Abstract

Destabilized multiethnic states and empires are environments that are highly susceptible to violent ethnonationalist conflict. Conflicts between states built on the ruins of such empires and their minorities are especially common. James Fearon has famously argued that these conflicts are the result of minorities' rational incentives to rebel, which in turn are the result of newly independent states' inability to guarantee that these minorities will not be discriminated against if they acquiesce to citizenship, as well as expectations that over time the balance of power will shift against minorities as states consolidate their institutions. States can, however, take steps to reassure their minorities. The puzzle is why they often fail to do so. In fact, states often adopt policies that confirm minorities' worst fears, pushing them toward rebellion. Such action may be precipitated by a state's belief that a minority is motivated by a separatist agenda rather than by the desire to have its concerns and grievances satisfactorily addressed. If secession is a minority's primary objective, then concessions intended to demobilize the minority will only make the state more vulnerable to future demands and separatist bids. The existence of third parties with incentives to support minority separatism exacerbates the problem. The violent and nonviolent minority disputes in post-Soviet Georgia illustrate these findings.

Introduction

Disintegrating multiethnic states and empires are environments that are highly prone to violent ethnonationalist conflict. Such conflict accompanied the weakening and subsequent demise of the Ottoman and Austro-Hungarian Empires in the nineteenth and early twentieth centuries, the collapse of the Russian Empire after 1917, the end of European colonialism in Africa and Asia after World War II, and the disintegrations of the Soviet Union and Yugoslavia in the early 1990s. A particularly common type of violent post-imperial conflict involves newly independent states and minorities trapped within their borders. All violent conflict is puzzling to the extent that belligerents have incentives to reach peaceful bargains.1 Post-imperial conflicts are especially puzzling, however, because the belligerents invariably face a host of urgent problems stemming from the economic and institutional collapse that inevitably accompanies the disintegration of empires. Yet, in many cases these problems are put aside, or are made far more difficult to solve, as newly independent states and minorities engage in violence.

In a highly influential essay on ethnonationalist conflict, James Fearon argued that such conflicts result from the inability of newly independent states to make credible commitments that they will not discriminate against their minorities if the latter acquiesce to citizenship. According to Fearon, that inability, combined with the expectation that over time the distribution of power will shift in the state's favor as it consolidates its institutions of governance and instruments of repression, confronts the minority with a terrible choice: either launch a war of secession before that shift takes place or risk discrimination—or worse—later. If the minority believes that it has a decent chance of success, then launching a war of secession becomes the optimal choice.2

Several strands of evidence support Fearon's argument. First, there are statements Fearon himself cites that constitute direct evidence of the commitment logic.3 Second, there is the evidence of certain minorities in the Soviet Union prioritizing security concerns over their “separatist” aspirations in voting overwhelmingly to preserve the Soviet Union in a 1990 referendum.4 Third is the evidence of separatists justifying their aspirations not only with appeals to the right to self-determination, but also with references to past persecution by the group in control of the state.5 Finally, there is the evidence of minorities in control of states who resisted the transition to democracy, fearing they would be persecuted if they ceded control.6

Nevertheless, three important empirical puzzles remain. First, even if a newly independent state cannot provide ironclad guarantees to its minorities that they will not be persecuted, it can reassure them by enacting measures that would make future persecution costly and thus less likely. For example, the state can grant minorities wide-ranging rights of self-governance; give them veto power over certain issues; or impose fixed quotas in legislatures, police forces, and the officer corps. The real puzzle, therefore, is why some states do not propose such measures. Worse yet, some newly independent states not only fail to reassure their minorities, but they adopt postures and policies that make violent escalation inevitable, or they initiate violence despite needing time to consolidate their institutions and instruments of repression. A second puzzle is why some newly independent states—for example, post-Soviet Russia—try to reassure their minorities, whereas others—for example, Azerbaijan and Croatia—respond harshly to the demands of mobilized minorities. The third puzzle is why the same state sometimes behaves differently toward different minorities—as did post-Soviet Georgia, which is the main empirical focus of this study.

This article aims to solve all three puzzles. Its argument rests on the premise that although some mobilized minorities are motivated by fears of future mistreatment, such fears are not the only possible motive for mobilization. Another motivation may be a minority's desire for statehood. I argue that this variation in minorities' motives produces a result similar to what Robert Jervis describes in his analysis of the deterrence and spiral models.7 Specifically, the state's optimal position is to grant concessions to a minority that is motivated by fear and grievances resulting from discrimination, because such concessions will reassure and demobilize the minority. If, however, the minority is motivated primarily by a separatist agenda, concessions will only embolden it to make more demands or to attempt secession from a position improved by those very concessions.8 As a result, the state's decision to conciliate or to confront a mobilized minority cannot be based on a straightforward calculation of the costs of concessions compared to the benefits of preventing the minority from seceding. Rather, it is a choice between two competing risks: confronting a minority motivated by fears and grievances, which will increase the probability of a violent rebellion, or making concessions to a separatist-driven minority, which will make the state vulnerable to more demands and more vigorous separatist challenges in the future. Further complicating the state's choice is the fact that the minority's preferences are subject to uncertainty: the state cannot be sure of the minority's preferences, and even if it could, those preferences could shift over time in response to a variety of exogenous factors.

Uncertainties regarding actor's preferences need not make cooperative outcomes impossible. The literature on costly signals and commitments argues that “good” actors can often credibly reveal their preferences either by taking costly actions that only they, and not “bad” actors, would have incentives to take,9 or by attaching costs to noncooperation. Such measures are not available to minorities, however, because it is unclear how a minority could guarantee that it will not pursue a separatist agenda in the future.10 Moreover, to extract concessions from the state, even a mobilized minority seeking redress of its grievances must resort to threats of secession and show a willingness to use violence if the state resists its demands. Such actions will make it even more difficult for the state to distinguish between security-seeking and independence-seeking minorities.11 The state cannot automatically assume, however, that it is facing a separatist minority either. Therefore it runs the risk not only of making concessions to separatist minorities, but also of unnecessarily adopting coercive strategies against minorities that would demobilize if they received concessions.

The question, then, is what makes state coercion the perferable option. I argue that it is the degree of the state's vulnerability to a successful secessionist challenge following its offer of concessions to the minority. That vulnerability is a function of several factors, including whether a minority is compactly settled or dispersed,12 whether it is urban or rural,13 and whether it resides in an area conducive to guerrilla warfare.14 The effects of these factors are well understood, and I have little to add to what other scholars have already said. Instead, I focus on the distribution of power between the state and the minority and the existence of a third part that is likely to support the minority in the event it decides to rebel. Specifically, I demonstrate that the risks associated with concessions are higher the weaker the state is and the more likely the minority is to receive support from a third party if it decides to seek independence.

The rest of the article is organized as follows. First, I explain the details of my argument with the help of a simple game-theoretic model. Second, relying on Mill's method of difference, I conduct a preliminary test of the argument to explain why post-Soviet Georgia went to war with its Abkhaz and Ossetian minorities, but not with its Azeri and Armenian minorities.15 Third, I discuss the plausibility of my argument in the context of other prominent state-minority disputes in the postcommunist space that did or did not turn violent. Fourth, I evaluate other theories that scholars have applied to explain this same set of conflicts. I conclude with a summary of my argument and its implications for the debate on the causes of violent ethnonationalist conflicts.

A Model of Bargaining between New States and Their Minorities

The most common category of disputes between newly independent states and their minorities involves the latter's political status—that is, the degree of state control over the territory where the minorities reside. Some scholars describe these as disputes over an indivisible object, because the territory in question represents more than mere real estate. “A homeland,” Monica Duffy Toft writes, “is … a special category: it is not an object to be exchanged but an indivisible attribute of group identity.”16 Indeed, the very definition of nationalism as “a political principle, which holds that the political and the national unit should be congruent,”17 suggests that political status may be indivisible.

It is not. Total subjugation and independent statehood are two extremes of a continuum. This is, in fact, a rather trivial observation given the existence of arrangements that accord minorities varying degrees of self-rule. Newly independent states, in particular, must have incentives to be generous because they need time to consolidate their institutions.

This argument is valid, however, only if the analytical horizon is limited to bargaining at a given point in time. But because states and minorities will continue to interact after their current dispute is settled and because the terms of settlement will affect their future interactions, the argument about states' incentives for being generous may no longer hold. Even if concessions succeed in preventing escalation in the short run, they improve the minority's capacity for collective action,18 making the minority more dangerous as a potential secessionist in the long run. Consequently, the state must incorporate into its decisionmaking calculus the prospect of confronting a more powerful minority in the future. This threat, combined with uncertainty regarding the minority's preferences and its inability to commit credibly not to pursue secession in the future, explains why newly independent states and minorities often fail to reach negotiated bargains. The presence of a third party with incentives to collude with the minority only exacerbates this problem.

Nevertheless, it is not irrational for a state to make concessions, because there is some probability that the minority does not want to secede and therefore would demobilize if the state were willing to address its fears and grievances. The real issue concerns risk, which is subject to variation. The question is what causes this variation.

To answer this question, I develop a formal model, which is a finite-horizon, incomplete information bargaining game. The game has three actors—the state (S), the minority (M), and a third party (TP). For expositional simplicity, however, I initially omit the third party from the analysis. Incomplete information concerns the minority's type, defined as the minority's motive for mobilizing. Specifically, the minority can mobilize either because it wants to secede or because it wants the state to address its fears and grievances. The first type, which I label “unappeasable,” will be made more aggressive by the state's concessions. In contrast, the second type, which I label “appeasable,” will demobilize in response to the same set of concessions. The minority knows its own type, whereas the state has only a probability estimate about it.

The game begins with the minority's decision to mobilize or to acquiesce to citizenship in the new state. If the minority acquiesces, the game ends with the outcome SQ, which stands for status quo and which is associated with the payoffs bM and bS for the minority and the state, respectively, where bM is the minority's share of the total benefit B, and bS is the state's share, such that B = bM + bS.19 If the minority chooses to mobilize, the game continues, with the state moving next and choosing between making concessions to appease the minority and cracking down. If the state cracks down, the minority moves next, choosing between launching a war of secession and backing down. If the minority backs down, the game ends with the outcome BD and with payoffs bM – ccM for the minority and bS – cSc for the state, where ccM and cSc are the minority's and the state's respective costs for the crackdown. If the minority launches a war of secession, the game ends with the outcome WAR1 with payoffs p1B – cM for the minority and (1 – p1)B – cS for the state, where p1 is the minority's probability of victory, cM is its cost for war, and cS is the state's cost for war (see figure 1).

Figure 1.

States and Minorities Bargaining in the Shadow of the Future

Figure 1.

