (e.g., Sambanis 2004a) and shows that all these immediate drivers and underlying factors are indeed part of escalatory dynamics.

The Four Master Drivers: Emotion, Interest, Capability, and Opportunity

The four master drivers are key to linking the numerous psychological and material factors into an integrated framework via the two meta-mechanisms described earlier. Much work underscores that these drivers are the more proximate drivers of conflictual behavior (e.g., Collins 2008, 2012; Horowitz 1985; Kaufman 2001, 2006; Lebow 2008; Mann 2005; Petersen 2002; Wimmer 2013), but here I bring them all together as the four master drivers.

Under the heading of emotion are four specific emotions (or emotional drivers): fear, honor, hatred, and anger (rage). Fear, especially fear of ethnic domination and annihilation, has been singled out by Horowitz (1985), Posen (1993), and Kaufman (2001), sometimes as part of SD/SM. In ethnic politics, fear is often triggered by a chauvinistic majority group, through either rhetoric (e.g., announcing a plan to dominate the minority) or actual action (e.g., implementing legal measures of domination). In Georgia, the nationalistic president Zviad Gamsakhurdia revoked the cultural and political autonomy of the minority Abkhaz immediately after claiming independence from the former Soviet Union. The same process happened in Croatia: Tuđman began to “Croatialize” the country immediately after the collapse of Yugoslavia. In both Georgia and Croatia, the outbreak of violence was initially driven by the majority group’s rhetoric and measures to impose or strengthen its domination (Brubaker 1996; Kaufman 1996b, 2001); minority groups then had to resist.

Likewise, honor, which links with prestige, grievance, resentment, and domination, has been singled out by Gurr (1968, 1970), Rothschild (1981), Horowitz (1985), Petersen (2002), Mann (2005), Lebow (2008), Cederman, Wimmer, and Min (2010), and Wimmer (2002, 2013). Both dominant and subordinate groups can harbor grievance or resentment: the former abhors the possible loss of their advantages, and the latter wants to (re)gain equality, honor, and glory (Rothschild 1981).

Hatred, whether ancient or modern, has been singled out by Kaplan (1994), Kaufman (2001), and Petersen (2002). Spurred by previous episodes of conflict, hatred should be one of the most potent predictors of conflict down the road (Cederman et al. 2010).

Relatively speaking, anger and rage or fury (i.e., strongly aroused and highly charged anger) have been underappreciated as key drivers of conflictual behavior (for exceptions, see Collins 2008, 2012; Petersen 2002; see also Brass 1997; Horowitz 2001; Kaufman 2001). Yet anger, especially rage (which is hard to predict), may be the most immediate driver of localized conflict. Directly, rage may induce spontaneous or mass-led violence. Indirectly, rage may be one of the most important factors facilitating ethnic mobilization.

Interest (or goals) has long been recognized as a key driver of human behavior, including conflictual behavior. For ethnic groups, tangible interest in ethnic war has been most prominently associated with territory (separation or autonomy), domination over or total annihilation of another group, natural resources such as oil, gas, and diamonds (Ross 2006), business opportunities (Horowitz 1985), and access to the state apparatus in terms of jobs and promotions (Cederman et al. 2010; Horowitz 1985). For a state (vs. an ethnic group within its boundaries), the objectives can vary from (re)imposing political domination, to repulsion, to outright genocide. Agents’ different objectives affect ethnic war differently.

Security (or freedom from insecurity), honor, and power (capability) can be understood as forms of symbolic interests. Here, I explicitly differentiate tangible interests from symbolic ones, and I restrict interests to tangible interests.¹⁵ Although power or capability can be understood as a kind of interest, I differentiate capabilities and power that are directly linked with fighting power from other tangible interests (e.g., access to bureaucratic positions or central purses). As such, interest here means tangible interest but not power or capability per se.

Opportunity or feasibility has been most prominently singled out by Fearon and Laitin (2003), Collier and Hoeffler (1998, 2004), and Collier et al. (2009). Factors that affect opportunity include weakening of the central state (e.g., collapse of central authority during regime transition, democratic or not), economic crises, and foreign invasion (Petersen 2001, 2002).

The key role of capability, especially military capability, is easy to understand: without some military capability, a state cannot terrorize its ethnic minorities, and an ethnic minority cannot rebel. Thus, the many ethnic wars in Central and Eastern Europe that occurred as the Soviet Union unraveled were facilitated by the fact that the crumbling nation-state left behind abundant weaponries (Zürcher 2007). Similarly, the first Sudanese ethnic war (1955–1972) between the northerners and southerners was underpinned by the fact that the southerners had a standing army (i.e., the Equatorial Corps) in place. Other factors that affect war-fighting capabilities include terrain, supporting diasporas, lootable resources within a territory held by a minority, and support by ethnic kin in a neighboring country.

Opportunity and capability interact with each other, and some of their underpinning factors may overlap (e.g., invasion by a foreign country), but they are not the same. Opportunity can be imposed or created, but capabilities can only be accumulated. Moreover, factors that underpin capabilities may be more important for understanding the duration than the onset of ethnic war. Collapsing opportunity and capability is thus misleading (e.g., Collier and Hoeffler 2004; Fearon and Laitin 2003).