States and Minorities Bargaining in the Shadow of the Future

If the state makes concessions instead of cracking down, the minority moves next, choosing between the options of demobilizing and launching a war of secession. If the minority demobilizes, the game ends with the outcome SQ+, which is associated with payoffs bM + θ and bS – θ for the minority and the state, respectively, where θ is the value of concessions that the state makes to the minority. If the minority launches a war of secession, the game ends with the outcome WAR2. The minority's payoff for this outcome depends on its type. If the minority is unappeasable, then the payoff is p2B – cM; if it is appeasable, then the payoff type's payoff is p2B – cM – ceM, where p2 is the minority's probability of victory, such that p2 > p1,20 and ceM is a cost in addition to the general cost for war incurred by the appeasable minority for rebelling after receiving concessions,21 such that p2B – cM > bM + θ > p2B – cM – ceM. The state's payoff for this outcome is (1 – p2)B – cS. Incomplete information about the minority's type is generated by assuming that the minority knows whether or not it is appeasable, whereas the state has only a probability estimate about it drawn by a nonstrategic actor, called Nature. It is assumed that the probability of facing an unappeasable minority is α, and that the corresponding probability of facing an appeasable minority is 1 – α.

What are the game's equilibria? Because the minority is the informed actor, it is possible to determine its best response strategy using backward induction. The minority will acquiesce if bM > p1B – cM—(i.e., if its payoff for the status quo outcome is larger than its payoff for fighting). If bM < p1B – cM, {mobilize, rebel} is the unappeasable minority's dominant strategy. The appeasable minority's best response strategies are {mobilize, demobilize} if the state plays {make concessions} and {mobilize, rebel} if the state plays {crack down}. The state's best response strategy cannot be determined using the same procedure, because the state does not know with certainty its location inside the information set following the minority's decision to mobilize. Rather, the state knows its location only probabilistically, which means that its best response strategy has to be determined by comparing the sure payoff for playing {crack down} with the expected payoff for playing {make concessions}, which is,

formula

Setting this value equal to the payoff for {crack down} and solving for α produces the threshold value of α, where the state is indifferent between the two strategies:

formula

The state's best response, then, is {make concessions} if α < α* and {crack down} if α > α*. The combination of the actors' best response strategies and the available information generates the equilibrium outcomes summarized in

figure 2, excluding the equilibrium outcome SQ when bM > p1B – cM. The equilibrium analysis confirms the intuition that it may be rational for newly independent states to drive hard bargains with their minorities; that it may also be rational to try conciliation; and that the optimality of either strategy depends on the state's expectations of the minority's future behavior.

Figure 2.

The Equilibrium Outcomes

Figure 2.

The Equilibrium Outcomes

More interesting are two comparative statics results. First, the state's willingness to make concessions is a function of the change in the distribution of power resulting from these concessions. Second, by relaxing the constraint imposed on the appeasable and unappeasable minorities, it is possible to see that the distribution of types of the minority is sensitive to the change in the distribution of power. Thus I have assumed that bM + θ > p2B – cM – ceM for certain fixed values of these parameters. These parameters are all variables, however; therefore, if the general costs of war—cM—were to decrease sufficiently, the relationship between the two payoffs may change. Suppose the threshold value of costs, which makes even the appeasable minority indifferent between the options of rebelling after it receives concessions and demobilizing, is

formula

Although it is still more costly for the appeasable minority to rebel, doing so will be optimal for any cM < c*M. In addition, c*M varies with the distribution of power, which means that the easier secession becomes, the more tempting it becomes despite the lower value that the appeasable minority attaches to secession.

What happens, then, if a third party biased in favor of the minority is added to the model? Suppose the game does not end with the WAR1 and WAR2 outcomes, but following the minority's decision to rebel, a third party enters the game. The third party's choice is to intervene or not to intervene. If the third party does not intervene, the game ends with the WAR1 and WAR2 outcomes. If it does intervene, the outcomes are INT1 and INT2 (see figure 3).

Figure 3.

States and Minorities Bargaining in the Shadow of Intervention

Figure 3.

States and Minorities Bargaining in the Shadow of Intervention

The introduction of a third party requires the modification of some prior assumptions about the minority's preferences and the addition of new ones. I discard the assumption about variation in the preferences of appeasable and unappeasable minorities and the corresponding assumption that there is incomplete information about those preferences. I assume instead that in response to concessions from the state, the minority prefers demobilization if it is not supported by a third party and a war of secession if it is. Doing so simplifies the analysis at no cost, putting the focus on how the presence of third parties affects the interaction (see figure 3).

I assume that third parties vary in type and that their preferences are subject to incomplete information. Thus some third parties are chiefly interested in protecting minorities if the latter are persecuted, whereas others see conflicts between states and minorities as opportunities to take advantage of the state.22 I label the third-party type interested only in protecting the minority “protector” and assume that it intervenes if the state fails to address the minority's fears and grievances, but not if the minority rebels after the state has addressed these fears and grievances. The other type, which I label “opportunist,” intervenes whenever an opportunity arises. In addition, I assume that the third party and the minority are fully informed about the third party's preferences, whereas the state can rely only on “Nature's” draw—β for the opportunist type and 1 — β for the protector type.

What is the payoff structure of this modified game? The payoffs for INT1 and INT2 are distributed as follows: qnB — cM and (1 — qn)B — cM, n = {1,2}, for the minority and the state, respectively, where qn is the minority's probability of victory with third-party support in stage n of the game, such that qn > pn. The assumption made earlier that the minority prefers to launch a war of secession with third-party support, but demobilization without it, formally means q2B — cM > bM + θ > p2B – cM. The third party's payoffs for these and all other outcomes of the game are defined as functions of the benefit portion of the minority's payoff and the third party's costs, depending on its type. Thus for INT1, the payoff for both types of third parties is Λq1B – cTP, where Λ is a coefficient pegging the benefit portion of the third party's payoff to the benefit portion of the minority's payoff, and cTP is the third party's cost for intervening. For INT2, the payoffs are Λq2B – cTP and Λq2B – cTP — cpTP for the opportunistic and protector type third parties, respectively, such that Λq2B — cTP > Λp2B > Λq2B — cTP — cpTP. cpTP is an additional cost that the protector type incurs for intervening on behalf of a minority whose grievances have been successfully addressed. Both types prefer intervening to not intervening following WAR1, which means Λq1B — cTP > Λp2B — cTP.

The process of determining the equilibria is similar to that of the game without a third party. The equilibrium behavior of the third party directly follows from the assumptions made about its preferences. Both third-party types intervene after WAR1 by assumption. After WAR2 the opportunistic third party intervenes, whereas the protector type does not.23 The minority's best response strategy is {mobilize, rebel} if the state plays {crack down} or if it plays {make concessions} and the third party is opportunistic, and if bM < q1B — cM. If the state plays {make concessions} and the third party is a protector type, the minority plays {mobilize, demobilize}. The state's best response strategy is again determined by comparing its payoff for {crack down} with its expected payoff for {make concessions}, which now is

formula

Setting this value equal to the payoff for {crack down} and solving for β results in a threshold value of β that makes the state indifferent between two strategies:

formula

Mathematically, (4) and (5) are equivalent to (1) and (2). The conclusions they produce, therefore, are similar. Namely, newly independent states are often reluctant to make concessions to mobilized minorities for fear that these concessions will encourage stiffer demands or a separatist rebellion, only this time with support from another state. The state's reluctance is a function of the degree of its vulnerability, which in turn is a function of the balance of power between the state and the coalition of the minority and its external supporter. I turn now to an empirical examination of the model beginning with post-Soviet Georgia's violent and nonviolent disputes with its minorities.

Georgia's Conflicts with Its Minorities

At first glance, Georgia's violent conflicts with its Abkhaz and Ossetian minorities in the early 1990s appear to present few puzzles. Both conflicts seem to be replays of a familiar script: as an empire's grip starts to loosen, some groups begin to demand independence, while groups that would become subordinate minorities if the first groups were to become independent demand separation from them. Given the incompatibility of these demands, violent conflict seems inevitable, especially if the subordinate group can count on third-party support.24 This narrative, however, obscures more than it reveals about Georgia's wars with its Abkhaz and Ossetian minorities in the early 1990s.

The conflicting nationalist aspirations of both the Abkhaz and the Ossetians do not explain why war, as opposed to a negotiated bargain, was the outcome of their disputes. Their failure to reach negotiated bargains is especially puzzling for three reasons. First, both minority groups were willing to negotiate statuses short of independence prior to the escalation of their conflicts. Second, Georgia had powerful incentives to make almost any concession to avoid war; the country was in the process of establishing statehood, building its institutions, settling a variety of intra-Georgian disputes, and trying to manage a catastrophic economic decline. Finally, Georgia should have been maximally generous to the Abkhaz and the Ossetians given the demographic realities in Abkhazia and South Ossetia. The Abkhaz constituted only 1.7 percent of the population of Georgia in 1989.25 In Abkhazia proper, they represented just 17.8 percent of the total population. Meanwhile Georgians were the largest group there, making up 45.7 percent of the population.26 While the proportion of the Abkhaz in Abkhazia had grown from 15.9 percent in 1970 to 17.8 percent in 1989, Georgians' growth from 41 percent to 45.7 percent during the same period was far more impressive.27 Ossetians constituted only 3 percent of the population of Georgia in 1989.28 In South Ossetia proper, they made up 66.2 percent of the population to Georgians' 28.9 percent, but while the percentage of Ossetians had been declining (from 66.5 percent in 1970), that of the Georgians had been increasing (from 28.3 percent in 1970).29 Bolstering the case for making concessions and reassuring both minorities was the fact that the Abkhaz and the Ossetians were on the receiving end of Georgian mistreatment the last time Georgia was independent, in 1918–21.30 Despite all of these factors, Georgia's response to the Abkhaz and Ossetian mobilizations in the late 1980s was intensely hostile and uncompromising.

The reasons why Georgia should have adopted a conciliatory posture toward the Abkhaz and the Ossetians seem so obvious that irrationality appears to be the only explanation for its behavior. It is, in fact, a popular explanation that is usually combined with parallel explanations focusing on the radicalism and incompetence of Georgia's leaders, especially its first postcommunist leader, Zviad Gamsakhurdia.31 Such claims, however, are mere tautologies masked as explanations. As a result, they raise more questions than they answer. Thus, if Gamsakhurdia was the problem, then how does one explain the fact that Georgia's behavior was not markedly different after his replacement by Edward Shevardnadze, an experienced statesman with decidedly more moderate views? It was under Shevardnadze that Georgia escalated the conflict in Abkhazia to full-scale war. Moreover, if decisionmakers' incompetence and radicalism are all that is needed to explain the wars in Abkhazia and South Ossetia, one must conclude that the same leaders' moderation and competent leadership explain why the disputes in the Azeri-populated Kvemo Kartli and the Armenian-populated Javakheti did not escalate. That clearly would be illogical.

What, then, explains Georgia's overly aggressive and impetuous reaction to the Abkhaz and Ossetian mobilizations? Why did Georgia not only take steps that further radicalized the Abkhaz and the Ossetians, but also escalated both conflicts to shooting wars? By contrast, why was it more circumspect and more willing to compromise when dealing with its Azeri and Armenian minorities?