Recognizing these master (immediate) drivers of conflictual behavior is a crucial first step. Yet the more critical task is to link the numerous factors identified in the literature on ethnic war with the four master drivers, and hence, ethnic war.

Factors Operating Independently

Numerous factors have been singled out as possibly contributing to ethnic war. In one exercise alone, Hegre and Sambanis (2006) identified 88 variables that have been postulated to contribute to the onset of civil war. Yet most of these factors have not been properly linked with the four master drivers of ethnic war. In this section I examine most of the factors identified in the literature and then sort out their possible links with ethnopolitics and ethnic war. I perform two main tasks here.

First, because different authors label the same factor differently, I reduce redundancy among labels. For instance, immigration of ethnic aliens to the core territory of a minority group, which almost inevitably induces fear and resentment within the minority group, has been labeled as “demographic invasion” (Ross 2005) or “sons of the soil” (Fearon and Laitin 2011; Weiner 1978). I group such factors under the same label. Second, I sort variables according to their potential links with the four master drivers. Table 1 displays the results.

Table 1.

From underlying factors to master drivers.

Master Drivers	Specific Underlying Factors
Emotion: fear, honor, hatred, and anger	Fear (insecurity): collapse of central authority; hijacking of the central authority by one group; withdrawal of colonial power leading to the fear of ethnic annihilation/subordination, especially with ethnic mixing and earlier episodes of violent conflict; nationalizing policies by a dominant group; influx of ethnic aliens; relative demographic decline
	Honor (resentment/grievance): ethnic solidarity; domination (subordination, exclusion, discrimination, repression); influx of ethnic aliens; economic inequality (real or perceived); redistribution policy
	Hatred: earlier episodes of violent conflict, especially ethnic war and cleansing
	Anger, especially rage/fury: local brutality; natural disasters that struck a group

Tangible interest/greed	Demand of equality and equal opportunity; demand of autonomy (cultural, economic, political); demand of secession/independence; access to state bureaucracy and military rank and file; natural resources, especially in the territory of the subordinate group; economic inequality and redistribution policy

Opportunity/feasibility (internal and external constrains)	Collapse of central authority; political instability at the center/regime transition; economic crises; a weakened state (e.g., by earlier episodes of violence, in-fighting within the ruling elite, or defeat in an interstate war); contagion/diffusion of conflict; other political contexts (e.g., regime type, civil society, or democratization/decentralization)

Capability/power	Access to weaponry; military organizations and combat experiences from earlier conflicts; absolute size of group and relative size of the group within the whole population; proportion of young men (within the subordinate group); support by external allies, either state or diasporas; distance between the rebelling region and the central government; terrain of the subordinate group; natural resources within the subordinate group; price of primary commodity goods; total gross domestic product (of the state, central government); overall state capacity weakened by earlier conflicts

This exercise yields some important payoffs. First, the redundancy among the factors identified in the literature becomes abundantly clear: the same factors have been labeled differently by various authors. Such redundancy muddles our understanding.

Second, some factors affect more than one master driver. For instance, earlier episodes of violent conflict affect fear, hatred, grievance, and interest. Similarly, immigration of ethnic aliens to the core territory of a minority group induces fear and resentment within the minority group, regardless of whether the minority group is dominant or subordinate within the region itself.

Third, some factors may be more potent than others in driving ethnic war because they affect several master drivers. Again, earlier episodes of conflict should be the most potent driver of future ethnic violence, because they not only affect fear, hatred, grievance, and interest, but they also leave behind military hardware, organizations, and combat experiences (cf. Fearon and Laitin 2003). Ethnolinguistic fractionalization (ELF), in contrast, likely contributes to the onset of ethnic war only marginally, because it is too far removed from the four master drivers. Not surprisingly, prior research does not show ELF to be significantly associated with the onset of ethnic war (for similar takes, see Wimmer 2013; Wimmer et al. 2009).

Fourth, some factors may have been unduly neglected or inappropriately operationalized. For instance, many earlier quantitative studies of civil war controlled for the terrain of a country. Yet the more appropriate indicator is the terrain of the region in which the rebelling group is concentrated; recent works have indeed borne out this hunch (e.g., Weidmann 2009; Wucherpfennig et al. 2011). Earlier quantitative studies that sought to link oil, gas, and other easily extractable mineral resources may have neglected the ethnogeography of these resources and thus obtained inconclusive results (for a more detailed discussion and robust quantitative evidences, see Tang and Li N.d.).

The Power of an Interactive Approach: An Illustration

Within any system, few factors operate independently. Rather, factors interact with each other and link with mechanisms to drive social outcomes such as ethnic war. Yet most existing studies, especially qualitative ones, tend to treat factors as if they work independently from each other (Sambanis 2004a).

Horowitz (1985:25–35) argued that whether two groups are ranked or unranked is a key variable for understanding ethnic war. He also noted that “whether ethnic politics is more parochial or more central is mainly a function of group size relative to state size” (p. 40).¹⁶ By combining the dimensions of majority group versus minority group with ranked versus unranked, four scenarios result.

Intuitively, scenarios in which the majority group dominates the minority group, or in which two groups are of roughly the same size and neither group dominates, should be fairly stable, although the latter scenario depends on a more delicate balance of power. In contrast, scenarios in which the minority group dominates the majority group, or two groups are of roughly the same size but one dominates the other, are quite unstable (Wimmer 2013).¹⁷ The former, however, should be more unstable than the latter, not only because minority domination is incompatible with the ideology of modern nationalism but also because a dominated majority group is more easily mobilized to usurp power. The odds of winning against a minority group appear very good, whereas the outcome of any potential conflict is less predictable in a scenario in which two groups are of roughly the same size.

I now turn to settlement patterns and the location of natural resources. For simplicity, I limit the discussion here to the scenarios with a clear majority-minority differentiation. Table 2 summarizes this discussion.

Table 2.

The power of an interactive approach: an illustration.

	Majority Domination/Minority in Subordination	Minority Domination/Majority in Subordination
Outcomes without other factors	Can be stable or unstable, separatist/secessionist movement possible	Highly unstableThe subordinated majority group will seek independence, if not try to take over the country outright

Discovery of oil and other substantial natural resources within the core territory of the subordinate group	Very likely to be a powder kegThe subordinate minority group might seek redistribution of resources or external supportThe dominant majority group might seek control of resources	Extremely likely to be a powder kegThe subordinate majority group might seek independence, and it can easily draw outside supportThe dominating minority group might also seek control of resources

Discovery of oil and other substantial natural resources within the core territory of the dominant group	Implications for onset: stabilizing, if not actually strengthening the status quo, because it provides more resources for the majority group to dominate and fight, if necessary	Implications for onset: extremely risky, the dominant (minority) group now has more resources for repression, whereas the subordinate (majority) group has more reason to resent and rebel
	Implications for duration: if war breaks out, war tends to be bloody but short (the majority group has more resources to fight)	Implications for duration: if war breaks out, it is likely to be bloody and long, partly because the minority group can gain external support by selling “booty futures”

Despite a voluminous literature, early studies of oil and gas and civil (ethnic and nonethnic) war have failed to produce consistent results (for reviews, see Ross 2006, 2014). A major cause behind this lack of consistency may have been the failure to see that the interaction between ethnic settlement patterns, geography, and oil/gas, rather than oil/gas alone, links oil/gas with ethnic war. With a more interactive approach, we can arrive at a new hypothesis: only oil/gas located in the core territory of a (subordinate) minority group tends to ignite or worsen ethnic war, whereas oil/gas located in territories held by a dominant majority group, or evenly dispersed ethnic groups (hence no group can claim oil/gas to be “its” oil/gas), should have little effect on ethnic war within a country. I and others have indeed advanced several hypotheses along this logic, and both quantitative and qualitative evidence strongly supports our hypotheses. By adopting a more interactive approach toward theorizing ethnic war, we are able to resolve a major controversy within the literature on oil/gas and ethnic war (for more detailed discussion and evidence, see Tang and Li N.d.).

Processes: Ethnic Mobilization And Elite-Mass Interaction

The idea that elite manipulation is one of the keys to intergroup conflict can be traced back to Simmel ([1908] 1955) and Coser (1956). After all, ethnic war would be extremely difficult, if not impossible, without some kind of organizational support, and to build this organizational support, elites and masses must work together. Not surprisingly, much work singles out ethnic mobilization by elites as a key process that drives ethnic war. Indeed, even in accounts of ethnic war with a primordialist or culturalist tone, the role of elites’ rhetoric and manipulation is all too evident, as Fearon and Laitin (2000:853–60) pointed out. Without understanding the actual process of ethnic mobilization, it is impossible to arrive at an adequate understanding of ethnic war, and policies based on favored factors will be ineffective at curbing such conflicts. Within the contentious politics literature, discussion of mobilization brings many useful insights, not least because it has already integrated many of these factors—such as social networks, community structure, group solidarity, and identity—into a coherent framework (e.g., McAdam, Tarrow, and Tilly 2001; Tarrow 2011; Tilly and Tarrow 2006), even though mobilization in ethnic politics may be quite different from mobilization in nonethnic politics (e.g., democratization, labor protests, and women’s rights movements).¹⁸

The crux of ethnic mobilization is to make group members willing to both sacrifice for the group and inflict violence and atrocities upon the other group. Ethnic mobilization mobilizes politics and violence (almost) exclusively alone ethnic lines (Mann 2005; Rothschild 1981). In successful ethnic mobilization, the more fearful, hateful, and hawkish voices win in the marketplace of ideas (Snyder and Ballentine 1996). In contrast, those who advocate and practice moderation will be treated “with the bitter contempt [if not hatred] reserved for brothers who betray a cause” (Hudson, quoted in Horowitz 1985:54; for earlier theoretical discussions in international relations, see Kahneman and Renshon 2007).

Existing discussions have identified several specific processes or tactics of ethnic mobilization, but the discussion has been far from systematic. I now systematically identify more fine-grained mechanisms of ethnic mobilization; four points should be stressed.