THE CONFLICTS IN ABKHAZIA AND SOUTH OSSETIA

Mikhail Gorbachev's policy of glasnost provided Georgians an opportunity to openly articulate their political demands and grievances. As elsewhere in the Soviet Union, they began with relatively safe subjects such as environmental problems and corruption. Sensing the possibility that they could take larger risks, however, they soon began to mobilize around distinctly nationalistic issues. Nationalist organizations that had been clandestine and illegal before glasnost began operating openly, and new organizations emerged. All of the leaders of these organizations, including Gamsakhurdia, were Soviet-era dissidents. The nationalist turn was initially confined to issues of cultural revival, reinterpretations of certain historical events, and a perennial concern of the Georgian intelligentsia—the threat of Russification. As Georgians became increasingly confident that they did not have to fear repression, their agenda grew more audacious. It expanded to include the idea of secession from the Soviet Union and rose in popularity at an impressive pace. By the end of 1988, the nationalist opposition was able to rally 200,000 Georgians in Tbilisi, the Georgian capital, to protest amendments to the Soviet constitution designed to allow the Supreme Soviet of the Soviet Union to override laws adopted by legislatures of the union republics.

The Abkhaz and the Ossetians, too, were beginning to mobilize, sensing both a threat from the prospect of a Georgia that was under anything less than tight Soviet control and an opportunity to enlist Moscow as an ally in their shared opposition to Georgian nationalism and separatism. In June 1988, a group of prominent Abkhaz communists traveled to Moscow carrying a letter explaining why Abkhazia should be removed from Georgia's jurisdiction and subordinated directly to Moscow. The watershed event, however, was the signing of a petition in March 1989 by 30,000 Abkhaz making the same case.32 Fueled by anger directed at both the Abkhaz and the leadership in Moscow, Georgians responded with mass rallies. Fearing a total loss of control, Georgian leaders asked Moscow to use force against the demonstrators. On April 9, units of the Soviet army dispersed the demonstrators, killing nineteen. This event further radicalized public opinion in Georgia, destroyed what little legitimacy the communists still enjoyed in the republic, and solidified the view among Georgians that Moscow was using the Abkhaz and the Ossetians to destroy Georgia's aspirations for independence.

Moscow tried to repair the damage by replacing the local communist bosses. The new leadership, however, adopted a strategy of outbidding the nationalist opposition, which was almost certainly not what Moscow intended. In quick succession, Georgia's Supreme Soviet passed a law making use of the Georgian language mandatory in all public institutions throughout the republic (August 1989); declared null and void a decision by South Ossetia's legislature to adopt Ossetian as the official language of the region (November 1989); rejected an appeal from the South Ossetian legislature to upgrade the status of the region from an autonomous oblast to an autonomous republic (November 1989); adopted a resolution declaring that Georgia would not recognize any Soviet law that contradicted the republic's interests (November 1989); and passed a resolution declaring Georgia an occupied state, while proclaiming full independence as Georgia's ultimate goal (March 1990).33

Georgian leaders made no effort to negotiate these measures with the Abkhaz and the Ossetians despite the obvious negative impact their implementation would have on the minorities' interests. Worse still, shortly after the Georgian Supreme Court rejected South Ossetia's request for upgraded status, 30,000 Georgians, led by Gamsakhurdia, marched to the capital of South Ossetia.34 The intervention of the Soviet army prevented the march from turning violent. Nevertheless, the march and the rejection of the request for upgraded status had radicalized the Ossetians, who began organizing paramilitary units. For their part, the Abkhaz adopted a resolution on August 25, 1990, declaring Abkhazia a union republic—an announcement tantamount to a call for secession.35 The Ossetian Supreme Soviet followed suit on September 20, declaring South Ossetia a union republic of the Soviet Union.36 On December 11, the Georgian parliament, which had come under the control of the nationalists after elections a month earlier, passed a law eliminating South Ossetia's autonomous status.37 The Georgian government imposed a blockade, and on January 5, 1991, a 5,000-strong Georgian paramilitary force attacked South Ossetia.38

Although Abkhazia posed the bigger problem initially, and even though it was the first to demand secession, the conflict there did not escalate as quickly as that in South Ossetia. Nor did it follow the same straightforward path toward escalation. Gamsakhurdia reached a power-sharing compromise with the Abkhaz in August 1991—a remarkable feat given the radicalized political atmosphere in Georgia and Gamsakhurdia's overt hypernationalism.39 The agreement guaranteed twenty-eight seats to the Abkhaz in a local parliament with sixty-five seats, even though the Abkhaz made up less than 17 percent of the population of Abkhazia. The Georgians, who made up more than 45 percent of the population, received twenty-six seats, with the remaining eleven going to the Armenians and Russians residing in Abkhazia.40 Shortly after the emergence of Georgia as an independent state and the ouster of Gamsakhurdia in January 1992, the deal broke down.

Initially, Gamsakhurdia's ouster seemed to create an opportunity for Georgia to reevaluate its policies toward Abkhazia and South Ossetia. Some in the new leadership made statements condemning Gamsakhurdia's confrontational posture toward Georgia's minorities, calling him a “parochial fascist.”41 The new leadership's policies, however, were not an improvement on Gamsakhurdia's. In fact, some were even more menacing. A month after Gamsakhurdia's ouster, for example, the Provisional Military Council, Georgia's governing body, readopted the country's 1921 constitution, first adopted by the Bolsheviks shortly before their takeover of the country the same year.42 Making such a decision without a referendum or negotiations with Georgia's minorities, and by a body with suspect legitimacy, was a serious provocation. In addition, the constitution contained no provision for minority self-rule.

Although displeased, the Abkhaz did not declare independence. Instead, they proposed a treaty that would have established what they called “confederal” relations with Tblisi and that would have preserved Georgia's territorial integrity. The Georgian leadership ignored the proposal, and on July 23, 1992, the Abkhaz parliament voted to secede.43 On August 14, Georgian troops attacked Abkhazia.

What conclusions can be drawn from the evidence presented above? First, peaceful bargains were available in both cases. Although it is true that the Abkhaz mobilization began with a demand to secede, parties to a dispute often start out with demands close to their ideal outcomes, revising them in response to counteroffers. This is precisely what happened with the Abkhaz; they basically agreed to forgo secession by accepting the 1991 quota system deal and then proposing the “confederation” treaty in 1992. The same observation is even more applicable to the South Ossetian dispute, which started with the Ossetians' demand for an upgrade to their autonomous status. Second, the Abkhaz and the Ossetians seemed to care more about secession than independence. Evidence for this includes their numerous petitions to be placed under Moscow's direct jurisdiction and a referendum on March 17, 1991, to preserve the Soviet Union, which both minorities voted for by huge margins. This evidence suggests that fear of a future under Georgian rule was an important motivating factor for the actions of both groups. Third, as mentioned earlier the Abkhaz and the Ossetians made up only a small percentage of the population of Georgia, which should have dampened the risks associated with making concessions. Even an Abkhazia or a South Ossetia made more powerful by concessions should not have been considered a formidable adversary. Fourth, Georgia's attacks on Abkhazia and South Ossetia seemed particularly ill-timed given the country's severe economic decline and the challenges of building a new state. Why, then, did Georgia's leaders respond with such impetuous and aggressive behavior?

The answer is that Georgians never considered the conflicts in Abkhazia and South Ossetia as conflicts with the Abkhaz and the Ossetians. Rather, they regarded them as mere extensions of Georgia's conflicts with Moscow. They did not view the Abkhaz and Ossetian demands as based on legitimate grievances. Instead, they believed that the Abkhaz and the Ossetians were acting as Moscow's “fifth columns.” Consequently, in Georgia's view making concessions would lead not to conciliation, but to more extravagant demands and more trouble. Evidence of this mind-set is overwhelming, as suggested in the following statement from the concluding resolution of the Second Congress of the National-Democratic Party of Georgia held in December 1989: “The national movement of Georgia is opposed not by the national movements of Ossetians and Abkhazians, not by their flag and national demands, [but] rather by the flag of the Soviet Union and demands of the unity of the empire. Their main slogan is directed against the unity of Georgia.”44 Similarly, the organizing body of the Georgian national movement declared: “[T]he Georgian People's Front ‘National Accord’ Association believes that, for each ethnic group residing on Georgian territory, it is now most important to determine its attitude toward the Georgian people: whether it will play the part of puppet in the hands of dark, hostile forces or embark together with them in the struggle for their flouted right and the restoration of an independent Georgian state.”45 And during a speech at a rally on December 25, 1989, Gamsakhurdia stated:

“The war against Georgia is about the conquest of Georgia by the Kremlin and continuation of the developments of 1921, even though the war … is waged through the hands of other nationalities in Abkhazia and South Ossetia. This is the ethnic conflict, the ethno-crisis for the Kremlin, whereas for Georgians this is the Russian conquest of our country.”46 In another speech, Gamsakhurdia declared: “[W]hen they criticize our policy for the processes occurring, say, in Samarchablo [the preferred Georgian name for South Ossetia], for the processes taking place in Abkhazia, they always gloss over the fact that all these conflicts were inspired by Moscow and that all tactics were aimed against us with premeditation in order to force Georgia to sign the Union treaty.”47 Georgi Gachechiladze, a prominent Georgian writer, observed: “Usually when an ethnic majority acts against a minority, a feeling of support for the minority arises instinctively. But the minority does not deserve any moral support when it is used as a blind weapon against a nation struggling for its freedom.”48 Mikheil Saakashvili, the third president of Georgia, echoed the same argument: “[Abkhazia and South Ossetia] are not ethnic conflicts. These are political conflicts imposed on us. They are linked to an attempt by post-Soviet forces, the remnants of the old Soviet mentality, to seize control of at least some of the neighboring territories—Georgia was the most attractive piece to gobble up—or, at the very least, to create problems for Georgia.”49

The question, however, is whether Georgians' perceptions were justified. Certainly, the Abkhaz and Ossetian grievances were genuine. And although Moscow exploited the conflicts in Abkhazia and South Ossetia to slow Georgia's drive toward independence, Boris Yeltsin's post-Soviet government in Moscow rejected interventionism. At the same time, facing its own separatist minorities, Russia vociferously opposed similar calls from other minorities. Dmitri Trenin describes Russian attitudes in the period immediately after the Soviet collapse as follows: “The leaders of the ‘new Russia’ saw that the prime national interest was to join NATO and the European Community, rather than restore the Soviet Union. For most of the other republics of the former USSR, this new attitude meant that the Russian government's policy was now one of benign neglect. The willingness to cut losses and withdraw—especially from areas of conflict, and in particular from the Caucasus—was very strong. This new policy was backed by the Russian people, who were growing weary of the country's involvement in petty wars along the Soviet periphery.”50

Consistent with the logic described above, President Yeltsin, Minister of Foreign Affairs Andrei Kozyrev, and other influential Russians made public statements in support of a united Georgia.51 More important, Moscow allowed Georgia to gain control over the Soviet weapons stockpiles stored on its territory. The seriousness of post-Soviet Russia's noninterventionist stance is further demonstrated in a letter written by Kozyrev to Georgia's minister of foreign affairs, Alexander Chikvaidze, shortly after the start of military operations in Abkhazia. In it, Kozyrev expressed support for what he called Georgia's fight against terrorists.52 Finally, Russia did nothing to support the Abkhaz in the initial phase of fighting.