First, elite competition via ethnic outbidding to speak for the group is a major cause (and outcome) of ethnic war. Discussions that assume a single voice within an ethnic group are thus deeply flawed, because they have assumed away one of the key dynamics of ethnic war (e.g., Lijphart 1977), as Horowitz (1985:573–74) pointed out. Moreover, elites can engage in outbidding as part of ethnic mobilization because they are true believers of the ethnonationalist cause (e.g., Tuđman of Croatia), for instrumental reasons (e.g., Milošević of Serbia), or for both reasons. All else being equal, the more severe the competition among elites within an ethnic group, the more likely some elites will resort to ethnic outbidding.¹⁹

Second, ethnic mobilization does not have to be successful: the actual outcome depends on the interaction between mobilization and countervailing polices (or counter-mobilization) by more conciliatory elites and external actors, in the context of other constraints. There is thus nothing inevitable about ethnic war, contrary to the notions that ancient hatred drives ethnic groups inexorably to actual conflict (e.g., Kaplan 1994) or that the mere social construction of violent group identities can lead to violence (e.g., Arfi 1998:152). For an ethnic war to break out, the masses must follow, and yet the masses do not always follow: masses are not easily duped (Fearon and Laitin 2000; Horowitz 1985; Petersen 2002). In short, elite mobilization is a necessary but insufficient cause of ethnic war.

Third, the four master drivers and the underlying factors singled out earlier are also critical for ethnic mobilization: mobilization does not operate within a vacuum. Indeed, without the master drivers and factors, ethnic mobilization will be extremely difficult, if not entirely impossible.

Finally, as emphasized in the discussion of SD/SM, ethnic mobilization inevitably changes the factors and immediate drivers themselves, and these changes then loop back to affect mobilization further down the road.

The mobilization tactics elites deploy to drive ethnic war can be grouped into two broad categories: intragroup and intergroup-intragroup.

Intragroup Tactics

Ethnic outbidding by elites is the key tactic within intragroup ethnic mobilization. More specific tactics of ethnic outbidding include the following²⁰:

Deploy mass media to paint conciliatory policies, especially policies proposed by opposing elites within one’s own group, as dangerous, if not traitorous.

Deploy mass media to paint the challenges against oneself, posed by opposing moderate elites within one’s own group, as challenges and threats against the whole group.

Deploy mass media to paint opposing moderate elites within one’s own group as softies, cowards, and even traitors who sacrifice the group for their personal gains.

Deploy violent means (including assassination) to eliminate opposing elites, whether moderate or hawkish, and then put the blame on the other group.

Intergroup-intragroup Tactics

In ethnic mobilization, intergroup-intragroup tactics often go hand in hand with ethnic outbidding within a group. Specific tactics of intergroup-intragroup manipulation include the following:

Deploy mass media to fan (i.e., construct) ethnic fear or insecurity (i.e., our group is in danger of being enslaved or annihilated, our culture is in danger) by painting the other group as inherently hostile and imminently threatening.

Deploy mass media to fan ethnic resentment or grievance by mythologizing the other group as discriminating, imposing, and undeservedly overachieving.

Deploy mass media to fan ethnic resentment or grievance by attributing one’s economic and other woes to the other group.

Deploy mass media to fan ethnic hatred by (a) mythologizing the other group’s real or fabricated crimes against one’s own group as part of the collective memory (e.g., the other group took our land, plundered our wealth, killed our fathers and brothers, and raped our mothers and sisters), (b) painting one’s own group as innocent victims while demonizing the other group as unforgivable perpetrators, and (c) painting one’s own group as heroic, powerful, and glorious and the other group as treacherous, weak, and inglorious.

Deploy mass media to arouse ethnic rage or anger by dramatizing and exaggerating the other side’s hostile behaviors as real or imagined atrocities even if some of the other side’s hostile behaviors are retaliatory measures against one’s own provocations and attacks.

Deploy thugs to inflict brutal violence against the other group, and thus provoke retaliations, and then use the retaliation as proof that the other group is hostile and evil (Gagnon 1994–1995; Mann 2005; Mueller 2000). This tactic achieves two objectives with one stroke and is thus the most devilish tactic within intergroup-intragroup ethnic mobilization: it not only inoculates hatred and rage among ethnic kin, but it can coerce formerly moderate ethnic kin to participate in ethnic violence and to seek protection within one’s own group because of fear of being targeted by the other group. Once this is achieved, masses on both sides are “rationally” compelled to support mass violence and ethnic war (De Figueiredo and Weingast 1999; Fearon and Laitin 2000).

Concluding Remarks

In this article, I advance a more integrated theory of ethnic war. The new theory brings together numerous underlying factors, immediate master drivers, and mechanisms scattered in the existing literature into a more coherent and dynamic synthesis. The theory thus holds important implications for future efforts in understanding ethnic war.

Foremost, a key lesson from this exercise is that factors and mechanisms interact with each other to drive social outcomes, including ethnic war. As such, banking on a single variable or mechanism, or pitting different variables and mechanisms against each other, is unlikely to produce much insight. An interactive and integrative approach provides a more fruitful path toward a better understanding of ethnic war.

Second, because ethnic mobilization is the key process leading to ethnic war, more attention should be paid to it, both in terms of policy and academically. Only with a deeper understanding of ethnic mobilization can we better understand ethnic war and design better policies to counter its potentially vicious and explosive fallout.

In terms of policy, because ethnic mobilization is the key process leading to ethnic war, a key measure for preventing ethnic war is to prevent radical ethnic elites—especially elites who already have substantial power bases (e.g., Dzhokhar Dudayev, Milošević, and Tuđman)—from successfully mobilizing the masses for violence. On this front, moderate ethnic elites and moderate masses must stay vigilant and ready to counter-mobilize (Tang 2011b; Zürcher 2007). Equally important, moderate elites in different groups need to work with one another and do this smartly (Tang 2011a; for evidence, see Zürcher 2007).