Nevertheless, the stance of the Georgian leadership was not the product of paranoia. The Abkhaz and Ossetian mobilizations may have been driven by certain fears and grievances, but there was no way to guarantee that either minority would not try to secede in the future. Without such guarantees, concessions were dangerous—especially because Georgian leaders had little confidence that Russia's non-imperialist, noninterventionist preferences would survive into the future. Such doubts were based in part on the realist logic that Russia would become more aggressive and more interventionist after its recovery from the Soviet collapse,53 and in part on the opposition to Yeltsin's foreign policy by a powerful and highly vocal element in the Russian political elite, known as the derzhavniki.54 The derzhavniki argued that Yeltsin, and Gorbachev before him, had squandered Russia's power by making too many unilateral concessions to the West. The group also argued that Yeltsin's policy toward Russia's “periphery” was too timid. Georgia was of particular concern because of its geopolitical position and because the conflicts in Abkhazia and South Ossetia had repercussions for the stability of the North Caucasus. In speeches and in articles, the derzhavniki—whose ranks included Yeltsin's vice president, Alexander Rutskoi; the head of the Communist Party, Gennady Zyuganov; an outspoken member of the parliament, Sergei Baburin; and the chairman of the Russian Supreme Soviet, Ruslan Khasbulatov—explicitly advocated intervening in Abkhazia.55

The fear that Russian foreign policy would eventually align with the preferences of the derzhavniki and the expectation that Yeltsin's Russia would not intervene created a temptation to solve the Abkhaz problem with force before it became impossible to do so. This “window of opportunity” logic may seem less relevant in the case of South Ossetia, because the Soviet Union was still intact when that conflict escalated. Still, fears of forceful intervention had already subsided considerably in response to the reluctance, clumsiness, and inconsistency of the Kremlin's earlier decisions to use force.56

THE DISPUTES IN KVEMO KARTLI AND JAVAKHETI

In contrast to its disputes with the Abkhaz and the Ossetians, Georgia's disputes with its Azeri and Armenian minorities did not escalate. In both cases, its leaders acted with greater circumspection and moderation. That behavior, however, begs for an explanation, especially because along certain dimensions the Azeris and especially the Armenians must have seemed no less dangerous than the Abkhaz and the Ossetians.

Azeris were a larger minority than either the Abkhaz or the Ossetians. In 1989 they represented 5.7 percent of the population of Georgia.57 Precise data from 1989 are not available, but according to the 2002 census, Azeris made up 45 percent of the population of the region designated as Kvemo Kartli.58 As with other minorities, the nationalist revival in Georgia caused considerable anxiety among Azeris. Georgian nationalists singled them out as particularly dangerous given their high birth rates. Gamsakhurdia and others openly spoke about the threat of “tatarization” of Georgia.59 Svante Cornell reports that as early as 1989 some nationalist groups blocked the sale of bread in the Bolnisi District, and that Georgian authorities fired Azeri doctors from hospitals in the towns of Bolnisi and Marneuli.60 These same nationalist groups also detonated small explosives in Azeri settlements,61 and in 1991 provoked sporadic violence in an effort to convince the Azeris to flee the country.62 Enraged, the Azeri community retaliated with demonstrations and demands for autonomous status. Gamsakhurdia responded by declaring a state of emergency in the region in November 1991.63

The tensions in Kvemo Kartli subsided over the next few months, for two mutually reinforcing reasons. First, with the escalation of the conflict between Armenia and Azerbaijan over Nagorno-Karabagh in the winter of 1991–92, the Azeris in Georgia realized that an insurgency in Kvemo Kartli would put their kin-state—Azerbaijan—in the position of having to fight a two-front war if it decided to come to their aid. Armenians were the more important adversary for Azerbaijan, so the Azeris of Kvemo Kartli had to be patient. Second, realizing that they did not need to worry about Azeri separatism in Kvemo Kartli and Azerbaijani irredentism as long as Azerbaijan was engaged in a conflict with Armenia, the Georgians decided to pull back. In other words, the Nagorno-Karabagh conflict credibly committed Azerbaijan to a non-irredentist policy for as long as that conflict remained unresolved and reduced the likelihood of Azeri separatism in Kvemo Kartli, which in turn made Georgians less trigger happy.

The same conflict and the same logic explain why the disputes in Armenian-populated Javakheti did not escalate, despite several factors that made Javakheti even more likely to explode than Kvemo Kartli. First, Armenians represented 8.1 percent of the total population of Georgia, making them an even larger minority than the Azeris.64 Nearly half of Georgia's Armenian population resided in a region that Georgia designates as Samtskhe-Javakheti, where Armenians make up approximately 55 percent of the population.65 In a smaller part of this region, which Armenians consider theirs, they represent around 90 percent of the population.66 Such high levels of concentration usually make minorities more prone to separatist demands.67 Second, Javakheti Armenians had created an organization—Javakhk—as early as 1988 that had both political and paramilitary functions and that had as its stated goal regional autonomy.68 Third, the Armenian community in Abkhazia sided with the Abkhaz in the Georgian-Abkhaz conflict. Fourth, in contrast to the Azeri case, Georgia and the Javakheti Armenians' kin-state—Armenia—had sharply divergent preferences on a number of issues, including Russia's continued strategic presence in the Caucasus, Georgia's close ties to Turkey, and whether the norm of territorial integrity trumped the norm of self-determination. Fifth, some nationalist intellectuals and groups in Armenia, the nationalist Dashnak Party most prominently among them, were and remain vocal supporters of Javakheti Armenians' demands for autonomy,69 and have periodically criticized the Armenian government for not backing such demands.

The tensions in Javakheti have reached crisis proportions on numerous occasions. Javakheti Armenians, for example, refused to accept the authority of several prefects appointed by the Gamsakhurdia government, and there have been many clashes and several dangerous standoffs. In December 1991, fighters loyal to the Provisional Military Council tried to enter the region in pursuit of Gamsakhurdia, who was fleeing to Armenia, but were not allowed to. Shortly thereafter, Armenians disarmed Gamsakhurdia's supporters, who were trying to reenter Georgia from Armenia. In 1998, armed Armenians confronted a Georgian military unit heading to the region to conduct military exercises. The government in Tbilisi ordered the unit to withdraw.70 Tensions flared again in 2004 after Mikheil Saakashvili came to power and reinstated Tbilisi's full authority over the region. Violent demonstrations ensued with renewed Armenian demands for autonomy. In 2005 and 2006, violent demonstrations again took place with renewed demands for autonomy, eliciting a stern reaction from Irakli Okruashvili, Georgia's defense minister at the time. Okruashvili threatened to “neutralize” groups that promoted separatist agendas and “anti-Georgian policies.” In 2008, Georgians arrested a leader of a Javakhk splinter group, Vahagn Chakhalyan, who was subsequently sentenced to a ten-year prison term for blowing up a building.71

Despite the high level of tensions, occasional clashes, and differences between Georgia and Armenia on a variety of issues, the disputes in Javakheti have not escalated into serious violence. Both sides have pulled back after edging too close to the precipice. Even though Georgia has not granted Armenians autonomy, it has taken conciliatory measures to reassure them and to address their grievances. Georgian leaders have pursued a strategy of co-opting local Armenian elites, made real attempts to address the socioeconomic problems of the people of Javakheti, and allowed a considerable degree of self-rule.72 They have also accepted Armenia's influence in the region. Georgia has made a practice of requesting the involvement of the Armenian government during crises in Javakheti, because Armenia has invariably played a moderating role. Resentful of Armenia's firm position on discouraging radicalism in Javakheti, a local politician complained: “The problem also comes from Armenia, which easily assumes the role of mediator when the Georgian government appeals for their support in dealing with the region. The Georgian government does not want to speak with us [Javakheti Armenians]. It is only interested in ignoring those who really reflect the people's will and using Armenia to silence us.”73

The significance of this evidence is difficult to overstate. A country as fearful of compromising its sovereignty and as apprehensive of third-party manipulation of its minorities as Georgia has essentially encouraged a third party—the Armenian government—to act as a mediator between it and part of its own population. One could argue that this is because successive Armenian governments have shown little interest in supporting separatism in Javakheti and have done their best to restrain their ethnic kin.74 This is certainly true. At the same time, however, it is important to remember that the moderate posture of the Russian government in 1992 was insufficient to change Georgia's stance on Abkhazia. What has mattered in addition to the moderate posture of successive Armenian governments has been Georgia's realization that this posture does not depend on particular leaders, their ideologies, or their political preferences and that Armenia will remain moderate and non-irredentist as long as the Karabagh conflict is unresolved. Armenia cannot afford a two-front war. Indeed, a war with Georgia, which would be the likely consequence of supporting separatism in Javakheti, would mean more than a second front for Armenia. Given the blockade of the country by Azerbaijan and Turkey and the limited value of the lines of communication running through Iran, Georgia provides Armenia with its only overland access to the outside world. A serious confrontation could strangle Armenia. Every sane Armenian understands this, and every sane Georgian knows that Armenians understand it.

Beyond Georgia

Does the argument put forward in this article apply to disputes beyond Georgia? A rigorous and systematic answer to this question would require a separate study. Here I offer a preliminary assessment in the context of ethnonationalist conflicts in other parts of the postcommunist space.