Academically, rather than merely focusing on initial conditions and constraining factors, as most quantitative exercises have done, we need to peer into the actual mobilization process to better understand ethnic war. In particular, quantitative studies without in-depth process tracing can carry us only so far: regressions simply do not allow us to visualize and thus differentiate different mechanisms, especially those of ethnic mobilization (cf. Petersen 2001; Zürcher 2007). Theories and hypotheses are better supported by a combination of quantitative and qualitative evidence (Collier and Sambanis 2005; Ross 2004; Sambanis 2004a).

Third, sophisticated theorizing should precede quantitative and qualitative empirical inquiries: without theoretical guidance, empirical studies produce mere correlations, evidence, or conclusions that are difficult to substantiate and interpret.

Fourth, we need to better grasp the escalatory dynamics leading to ethnic war, because years of low-level ethnopolitics tend to precede and stir up the eventual violent conflict (Sambanis 2004a). As such, data sets on the onset of ethnic war that do not capture escalatory dynamics are of limited value. We need to construct data sets that at least partially capture the (de)escalatory dynamics of ethnic politics.²¹

Finally, despite our effort and much progress, we still have a long way to go to form an adequate understanding of ethnic war. I thus end this discussion with a cautionary note: because of the enormous complexities of ethnic war, some overly confident policy prescriptions should be treated with extreme caution (e.g., Collier 2009; for a critique, see Keen 2012).

Footnotes

Acknowledgements

For critical discussions and comments, I thank Hui Li, Roger Petersen, Nils Weidmann, and two anonymous reviewers. Kai Wang, Shan Yang, and Min Zhang provided excellent research assistance.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research is supported by a 985 Project third-phase bulk grant (2011–2013) and an Academic Excellence bulk grant (2014–2016) from Fudan University to the author.

Notes

Author Biography

Shiping Tang is Fudan Distinguished Professor and Dr. Seaker Chan Chair Professor at the School of International Relations and Public Affairs (SIRPA), Fudan University, Shanghai, China. He has very broad research interests. His most recent book, The Social Evolution of International Politics (2013), received the International Studies Association (ISA) “Annual Best Book Award” in 2015. He is also the author of A General Theory of Institutional Change (2011), A Theory of Security Strategy for Our Time: Defensive Realism (2010), and many articles in journals in international relations, comparative politics, and philosophy of the social sciences.

References

Arfi

Badredine

. 1998. “Ethnic Fear: The Social Construction of Insecurity.” Security Studies 8(1):151–203.

Bessinger

Mark R.

2002. Nationalist Mobilization and the Collapse of the Soviet State. Cambridge, UK: Cambridge University Press.

Blattman

Christopher

Miguel

Edward

. 2010. “Civil War.” Journal of Economic Literature 48(1):3–57.

Brass

Paul R.

1991. Ethnicity and Nationalism: Theory and Comparison. Beverly Hills, CA: Sage.

Brass

Paul R.

1997. Theft of an Idol: Text and Context in the Representation of Collective Violence. Princeton, NJ: Princeton University Press.

Brubaker

Rogers

. 1996. Nationalism Reframed: Nationalhood and the National Question in the New Europe. Cambridge, UK: Cambridge University Press.

Brubaker

Rogers

Laitin

David D.

1998. “Ethnic and Nationalist Violence.” Annual Review of Sociology 24:423–52.

Bunge

Mario

. 1997. “Explanation and Mechanisms.” Philosophy of the Social Sciences 27(4):410–65.

Butterfield

Herbert

. 1951. History and Human Relations. London: Collins.

10.

Cederman

Lars-Erik

Wimmer

Andreas

Min

Brian

. 2010. “Why Do Ethnic Groups Rebel? New Data and Analysis.” World Politics 62(1):87–119.

11.

Chandra

Kanchan

. 2005. “Ethnic Parties and Democratic Stability.” Perspectives on Politics 3(2):235–52.

12.

Chandra

Kanchan

. 2006. “What Is Ethnic Identity and Does It Matter?” Annual Review of Political Science 9:397–424.

13.

Chandra

Kanchan

Wilkinson

Steven

. 2008. “Measuring the Effect of ‘Ethnicity.’” Comparative Political Studies 41(4–5):515–63.

14.

Collier

Paul

. 2009. Wars, Guns and Votes: Democracy in Dangerous Places. New York: HarperCollins.

15.

Collier

Paul

Hoeffler

Anke

. 1998. “On Economic Causes of Civil War.” Oxford Economic Papers 50(4):563–73.

16.

Collier

Paul

Hoeffler

Anke

. 2004. “Greed and Grievance in Civil War.” Oxford Economic Papers 56(4):563–95.

17.

Collier

Paul

Hoeffler

Anke

Rohner

Dominic

. 2009. “Beyond Greed and Grievance: Feasibility and Civil War.” Oxford Economic Papers 61(1):1–27.

18.

Collier

Paul

Sambanis

Nicholas

, eds. 2005. Understanding Civil Wars: Evidence and Analysis. Washington, DC: World Bank.

19.

Collins

Randall

. 2008. Violence: A Micro-sociological Theory. Princeton, NJ: Princeton University Press.

20.

Collins

Randall

. 2012. “C-escalation and D-escalation: A Theory of the Time-dynamics of Conflict.” American Sociological Review 77(1):1–20.

21.

Connor

Walker.

[1984] 1994. “Eco- or Ethno-nationalism?” Pp. 144–64 in Ethnonationalism: The Quest for Understanding, by Walker

Connor

. Princeton, NJ: Princeton University Press.

22.

Coser

Lewis A.

1956. The Functions of Social Conflict: Political Opportunism and Ethnic Conflict. Glencoe, IL: Free Press.

23.

Cramer

Christopher

. 2002. “Homo Economicus Goes to War: Methodological Individualism, Rational Choice and the Political Economy of War.” World Development 30(11):1845–64.

24.

De Figueiredo

Rui J. P.

Jr Weingast

Barry R.

1999. “Rationality of Fear: Political Opportunism and Ethnic War.” Pp. 261–302 in Civil Wars, Insecurity, and Intervention, edited by Walter

B. F.

Snyder

New York: Columbia University Press.

25.

Eifert

Benn

Miguel

Edward

Posner

Daniel

. 2011. “Political Competition and Ethnic Identification in Africa.” American Journal of Political Science 54(2):494–510.

26.

Fearon

James D.

1995. “Rationalist Explanations for War.” International Organization 49(2):379–414.

27.

Fearon

James D.

Laitin

David D.

1996. “Explaining Interethnic Cooperation.” American Political Science Review 90(4):715–35.

28.

Fearon

James D.

Laitin

David D.

2000. “Violence and the Social Construction of Ethnic Identity.” International Organization 54(4):845–77.

29.

Fearon

James D.

Laitin

David D.

2003. “Ethnicity, Insurgency, and Civil War.” American Political Science Review 97(1):75–86.

30.

Fearon

James D.

Laitin

David D.

2011. “Sons of the Soil, Migrants, and Civil War.” World Development 39(2):199–211.

31.

Frjida

Nico H.

1987. Emotions. Cambridge, UK: Cambridge University Press.

32.

Gagnon

V. P.

Jr.

1994–1995. “Ethnic Nationalism and International Conflict: The Case of Serbia.” International Security 19(3):130–66.

33.

Gellner

Ernest

. 1983. Nations and Nationalism. Oxford, UK: Blackwell.

34.

Goertz

Gary

Mahoney

James

. 2012. A Tale of Two Cultures: Qualitative and Quantitative Research in the Social Sciences. Princeton, NJ: Princeton University Press.

35.

Gourevitch

Peter

. 1978. “The Second Image Reversed: The International Sources of Domestic Politics.” International Organization 32(4):881–912.

36.

Grigoryan

Arman

. 2007. “Correspondence: Hate Narratives and Ethnic Conflict.” International Security 31(4):180–91.

37.

Gurr

Ted Robert

. 1968. “A Causal Model of Civil Strife: A Comparative Analysis Using New Indices.” American Political Science Review 62(4):1104–24.

38.

Gurr

Ted Robert

. 1970. Why Men Rebel. Princeton, NJ: Princeton University Press.

39.

Gurr

Ted

Harff

Barbara

. 1994. Ethnic Conflict in World Politics. Boulder, CO: Westview.

40.

Hale

Henry E.

2008. The Foundation of Ethnic Politics: Separatism of States and Nations in Eurasia and the World. Cambridge, UK: Cambridge University Press.

41.

Hall

John A.

(Ed.) 1998. The State of the Nation: Gellner and Theory of Nationalism. Cambridge, UK: Cambridge University Press.

42.

Hegre

Havard

Sambanis

Nicholas

. 2006. “Sensitivity Analysis of Empirical Results on Civil War Onset.” Journal of Conflict Resolution 50(4):508–35.

43.

Herz

John

. 1951. Political Realism and Political Idealism. Chicago: University of Chicago Press.

44.

Horowitz

Donald L.

1985. Ethnic Groups in Conflict. Berkeley: University of California Press.

45.

Horowitz

Donald L.

2001. Deadly Ethnic Riots. Berkeley: University of California Press.

46.

Horowitz

Shale

. 2008. “Mapping Pathways of Ethnic Conflict Onset: Preferences and Enabling Conditions.” Ethnopolitics 7(2–3):307–20.

47.

Jervis

Robert

. 1976. Perception and Misperception in International Politics. Princeton, NJ: Princeton University Press.

48.

Jervis

Robert

. 1978. “Cooperation under the Security Dilemma.” World Politics 30(2):167–214.

49.

Kahneman

Daniel

Renshon

Jonathan

. 2007. “Why Hawks Win.” Foreign Policy 158(1):34–38.

50.

Kalyvas

Stathis N.

2006. The Logic of Violence in Civil War. Cambridge, UK: Cambridge University Press.

51.

Kalyvas

Stathis N.

2008. “Promises and Pitfalls of an Emerging Research Program: The Microdynamics of Civil War.” Pp. 397–421 in Order, Conflict, Violence, edited by Kalyvas

S. N.

Shapiro

Masoud

Cambridge, UK: Cambridge University Press.

52.

Kaplan

Robert D.

1994. Balkan Ghosts: A Journey through History. New York: Random House.

53.