A cursory review of the conflicts in Krajina, Bosnia-Herzegovina, Nagorno-Karabagh, and Transnistria reveals important similarities with the conflicts in Abkhazia and South Ossetia. First, contrary to narratives popular in the West, Serbs in Krajina and Bosnia-Herzegovina, Armenians in Nagorno-Karabagh, and the Russophone population in Transnistria were willing to settle for something less than full independence or unification with their kin-states at different stages of their conflicts. As Susan Woodward writes, the Serbs of Krajina “wanted restoration of the same right to self-determination that the Croats had. When their petitions fell on deaf ears, they began to demand local autonomy.”75 Bosnian Serbs were similarly willing to forgo secession, but vehemently opposed the idea of a unitary constitution if Bosnia-Herzegovina were to become independent. They consequently agreed to participate in negotiations on the “cantonization” of Bosnia sponsored by the European Union.76 Like Krajina's Serbian population, the Armenians of Nagorno-Karabagh began with a demand for unification with Armenia in 1988. But as the conflict evolved and as the obstacles to the realization of that goal became all too apparent in 1990–91, they started to show unmistakable signs of flexibility. As Thomas de Waal reports, “On June 19, 1991, Party officials from the (still suspended) Regional Soviet passed a resolution expressing their intention to change ‘the course from a policy of confrontation to a policy of dialogue and negotiation.’ They appeared to be offering a bargain: in return for the recreation of Party organs and ‘demilitarization’ of Nagorno-Karabagh, they would vote to suspend their secession from Azerbaijan.”77 In addition, after hopes of a Moscow-imposed solution were dashed in Armenia, a group of young intellectuals assumed the leadership of the Karabagh Movement and gradually shifted its agenda toward independence from the Soviet Union. The group then began to claim that, to survive as an independent state, Armenia would have to resolve its conflicts with its neighbors, including Azerbaijan, through peaceful compromise. It remained committed to this argument after coming to power in 1990.78 The mobilization of the Russophone population of Transnistria, which was in large part a reaction to the Moldovan mobilization that had as one of its goals unification with Romania, started out with a plebiscite “on the desirability of creating an autonomous Transnistrian republic as a constituent of the Moldovan Soviet Socialist republic.”79 Later the Transnistrian movement declared “sovereignty,” but insisted that the word should not be interpreted as reflecting its intention to secede from Moldova. In his electoral platform, the Transnistrian leader Igor Smirnov emphasized his preference for the “retention of a single Moldova as a federal state with a single economic, political, legal, and cultural space.”80

Second, the reactions of the states in each of these cases were similar to Georgia's responses to the Abkhaz and Ossetian mobilizations in both their aggressiveness and their unwillingness to make compromises. Azerbaijan responded to the Armenian mobilization with a pogrom and then took escalatory steps at different stages of the conflict.81 It is not clear who started shooting first in Moldova, but Pål Kolstø and Andrei Malgin state that “in the second half of 1991 the Moldovan authorities tried to regain control over various institutions and organs [in Transnistria] such as courts, the prosecutor's office, and police station by using coercion.”82 Croats and Bosnian Muslims may not have initiated the violent escalations in their conflicts with the Serbs, but they made political decisions—unilateral declarations of independence as unitary states without agreement on the status of their Serbian minorities— that were tantamount to declarations of war.83

Third, as in the Abkhaz and Ossetian cases, third parties stood behind the Serbs of Krajina and Bosnia-Herzegovina, Armenians of Nagorno-Karabagh, and the Russophone population of Transnistria. And as in Georgia, the states in these cases linked their minorities' demands to more offensive and sinister projects involving those third parties—Greater Serbia in the cases of Krajina and Bosnia-Herzegovina, Greater Armenia in the Nagorno-Karabagh case, and Russian imperialism in the Transnistria case.

Fourth, the states combined their fears of current or future collusion of their minorities with hostile third parties and kin-states with assessments that the conflicts could be resolved by force in the short run. Croats grew increasingly emboldened and reluctant to compromise after Germany offered assurances that it would recognize their independence before the status of Krajina could be settled.84 Bosnian Muslims similarly thought that they could afford to abandon the negotiations on the “cantonization” of Bosnia-Herzegovina after the United States decided to support the idea of Bosnian independence without an agreement on the status of Bosnia's Serbs.85 Azerbaijan's leaders thought that Armenia was weak, vulnerable, and without allies, while the Azeris had Turkey on their side—a catastrophically wrong but not an unreasonable assessment at the time. Moldovans thought that the chaos in Moscow and the government's loss of nerve gave them an opportunity to impose a solution in Transnistria.

The nonviolent outcomes of Serbia's, Slovakia's, and Romania's disputes with their Hungarian minorities stand in sharp contrast to the violent cases discussed above. As the grip of the Soviet Union and local communist regimes began to loosen, Hungarian minorities in all three states started to mobilize, issuing demands ranging from increased political rights to separation and unification with Hungary—demands that were initially supported by an irredentist-minded Hungary. All three states responded with discriminatory legislation and curtailed the rights of their Hungarian minorities. A violent clash in 1990 led to the deaths of eight people in Transylvania and the dispatch of the Romanian army to the region. None of these disputes led to full-scale war, however, and all eventually de-escalated. The primary reason was Budapest's decision to pursue integration with the West and the consequent realization by all concerned that such a course would be incompatible with an irredentist policy toward Hungary's neighbors. Hungarian minorities moderated their demands accordingly, while Serbia, Slovakia, and Romania rolled back their discriminatory laws and policies.86

The peaceful outcomes in most of post-Soviet Russia's disputes with its minorities, and the fact that some disputes remained latent, provide another opportunity for comparison with the violent outcomes discussed above. Post-Soviet Russia chose the path of conciliation and reassurance vis-à-vis its restless minorities, in stark contrast to the behavior of Georgia, Croatia, Bosnia-Herzegovina, Azerbaijan, and Moldova. Moscow negotiated a new federal constitution and a federal treaty with the subjects of the Russian Federation, including its ethnically defined units, and devolved considerable authority to them. President Yeltsin famously instructed the leaders of the federal units “to take as much sovereignty, as [they] could swallow.”87

Why was Russia more willing to reassure its restive minorities and make concessions that would strengthen local authorities at the expense of the government in Moscow? The answer is that the risks associated with concessions to minorities were not the same for Russia as they were for the aforementioned states. No matter how empowered and emboldened, no minority would be a match for Russia in a secessionist war. More important, Russian minorities could hardly hope for third-party support, because no third party would want to risk a war with a nuclear-armed great power, which such support would make likely. Even the violent outcome in Chechnya does not undermine this claim. Russia's often brutal behavior in Chechnya after the conflict there escalated deserves the harshest criticism, but the escalation itself was not a consequence of Russia's reluctance to make concessions. Rather, it was the result of the Chechen decision to insist on de jure independence as the minimum acceptable in any bargain.88

The same uncompromising stance by the Albanian minority in Kosovo explains the violent outcome of the conflict there. According to the standard narrative in the West, the conflict erupted as a result of Belgrade's decision in 1989 to downgrade Kosovo's status. The conflict, however, predated that decision and in part was the cause rather than the consequence of Belgrade's decision. The search for a compromise did not end with that decision even if at times it was combined with heavy-handed coercion. In 1992, for example, Serbian Prime Minister Milan Panić met with the leader of Kosovo's Albanians, Ibrahim Rugova, and offered to restore the province's autonomous status. Rugova told Panić that Kosovar independence was the only acceptable solution. Further attempts at compromise also failed, because like the Chechens, Kosovo Albanians insisted on de jure independence as the minimum they would accept in any agreement.89

Alternative Theories

There are several alternative theories of ethnonationalist conflict that scholars have employed to explain the conflicts in the postcommunist space. In this section, I provide a brief overview and critique of these theories.

SEPARATISM AND DETERRENCE FAILURE

The group of theories that regards ethnic conflict as a failure of deterrence is perhaps the most obvious alternative to the argument put forward in this article. Underpinning these theories is the idea that ethnic groups view national self-determination as their ultimate objective, and unless deterred by the threat of high costs and the prospect of failure, they will launch secessionist wars to achieve it. The presence of third parties—typically kin-states—that are willing to support minorities increases the probability of such wars.90

There is much to like about this literature, but it has three important limitations. First, even though contributors to this literature employ bargaining theory, they fail to provide an explicit mechanism that explains why war, as opposed to a negotiated bargain, should be the outcome in these conflicts. Second, this literature does not explain why the addition of a third party should increase the probability of violence. The argument that minorities that would otherwise be deterred are likely to attempt secession if they have third-party support is not convincing, because the addition of a third party—hence a change in the distribution of power—should explain changes in the terms of bargaining, not whether bargaining succeeds or fails.91 Third, this literature implies an escalation scenario, where it is the minority that initiates violence and where it does so solely on the basis of calculations regarding the odds of victory. I have argued, however, that it is often states that initiate violence or assume bargaining positions that make violence inevitable, whereas minorities are willing to compromise.

NATIONALIST NARRATIVES AND CONFLICT

Some scholars who hold nationalism responsible for ethnonationalist conflicts define nationalism as the content of a group's identity or its nationalist narrative, rather than as a principle of political legitimacy and territorial control. Nationalist narratives typically consist of emotionally salient symbols, mythologized stories of victimhood, and claims to special providence. These narratives often refer to those who feel victimized, groups associated with these so-called victims, and groups that stand in the way of the nation's providential mission. Pronounced feelings of hatred against such groups or irrational fears of being victimized by them are corollaries of such narratives, and violence is their direct consequence. Such hatreds are often evident in multiethnic states and empires. When these states and empires become destabilized, however, these hatreds drive some groups to kill those in other groups.92

This theory has been criticized in detail elsewhere.93 Here I examine only an empirical claim that ostensibly confirms the theory and directly contradicts my argument about the conflicts in postcommunist Georgia. According to Stuart Kaufman, Georgia went to war against the Abkhaz and the Ossetians, but not the Armenians and the Azeris, because Georgian attitudes were more hostile to the first two groups. To support his claim, Kaufman cites hostile beliefs about the Abkhaz and the Ossetians that characterized Georgia's nationalist discourse, as well as the results of a survey of Georgian attitudes toward minorities conducted in 1990. The results reveal higher percentages of negative attitudes toward the Abkhaz and the Ossetians than toward the Azeris and the Armenians. According to this survey, 39.9 percent and 34 percent of the respondents who identified themselves as supporters of the nationalist movement had negative attitudes toward the Abkhaz and the Ossetians, respectively. In contrast, only 27.6 percent and 16.4 percent of respondents held negative attitudes toward the Azeris and the Armenians, respectively.94

These figures cannot explain why Georgia fought the Abkhaz and the Ossetians, but not the Armenians and Azerbaijanis. First, it is unclear whether a 6 percent difference between the negative attitudes toward the Ossetians and the Azeris should be accepted as an explanation for why there was war in one case and peace in the other. Second, the survey results say nothing about the direction of causality. It is at least plausible that the higher level of negative attitudes was the consequence of the higher intensity of the conflicts with these two groups, and not the other way around. Third, what Kaufman does not report about the survey is more important than what he does report—namely, the cited numbers cover only respondents who identified themselves as nationalists, which means that at a time of intense political confrontation with the Abkhaz and the Ossetians, only 39.9 percent and 34 percent of nationalist respondents felt hostile toward these two groups. Meanwhile, 22.3 percent of the respondents from the same category had a positive attitude toward the Ossetians, with 40.3 percent feeling neutral. For the Abkhaz the numbers were 19.8 percent and 37.8 percent, respectively.95 In other words, holders of negative attitudes were in a minority, and it bears repeating that these respondents were self-identified supporters of the nationalist movement and that the survey was conducted at a time of especially high tensions with these two minorities. The real puzzle, then, is why the percentages of respondents with negative attitudes toward the Abkhaz and the Ossetians were so small.

As for the hostile myths about the Abkhaz and the Ossetians, they certainly existed, but such myths existed about the other groups as well. Indeed, Kaufman identifies Georgian elites' fear of a demographic decline as a particularly important element in the Georgian “myth-symbol complex.”96 Given the Azeris' high birth rate and Georgian nationalists' frequently expressed fears that Georgia was being “tatarized,” the Azeri community should have been the nationalists' main target. Hostile myths about Armenians also have a long tradition among Georgian nationalists.97 Yet, Georgia chose to de-escalate its disputes with these minorities and escalate those with the Abkhaz and the Ossetians.