Kaufmann

Chaim D.

2005. “Rational Choice and Progress in the Study of Ethnic Conflict: A Review Essay.” Security Studies 14(1):178–207.

54.

Kaufman

Stuart J.

1994–1995. “The Irresistible Force and the Imperceptible Object: The Yugoslav Breakup and Western Policy.” Security Studies 4(2):281–319.

55.

Kaufman

Stuart J.

1996a. “An ‘International’ Theory of Inter-ethnic War.” Review of International Studies 22(2):149–71.

56.

Kaufman

Stuart J.

1996b. “Spiraling to Ethnic War.” International Security 21(2):108–38.

57.

Kaufman

Stuart J.

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

58.

Kaufman

Stuart J.

2006. “Symbolic Politics or Rational Choice? Testing Theories of Extreme Ethnic Violence.” International Security 30(1):45–86.

59.

Kaufman

Stuart J.

2009. “Narratives and Symbols in Violent Mobilization: The Palestinian-Israeli Case.” Security Studies 18(3):400–34.

60.

Kaufman

Stuart J.

2011. “Symbols, Frames, and Violence: Studying Ethnic War in the Philippines.” International Studies Quarterly 55(4):937–58.

61.

Keen

David

. 2012. “Greed and Grievance in Civil War.” International Affairs 88(4):757–77.

62.

Lake

David A.

Rothchild

Donald

. 1996. “Containing Fear: The Origins and Management of Ethnic Conflict.” International Security 21(2):41–75.

63.

Lebow

Richard Ned

. 2008. Cultural Theory of International Relations. Cambridge, UK: Cambridge University Press.

64.

Lijphart

Arent

. 1977. Democracy in Pluralist Societies. New Haven, CT: Yale University Press.

65.

Mahoney

James

. 2010. “After KKV: The New Methodology of Qualitative Research.” World Politics 62(1):120–47.

66.

Mann

Michael

. 2005. The Dark Side of Democracy: Explaining Ethnic Cleansing. Cambridge, UK: Cambridge University Press.

67.

McAdam

Doug

Tarrow

Sidney

Tilly

Charles

. 2001. Dynamics of Contention. Cambridge, UK: Cambridge University Press.

68.

McKay

James

. 1982. “An Exploratory Synthesis of Primordial and Mobilizationist Approaches to Ethnic Phenomena.” Ethnic and Racial Studies 5(4):395–420.

69.

Mercer

Jonathan

. 2005. “Rationality and Psychology in International Politics.” International Organizations 59(1):77–106.

70.

Mercer

Jonathan

. 2010. “Emotional Beliefs.” International Organizations 64(1):1–31.

71.

Mitzen

Jennifer

. 2006. “Ontological Security in World Politics: State Identity and the Security Dilemma.” European Journal of International Relations 12(3):341–70.

72.

Mueller

John

. 2000. “The Banality of ‘Ethnic War.’” International Security 25(1):42–70.

73.

Mylonas

Harris

. 2013. The Politics of Nation Building. Cambridge, UK: Cambridge University Press.

74.

Petersen

Roger D.

2001. Resistance and Rebellion: Lessons from Eastern Europe. Cambridge, UK: Cambridge University Press.

75.

Petersen

Roger D.

2002. Understanding Ethnic War: Fear, Hatred, Resentment in Twentieth-century Eastern Europe. Cambridge, UK: Cambridge University Press.

76.

Posen

Barry

. 1993. “The Security Dilemma and Ethnic War.” Survival 35(1):27–47.

77.

Putnam

Robert D.

1988. “Diplomacy and Domestic Politics: The Logic of Two-level Games.” International Organization 42(3):427–60

78.

Ragin

Charles

. 2000. Fuzzy Set Social Sciences. Chicago: University of Chicago Press.

79.

Ross

Michael L.

2004. “How Do Natural Resources Influence Civil War? Evidence from 13 Cases.” International Organization 58(1):35–67

80.

Ross

Michael L.

2005. “Resources and Rebellion in Aceh, Indonesia.” Pp. 35–58 in Understanding Civil Wars: Evidence and Analysis, Vol. 2 (Europe, Central Asia, and Other Regions), edited by Collier

Sambanis

Washington, DC: World Bank.

81.

Ross

Michael L.

2006. “A Closer Look at Oil, Diamonds, and Civil War.” Annual Review of Political Science 9:265–300.

82.

Ross

Michael L.

2014. “What Have We Learned about the Resource Curse?” Unpublished manuscript, Department of Political Sciences, University of California, Los Angeles.

83.

Rothschild

Joseph

. 1981. Ethnopolitics: A Conceptual Framework. New York: Columbia University Press.

84.

Sambanis

Nicholas

. 2001. “Do Ethnic and Nonethnic Civil Wars Have the Same Causes? A Theoretical and Empirical Inquiry (Part 1).” Journal of Conflict Resolution 45(3):259–82.

85.

Sambanis

Nicholas

. 2004a. “Using Case Studies to Expand Economic Models of Civil War.” Perspectives on Politics 2(2):259–79.

86.

Sambanis

Nicholas

. 2004b. “What Is Civil War? Conceptual and Empirical Complexities of an Operational Definition.” Journal of Conflict Resolution 48(4):814–58.

87.

Scheff

Thomas J.

1994. Emotions, Nationalism, and War. Boulder, CO: Westview.

88.

Scott

George M.

1990. “A Resynthesis of the Primordial and Circumstantial Approaches to Ethnic Group Solidarity.” Ethnic and Racial Studies 13(2):147–71.

89.

Simmel

Georg.

[1908] 1955. Conflict and the Web of Group Affiliations. Translated and edited by Bendix

Wolff

K. H.

Glencoe, IL: Free Press.

90.

Smith

Anthony

. 1998. Nationalism and Modernism. London: Routledge.

91.

Snyder

Glenn H.

1984. “The Security Dilemma in Alliance Politics.” World Politics 36(4):461–95.

92.

Snyder

Jack

Ballentine

Karen

. 1996. “Nationalism and the Marketplace of Ideas.” International Security 21(2):5–40.

93.

Snyder

Jack

Jervis

Robert

. 1999. “Civil War and the Security Dilemma.” Pp. 15–37 in Civil Wars, Insecurity, and Intervention, edited by Walter

B. F.

Snyder

New York: Columbia University Press.

94.

Tang

Shiping

. 2009. “The Security Dilemma: A Conceptual Analysis.” Security Studies 15(3):587–623.

95.

Tang

Shiping

. 2010. A Theory of Security Strategy for Our Time: Defensive Realism. New York: Palgrave Macmillan.

96.

Tang

Shiping

. 2011a. “Review Article: Reconciliation and the Remaking of Anarchy.” World Politics 63(4):713–51.

97.

Tang

Shiping

. 2011b. “The Security Dilemma and Ethnic Conflict: Toward a Dynamic and Integrative Theory of Ethnic Conflict.” Review of International Studies 37(3):511–36.

98.

Tang

Shiping

. 2014. “International System, Not International Structure: Against the Agent-structure Problématique in IR.” Chinese Journal of International Politics 7(4):483–506.

99.

Tang

Shiping

. N.d.-a. “Factors and Mechanisms in Social Sciences: From Ontology to Epistemology.” Manuscript in progress.

100.

Tang

Shiping

. N.d.-b. “Uncovering Mechanisms: Toward a New Methodology.” Manuscript in progress.

101.

Tang

Shiping

Hui

. N.d. “Location, Location, Location: The Ethno-geography of Oil and the Onset of Ethnic War.” Unpublished manuscript.

102.

Tarrow

Sidney G.

2007. “Inside Insurgencies: Politics and Violence in an Age of Civil War.” Perspectives on Politics 5(3):587–600.

103.

Tarrow

Sidney G.

2011. Power in Movement: Social Movements and Contentious Politics. 3rd ed. Cambridge, UK: Cambridge University Press.

104.

Tilly

Charles

Tarrow

Sidney

. 2006. Contentious Politics. Oxford, UK: Oxford University Press.

105.

Toft

Monica Duffy

. 2003. The Geography of Ethnic Violence. Princeton, NJ: Princeton University Press.

106.

Van Houten

Pieter

. 2008. “Comment for Horowitz: Responses.” Ethnopolitics 7(2/3):321–36.

107.

Walter

Barbara F.

2001. Committing to Peace: The Success Settlement of Civil Wars. Princeton, NJ: Princeton University Press.

108.

Weidmann

Nils B.

2009. “Geography as Motivation and Opportunity: Group Concentration and Ethnic Conflict.” Journal of Conflict Resolution 53(4):526–43.

109.

Weiner

Myron

. 1978. Sons of the Soil: Migration and Ethnic War in India. Princeton, NJ: Princeton University Press.

110.

Wilkinson

Steven I.

2005. Votes and Violence: Electoral Competition and Ethnic Riots in India. Cambridge, UK: Cambridge University Press.

111.

Williams

Robin M.

Jr.

1994. “The Sociology of Ethnic Conflict: Comparative International Perspectives.” Annual Review of Sociology 20:49–79.

112.

Wimmer

Andreas

. 2002. Nationalist Exclusion and Ethnic Conflict: Shadows of Modernity. Cambridge, UK: Cambridge University Press.

113.

Wimmer

Andreas

. 2008. “The Making and Unmaking of Ethnic Boundaries: A Multi-level Process Theory.” American Journal of Sociology 113(4):970–1022.

114.

Wimmer

Andreas

. 2013. Waves of War: Nationalism, State Formation, and Ethnic Exclusion in the Modern World. Cambridge, UK: Cambridge University Press.

115.

Wimmer

Andreas

Cederman

Lars-Erik

Min

Brian

. 2009. “Ethnic Politics and Armed Conflict: A Configurational Analysis of a New Global Dataset.” American Sociological Review 74(2):316–37.

116.

Woodward

Susan

. 1995. Balkan Tragedy. Washington, DC: Brookings Institution.

117.

Wucherpfennig

Julian

Weidmann

Nils B.

Girardin

Luc

Cederman

Lars-Erik

Wimmer

Andrews

. 2011. “Political Relevant Ethnic Groups across Space and Time: Introducing the Geo-EPR Dataset.” Conflict Management and Peace Science 28(5):423–37.

118.

Zürcher

Christoph

. 2007. The Post-Soviet Wars: Rebellion, Ethnic War, and Nationhood in the Caucasus. New York: New York University Press.