THE SECURITY DILEMMA AND ETHNIC CONFLICT

Barry Posen was one of the first scholars to try to make theoretical sense of the conflicts in the postcommunist space. Posen argues that the conflicts were the consequence of the region's “emerging anarchy” and a special type of security dilemma resulting from it. According to Posen, the disintegration of multiethnic empires confronts ethnic groups with an imperative to provide for their own security. In anarchy, however, trying to maximize one's security automatically undermines the security of others. As Robert Jervis argues in his classic analysis of this phenomenon, the security dilemma, though always present, varies in intensity as a function of the offense-defense balance and the actors' ability to distinguish offense from defense, which in turn are functions of geography and technology.98 Posen focuses on the newly independent groups' patterns of intermingling and their degree of military sophistication to measure the offense-defense balance and whether offense and defense can be distinguished. Specifically, he argues that offense has the advantage if pockets of one ethnic group are surrounded by another group. Such pockets are vulnerable to attack, which makes them tempting targets, which in turn generates preemptive rescue incentives. Posen also argues that the main military asset of newly independent groups is the motivation of the infantry soldier, which can be bolstered by nationalistic fervor. Nationalistic fervor, however, can be a useful motivator for both offense and defense.99

The problem with Posen's argument is that it is designed to explain how short-term vulnerabilities may create incentives for preemptive attacks as ethnic groups find themselves deprived of the protection of a central authority. The conflicts that I have discussed in this article, however, erupted before the central authority had vanished and before such vulnerabilities had become a concern. The evidence suggests that the parties to these conflicts were concerned about long-term vulnerabilities and that it was obstacles to addressing these vulnerabilities that made peaceful compromises difficult to reach.

ELITE MANIPULATION

Elite manipulation theory focuses on the fact that many ethnonationalist conflicts take place in times of societal upheaval and transformation. Such upheavals are typically characterized by elites' simultaneous inability to address society's needs and to suppress its discontent. Elites that find themselves in this predicament acquire incentives to embrace aggressive nationalism, because aggressive nationalism entails a shift in political priorities from elites' failures to conflicts with outsiders. The conflicts themselves can be either manufactured or fueled by the hard-line behavior of these elites. Either way, these conflicts are likely to produce a “rally ‘round the flag” effect and to sideline internal conflict. The “rally ‘round the flag” effect in turn increases the probability of violence.100

The theory of elite manipulation provides an accurate depiction of right-wing nationalism, and not only in times of societal upheaval and political transition. Elites almost by definition have an interest in preserving the political and socioeconomic status quo, which is always made easier when challenges stemming from socioeconomic disparities or political grievances are countered with arguments for national unity. Nationalism is not always a consequence of elite manipulation, however. It can be an instrument for organizing collective action that benefits the nation, not just the elites, and external conflicts are sometimes real. The theory is problematic empirically as well, at least insofar as its application to postcommunist cases, because it generates empirical predictions regarding the evolution of mobilizations and conflicts that are not borne out by the record. If the theory is right, the cases of mobilization discussed in this article should have initially centered on socioeconomic issues and put the elites in their crosshairs. After being manipulated by these very elites, the masses should have made socioeconomic issues a lower priority and embraced a nationalist agenda, which then should have led to violence. I am aware of no case in the postcommunist space that followed this path. In fact, every case started out as a bottom-up nationalist mobilization to which communist elites had to adjust reluctantly.

Consider the Serbian mobilization, which is the paradigmatic case for the proponents of the elite manipulation theory. They maintain that the Serbian masses had mobilized to protest the failed economic policies of Yugoslavia's Communist Party, threatening its grip on power. In response, a segment of the Party, headed by the unscrupulous and opportunistic Slobodan Milošević, embraced the politics of nationalism. Milošević and his loyalists succeeded in marginalizing the old guard of the Communist Party, then used their leadership positions and control of the media to revise the Serbian agenda. That agenda included irredentist demands against Croatia and Bosnia-Herzegovina, as well as the harsh repression of Kosovo Albanians. Milošević neutralized the opposition with this new agenda, but he also opened the floodgates of Serbian nationalism, which plunged Yugoslavia into a series of wars.

The actual record, meanwhile, suggests a different interpretation. First, Serbian nationalism in the 1980s was not manufactured by Milošević. Serbs had already mobilized over the plight of their ethnic kin in Kosovo; what made Milošević famous and popular was his embrace of that mobilization. It is true that Milošević was an opportunist who, sensing political promise, switched from embracing communism to Serbian nationalism. At best, however, this means that he was the product, not the producer, of Serbian nationalism. Second, the political shift among Serbian elites was aided by the loss of incentives to contain Serbian nationalism. That loss, however, was not the result of society's mobilization with a socioeconomic agenda and the need to divert the public's attention. Rather, it was the result of Croatia's and Slovenia's decisions to secede from Yugoslavia, which undermined the “Yugoslav bargain.”101 Third, the disputes escalated because of the interaction of Serbian positions with those of their opponents, not because of Serbian actions alone. And as I have already pointed out, the hard bargains that Croats, Bosnian Muslims, and Kosovar Albanians drove contributed to the failure to settle the disputes in Yugoslavia peacefully.

The Soviet cases are even more problematic for the elite manipulation theory. Some of the mobilizations in these cases did start out as protests against environmental degradation or corruption. As I have argued, however, these were attempts to test the limits of what Moscow would tolerate. As soon as it became clear that the limits were not very strict, the nationalist agendas of these mobilizations were quickly revealed. In every case, the local Communist Party leadership vehemently resisted these mobilizations. Some—like Georgian communists—eventually embraced the agendas of nationalist movements and even tried to outbid the opposition's nationalist leaders, but that happened under pressure from nationalist masses. Some of the conflicts turned violent after the popular movements came to power even though they did not need to drum up political support by assuming hard-line positions. Finally, elite manipulation theory has difficulty explaining why a country such as Georgia would fight some minorities but not others, or why nationalism produced conflicts in some transitional countries but not others.102

AUTONOMY

In an earlier study, Cornell observes the same variation in the outcomes of Georgia's disputes with its minorities and links it to the fact that the violent disputes involved minorities with autonomous status, whereas the disputes that were managed peacefully involved minorities with no such status. In addition, he notes that the other violent cases in the Soviet Union, with the exception of Transnistria, also involved autonomous minorities, and he hypothesizes that the existence of autonomous institutions explains ethnonationalist violence.103

Cornell's study is representative of a larger body of scholarship claiming that ethnofederal institutions facilitate separatism and conflict.104 As he writes,

Autonomous regions typically possess statelike institutions that can be crucial factors in promoting ethnic mobilization. Unlike nonautonomous minorities, minorities in autonomous regions typically have governments and parliaments that act as legitimate representatives of their ethnic constituencies and constitute legitimate decision-making bodies. Parliaments can pass language laws, refuse legislation from the central government, and issue declarations of sovereignty and independence. A minority with autonomous status hence has institutions for challenging state authorities in general and its specific policies and actions in particular. A minority lacking such institutions, by contrast, would find mounting such a challenge more difficult.105

Below is Cornell's explanation of the different outcomes in Javakheti and South Ossetia:

The absence of conflict in Javakheti must be associated with the lack of a strong and legitimate leadership, especially in comparison with South Ossetia, which had a comparable conflict potential. Ademon Nykhas [the popular South Ossetian organization] did not have a higher degree of initial popular legitimacy than Javakhk; the key difference in the development of the two organizations was that the autonomy enjoyed by South Ossetia facilitated cohesion and the strengthening of the nationalist/separatist movement around the governmental institutions of the region. In Javakheti the Javakhk movement needed to build up its position on its own, including the creation of provisional administrative structures; in South Ossetia, such institutions already existed with a rigid hierarchy and an accepted decision-making process. The legitimacy of national leaders was determined not only by their personality and achievements, but also by the posts they held. Moreover, when unfavorable laws were introduced by the Georgian parliament, Javakheti Armenians had little to respond with except petitions and popular demonstrations. Ossetians by contrast possessed a legislative body, the Supreme Soviet of the autonomous province, which provided them with an institutional channel for the struggle against Georgian actions.106

I have argued that states are often reluctant to make concessions to mobilized minorities because these concessions can strengthen the minorities' ability to mount separatist challenges in the future. In a sense, Cornell's argument is about that future (i.e., about the improved ability to organize collective action accorded to certain minorities by their autonomous institutions). Although I agree with Cornell that these institutions facilitate the ability of minorities to organize and challenge states more effectively, I am not persuaded that they explain the variation that he has set out to address.

There are three reasons for my skepticism. First, Cornell's argument implies that the minorities' ability to organize and make demands is not only necessary but also sufficient for explaining violent state-minority conflicts. To more fully explain these conflicts, however, scholars must account for states' behavior as well. State-minority conflicts are typically the consequence of interaction, not of the actions of only one of the parties. Cornell's argument has little appreciation for this fact. Second, the lack of autonomous institutions did little to prevent Javakheti Armenians from organizing. The evidence is fairly strong that the combination of Armenia's moderating role and Tbilisi's willingness to co-opt the local population and elites by offering concessions deradicalized the Javakheti Armenians. Cornell himself acknowledges this fact,107 which means that his claim about the violent and nonviolent outcomes in Georgia at the very least has a problem of equifinality. Third, a more complete examination of the empirical record does not support Cornell's claim that the presence of autonomous institutions explains violent ethnonationalist conflicts in the postcommunist space. Yugoslavia and the Soviet Union contained a total of fifty autonomous regions, but only five—Abkhazia, South Ossetia, Chechnya, Nagorno-Karabagh, and Kosovo—became embroiled in violent conflict. At the same time, three out of the eight violent conflicts in Yugoslavia and the Soviet Union—Bosnia-Herzegovina, Krajina, and Transnistria—involved minorities with no autonomous status. This is a weak empirical foundation for claiming that autonomy was at the root of ethnonationalist conflicts in the postcommunist space.

Conclusion

In explaining ethnonationalist violence, many studies focus on the role played by villainous leaders, clashing cultures, nationalist narratives, and flawed political institutions in states where such violence takes place. James Fearon's call to examine more closely the strategic environment in which these conflicts occur and his argument that they result from the inability of states to make credible commitments to minorities is a welcome corrective to these claims. Fearon's argument is also a persuasive explanation for minority mobilizations in destabilized empires and multiethnic states, but it highlights only part of a more complex strategic interaction. In particular, it does not explain why some states adopt policies that exacerbate minorities' fears or respond violently to minority mobilizations, and why state responses to minority mobilizations vary.

In this article, I have argued that not only minorities but also states have rational fears of exploitation, which may dramatically complicate the search for peaceful bargains. Offering concessions to settle disputes with minorities may make states vulnerable to still more demands or, worse, to separatist bids by minorities that have become empowered and emboldened by those very concessions. At the same time, the intensity of such fears is not constant. It varies depending on whether there are third parties likely to support minorities' future secessionist aspirations as well as on the balance of power between the state, on the one hand, and the minority and the third party, on the other.

Evidence from post-Soviet Georgia's violent and nonviolent disputes with its minorities in the early 1990s supports my argument. Georgia responded violently to mobilizations by the Abkhaz and the Ossetians, despite signals that these minorities would be willing to make compromises, because of government fears that compromise would embolden them to pursue secession more vigorously. Expectations that Russia would eventually reverse its noninterventionist posture made the prospect of Abkhaz and Ossetian secessionism intolerably dangerous. By contrast, Georgian leaders held very different expectations regarding the Azeris and the Armenians. The conflict in Nagorno-Karabagh caused Georgian elites to expect more restrained behavior from these minorities, as well as non-irredentist postures from their kin-states—Azerbaijan and Armenia. Indeed, they estimated correctly that Azerbaijan and Armenia would help to restrain their ethnic kin in Georgia for as long as the Karabagh conflict remained unresolved. My brief analysis of other violent and nonviolent state-minority conflicts in the post-Soviet space is also consistent with the logic of the argument.

The article makes four general theoretical contributions to the literature on ethnonationalist conflict. First, it provides additional support for the general claim that violent state-minority conflicts can be tragic, in the sense that these conflicts can take place even if the belligerents are perfectly rational and prefer a peaceful bargain to violence. Second, the article draws attention to a common error in the study of state-minority conflicts, which is the treatment of minority mobilization as a sufficient cause of violence. It shows both theoretically and empirically that minority mobilization can be at best a necessary but not a sufficient cause. A good theory of violent state-minority conflict must account for states' responses to these mobilizations instead of equating mobilization with rebellion or implicitly assuming that violence is the state's natural response. Third, it shows that a good theory should avoid the opposite error as well, namely, the error of assuming that a state's violent response is a manifestation of its ceteris paribus preference for persecuting the minority. Finally, contrary to the popular view that unchecked power is the reason why states persecute minorities, this article implies that tolerance and generosity toward minorities may be positively correlated with the state's power position in the international system: powerful states that do not fear their minorities colluding with third parties can afford to be more tolerant and generous. Of course, unchecked power can be dangerous but only, or especially, when it is threatened in the long run.

Acknowledgments

The author is grateful to Fotini Christia, Robert Jervis, Chaim Kaufmann, Charles King, Andrew Kydd, Rajan Menon, Kevin Narizny, Constantino Pischedda, Holger Schmidt, Jack Snyder, Robert Trager, and the anonymous reviewers for helpful comments on various drafts of the article. David Matsaberidze ably served as a research assistant and a translator of Georgian-language documents. The author also acknowledges institutional support from the Center for Comparative and International Studies at the Swiss Federal Institute of Technology, the Armenian Studies Program at the University of Michigan, and Lehigh University during different stages of work on the article.

Notes

1. 

James D. Fearon, “Rationalist Explanations for War,” International Organization, Vol. 49, No. 3 (Summer 1995), p. 380.

2. 

James D. Fearon, “Commitment Problems and the Spread of Ethnic Conflict,” in David A. Lake and Donald Rothchild, eds., The International Spread of Ethnic Conflict: Fear, Diffusion, and Escalation (Princeton, N.J.: Princeton University Press, 1998), pp. 107–126.

3. 

The Armenian president stated that international guarantees for the security of Karabagh Armenians would be the only way to solve the conflict. See ibid., p. 120.

4. 

Majorities of Abkhaz, Ossetians, and Russian speakers in Transnistria voted to preserve the Soviet Union in the referendum of March 17, 1991, even as they were campaigning to secede from Georgia and Moldova.

5. 

References to past persecution are usually perceived as evidence for irrational “ancient hatreds” motivating belligerents in ethnic conflicts. As defined, however, rationality requires the updating of beliefs on the basis of the available information. Therefore, ignoring past persecution would actually be irrational. See Jon Elster, Sour Grapes: Studies in the Subversion of Rationality (Cambridge: Cambridge University Press, 1977), pp. 1–4.

6. 

See Donald L. Horowitz, Ethnic Groups in Conflict (Berkeley: University of California Press, 1985), pp. 186–194.

7. 

Robert Jervis, Perception and Misperception in International Politics (Princeton, N.J.: Princeton University Press, 1976), chap. 3.

8. 

Most minorities would probably prefer secession if it involved no costs and not seceding offered no benefits. Secession, however, incurs a risk of state resistance and failure to achieve recognition, whereas not seceding can have benefits, such as access to a larger market. Thus, it is rare that minorities' motives are purely fear-driven or purely secessionist.

9. 

The seminal works on costly signals include Thomas C. Schelling, Arms and Influence (New Haven, Conn.: Yale University Press, 1966); Robert Jervis, The Logic of Images in International Relations (New York: Columbia University Press, 1970); George A. Akerloff, “The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism,” Quarterly Journal of Economics, Vol. 84, No. 3 (August 1970), pp. 488–500; Michael Spence, “Job Market Signaling,” Quarterly Journal of Economics, Vol. 87, No. 3 (August 1973), pp. 355–374; James D. Morrow, “Capabilities, Uncertainty, and Resolve: A Limited Information Model of Crisis Bargaining,” American Journal of Political Science, Vol. 33, No. 4 (November 1989), pp. 941–972; and James D. Fearon, “Signaling Foreign Policy Interests: Tying Hands vs. Sinking Costs,” Journal of Conflict Resolution, Vol. 41, No. 1 (February 1997), pp. 68–90.

10. 

For a related point, see Isak Svensson, “Bargaining, Bias, and Peace Brokers: How Rebels Commit to Peace,” Journal of Peace Research, Vol. 44, No. 2 (March 2007), pp. 177–194.

11. 

Security-seeking minorities, in other words, face what is known as the “credibility paradox”: the same actions designed to make their threats of escalation credible undermine their promise of cooperation if those demands are met. For an analysis of the problem, see Max Abrahms, “The Credibility Paradox: Violence as a Double-Edged Sword in International Politics,” International Studies Quarterly, Vol. 57, No. 4 (December 2013), pp. 660–671.

12. 

Alicia Levine, “Political Accommodation and the Prevention of Secessionist Violence,” in Michael E. Brown, ed., The International Dimensions of Internal Conflict (Cambridge, Mass.: MIT Press, 1996), pp. 311–340.

13. 

Monica Duffy Toft, The Geography of Ethnic Violence: Identity, Interests, and the Indivisibility of Territory (Princeton, N.J.: Princeton University Press, 2003), pp. 21–26.

14. 

James D. Fearon and David Laitin, “Ethnicity, Insurgency, and Civil War,” American Political Science Review, Vol. 97, No. 1 (February 2003), pp. 75–90.

15. 

For a discussion of the method, see Alexander L. George and Andrew Bennett, Case Studies and Theory Development in the Social Sciences (Cambridge, Mass.: MIT Press, 2005), pp. 153–160.

16. 

Toft, The Geography of Ethnic Violence, p. 20. On the indivisibility of territory endowed with special meaning, see also Ron E. Hassner, War on Sacred Grounds (Ithaca, N.Y.: Cornell University Press, 2009).

17. 

Ernest Gellner, Nations and Nationalism (Ithaca, N.Y.: Cornell University Press, 1983), p. 1.

18. 

Concessions usually involve increased self-rule, control over local resources, and the ability to make decisions on educational and cultural matters.

19. 

The variable B can be conceptualized as total political control of the territory where the minority resides.

20. 

The constraint p2 > p1 reflects the assumption that concessions improve the minority's capacity to organize collective action and to fight. Exactly how concessions have this effect is an interesting question in its own right. It would be reasonable to assume, for example, that p2 is a logistic function of p1, which means that the same concession has a relatively small effect if it is made to a very weak or a very strong minority, and a much larger effect on minorities that are neither very weak nor very strong. A very weak minority is likely to be unable to mount a secessionist challenge even after concessions, and a very strong one will be able to do so even without concessions. For the purposes of this article, however, the assumption about the precise relationship between p2 and p1 has no consequence as long as p2 > p1.

21. 

The parameter cem is a convenient mathematical expression of the idea that the appeasable minority attaches a smaller value to independence than to addressing its fears and grievances. It is not literally an additional cost.

22. 

The latter category would include both irredentist kin-states and states or third parties that have disputes with the state unrelated to the minority.

23. 

The assumption that a protector-type third party will refuse to intervene following WAR2 may be controversial, because an argument can be made that a third party that is biased in favor of the minority cannot turn its back on that minority, even if the minority has caused a conflict that the third party did not want. This “moral hazard” logic, however, is unpersuasive, because third parties that are not kin-states and that are interested only in protecting the minority against mistreatment can indeed credibly threaten to abandon the minority if the latter behaves in ways not consistent with the third parties' interests. Kin-states may be more susceptible to moral hazard, but this is counteracted by their considerable abilities to restrain the minority without threatening abandonment. For a study defending the moral hazard logic, see Alan Kuperman, “The Moral Hazard of Humanitarian Intervention: Lessons from the Balkans,” International Studies Quarterly, Vol. 52, No. 1 (March 2008), pp. 49–80. For a critique of the moral hazard logic, see Arman Grigoryan, “Third-Party Intervention and the Escalation of State-Minority Conflicts,” International Studies Quarterly, Vol. 54, No. 4 (December 2010), pp. 1145–1147.

24. 

For studies of Georgia's conflicts with its minorities that rely on the logic of this narrative, see Svante E. Cornell, Small Nations and Great Powers: A Study of Ethnopolitical Conflict in the Caucasus (Surrey, U.K.: Curzon, 2001), chap. 4; Thomas de Waal, The Caucasus: An Introduction (Oxford: Oxford University Press, 2010), chap. 5; and Ghia Nodia, “Political Turmoil in Georgia and the Ethnic Politics of Zviad Gamsakhurdia,” in Bruno Coppieters, ed., Contested Borders in the Caucasus (Brussels: VUB University Press, 1996), pp. 81–85.

27. 

Ibid.

28. 

“Demographics of Georgia (Country).”

30. 

Cornell, Small Nations and Great Powers, pp. 148–149.

31. 

According to Liz Fuller, “[T]he intelligentsia … were alarmed at Gamsakhurdia's irrational and dictatorial leadership style.” See Fuller, “Former Georgian Defense Minister Sentenced,” OMNI Analytical Briefs, October 11,1996, Records of the Open Media Research Institute, Open Society Archives, Central European University, Budapest, http://fa.osaarchivum.org/ft?col=210&i=383. Ronald D. Asmus similarly refers to Gamsakhurdia's rule as “increasingly irrational and authoritarian.” See Asmus, A Little War That Shook the World: Russia, Georgia, and the Future of the West (New York: Palgrave Macmillan, 2010), p. 61. References to Gamsakhurdia as a fascist were common even among those sympathetic to Georgia. See Scott Shane, “Nationalist Leader in Soviet Georgia Turns Georgians against Minorities,” Baltimore Sun, March 7, 1991; and Christoph Zürcher, Post-Soviet Wars: Rebellion, Ethnic Conflict, and Nationhood in the Caucasus (New York: New York University Press, 2007), p. 127.

32. 

This petition is known as the Lykhni Declaration. See de Waal, The Caucasus, p. 131.

33. 

Tamaz Diasamidze, ed., Regional Conflicts in Georgia (The Autonomous Oblast of South Ossetia and the Autonomous SSR of Abkhazia 1989–2008): The Collection of Political-Legal Acts (Tbilisi: Pirveli Stamba, 2008), pp. 8, 15–24.

34. 

Zürcher, Post-Soviet Wars, p. 124.

35. 

Diasamidze, ed., Regional Conflicts in Georgia, pp. 26–28.

36. 

Ibid., pp. 31–34.

37. 

Ibid., pp. 38–39.

38. 

Zürcher, Post-Soviet Wars, p. 125.

39. 

The cause of Gamsakhurdia's newfound moderation was the considerable erosion of his popularity in Georgia, along with the ascendance of reactionaries in the Soviet government in 1991 and the corresponding increase in the likelihood that they would exploit opportunities to intervene against republics with noncommunist governments.

40. 

Zürcher, Post-Soviet Wars, p. 130.

41. 

Ghia Nodia, “Georgia: Dimensions of Insecurity,” in Bruno Coppieters and Robert Legvold, eds., Statehood and Security: Georgia after the Rose Revolution (Cambridge, Mass.: MIT Press, 2005), p. 47.

42. 

Diasamidze, Regional Conflicts in Georgia, pp. 95–96.

43. 

Ibid., pp. 120–122.

44. 

Sakartvelos Khronikebi, December 10, 1989.

45. 

“Manifesto of the Georgian People's Front ‘National Accord’ Association—14 November, 1989,” in Charles F. Furtado Jr. and Andrea Chandler, eds., Perestroika in the Soviet Republics: Documents on the National Question (Boulder, Colo.: Westview, 1992), p. 382.

46. 

Sakartvelos Respublika, December 20, 1990.

47. 

Furtado and Chandler, Perestroika in the Soviet Republics, p. 396.

48. 

Ronald Grigor Suny, The Making of the Georgian Nation (Bloomington: Indiana University Press, 1994), p. 325.

49. 

International Crisis Group, “Abkhazia Today,” Europe Report No. 176 (Brussels: International Crisis Group, September 15, 2006), p. 7.

50. 

Dmitri Trenin, “Russia's Security Interests and Policies in the Caucasus Region,” in Coppieters, Contested Borders in the Caucasus, p. 93.

51. 

Izvestia, August 26, 1992.

52. 

See Oksana Antonenko, “Frozen Uncertainty: Russia and the Conflict in Abkhazia,” in Coppieters and Legvold, Statehood and Security, p. 211.

53. 

Doubts about whether Russia would remain a non-imperialist power were not confined to Georgia. Some Western policy analysts expressed similar concerns. See Zbigniew Brzezinski, “The Premature Partnership,” Foreign Affairs, Vol. 73, No. 2 (March/April 1994), pp. 67–82; and Peter Rodman, “4 More for NATO,” Washington Post, December 13, 1994.

54. 

The word derzhavnik comes from derzhava, which is the Russian word for great power. Derzhavnik can be loosely translated as “proponent of a great power status.”

55. 

Nicole J. Jackson, Russian Foreign Policy and the CIS: Theories, Debates, Actions (London: Routledge, 2003), p. 124.

56. 

As noted above, reactionaries gained increased influence in the Soviet government in the months preceding the attempted coup d'état in August 1991, making such fears real again. The Ossetian escalation, however, occurred before the coup attempt.

57. 

“Demographics of Georgia (Country).”

58. 

International Crisis Group, “Georgia's Armenian and Azeri Minorities,” Europe Report No. 178 (Brussels: International Crisis Group, November 22, 2006), p. 4.

59. 

Stuart J. Kaufmann, Modern Hatreds: The Symbolic Politics of Ethnic War (Ithaca, N.Y.: Cornell University Press, 2001), p. 124. “Tatar” is what Azeris were called in tsarist Russia. The word is now an ethnic slur.

60. 

Svante E. Cornell, Autonomy and Conflict: Ethnoterritoriality and Separatism in the South Caucasus— Cases in Georgia (Uppsala, Sweden: Uppsala University Press, 2001), p. 159.

61. 

Ibid.

62. 

Elizabeth Fuller, “Azerbaijani Exodus from Georgia Imminent?” RFE/RL Report on the USSR, February 15, 1991.

63. 

Ibid.

64. 

“Demographics of Georgia (Country).”

65. 

International Crisis Group, “Georgia's Armenian and Azeri Minorities,” p. 3.

66. 

Georgia created the administrative unit Samtskhe-Javakheti by merging the primarily Georgian Samtskhe with the primarily Armenian Javakheti to avoid having an administrative district that was overwhelmingly Armenian. See Jonathan Wheatly, “Obstacles Impeding the Regional Integration of the Javakheti Region of Georgia,” Working Paper No. 22 (Flensburg, Germany: European Center for Minority Issues, 2004), p. 14.

67. 

Levine, “Political Accommodation and the Prevention of Secessionist Violence,” p. 313.

68. 

Ibid., p. 13.

69. 

Armenian Revolutionary Federation-Dashnaktsutyun, “Foreign Policy and Strategy: Armenia-Turkey and Armenian-Turkish Relations,” http://www.arfd.info/arf-d-foreign-policy-strategy/.

70. 

Cornell, Autonomy and Conflict, pp. 206–207.

71. 

International Crisis Group, “Georgia: The Javakheti Region's Integration Challenges” (Brussels: International Crisis Group, May 23, 2011), pp. 3–5.

72. 

Ibid., p. 5.

73. 

“Georgia's Armenian and Azeri Minorities,” p. 19.

74. 

International Crisis Group, “Georgia,” pp. 11–13.

75. 

Susan L. Woodward, Balkan Tragedy: Chaos and Dissolution after the Cold War (Washington, D.C.: Brookings Institution Press, 1995), p. 108.

76. 

See Stephen L. Burg and Paul S. Shoup, The War in Bosnia-Herzegovina: Ethnic Conflict and International Intervention (New York: M.E. Sharpe, 1999), pp. 108–117.

77. 

Thomas de Waal, Black Garden: Armenia and Azerbaijan through War and Peace (New York: New York University Press, 2003), p. 118. On the increased flexibility of the Armenian position in 1991, see also Erik Melander, “The Nagorno-Karabagh Conflict Revisited: Was the War Inevitable?” Journal of Cold War Studies, Vol. 3, No. 2 (Spring 2001), pp. 69–70.

78. 

For a set of fascinating documents tracing this transformation, see Gerard J. Libaridian, ed., Armenia at the Crossroads: Democracy and Nationhood in the Post-Soviet Era (Watertown, Mass.: Blue Crane, 1991).

79. 

Pål Kolstø and Andrei Malgin, “The Transnistrian Republic: A Case of Politicized Regionalism,” Nationalities Papers, Vol. 26, No. 1 (March 1998), p. 107.

80. 

Ibid., p. 109.

81. 

The pogrom took place in late February 1988 in the town of Sumgait. It was followed by sporadic violence, expulsions of Armenians from several Azerbaijani cities, and expulsions of Azerbaijanis from Armenia. While Armenia began to moderate its position, Azerbaijan's position kept hardening. In the spring of 1991, Soviet interior ministry troops, together with the special forces of the Azerbaijani interior ministry, conducted so-called Operation Ring, which resulted in the expulsion of Armenians from a dozen villages near Nagorno-Karabagh. The Armenians correctly regarded this operation as a prelude to what was in store for Nagorno-Karabagh itself. On November 26, 1991, Azerbaijan abolished the autonomous status of Nagorno-Karabagh, and it began shelling the residential areas of the capital, Stepanakert, soon after the collapse of the Soviet Union. For more details, see de Waal, Black Garden, chaps. 8, 11.

82. 

Kolstø and Malgin, “The Transnistrian Republic,” p. 109.

83. 

See Woodward, Balkan Tragedy, chap. 5, and Burg and Shoup, The War in Bosnia and Herzegovina, chap. 3.

84. 

John Zametica, “The Yugoslav Conflict,” Adelphi Paper No. 270 (London: International Institute for Strategic Studies, 1992), pp. 18–19.

85. 

Burg and Shoup, The War in Bosnia-Herzegovina, p. 113.

86. 

Erin K. Jenne, Ethnic Bargaining: The Paradox of Minority Empowerment (Ithaca, N.Y.: Cornell University Press, 2006), chaps. 4, 6.

87. 

“Beyond the Kremlin's Reach,” Economist, January 30, 2010, p. 63.

88. 

See Valery Tishkov, Chechnya: Life in a War-Torn Society (Berkeley: University of California Press, 2004), chap. 5; and Anatol Lieven, Chechnya: Tombstone of Russian Power (New Haven, Conn.: Yale University Press, 1999), chap. 2.

89. 

Space constraints do not permit a more detailed discussion of this claim. See Grigoryan, “Third-Party Intervention and the Escalation of State-Minority Conflicts,” pp. 1155–1163.

90. 

Myron Weiner, “The Macedonian Syndrome: An Historical Model of International Relations and Political Development,” World Politics, Vol. 23, No. 4 (July 1973), pp. 665–683; Alexis Heraclides, “Secessionist Minorities and External Involvement,” International Organization, Vol. 44, No. 3 (Summer 1990), pp. 341–378; Stephen M. Saideman, The Ties That Divide: Ethnic Politics, Foreign Policy, and International Conflict (New York: Columbia University Press, 2001); Jenne, Ethnic Bargaining; and Harris Mylonas, The Politics of Nation-Building: Making Co-nationals, Refugees, and Minorities (Cambridge: Cambridge University Press, 2013).

91. 

Rupen Cetinyan, “Ethnic Bargaining in the Shadow of Third-Party Intervention,” International Organization, Vol. 56, No. 3 (Summer 2002), p. 647.

92. 

The most straightforward statement of this argument is Kaufman, Modern Hatreds. For similar arguments, see Roger D. Petersen, Understanding Ethnic Violence: Fear, Hatred, and Resentment in Twentieth-Century Eastern Europe (Cambridge: Cambridge University Press, 2002); and Badredine Arfi, “Ethnic Fear: The Social Construction of Insecurity,” Security Studies, Vol. 8, No. 1 (Autumn 1998), pp. 151–203.

93. 

See Arman Grigorian, “Correspondence: Hate Narratives and Ethnic Conflict,” International Security, Vol. 31, No. 4 (Spring 2007), pp. 180–186; and James D. Fearon and David D. Laitin, “Violence and the Social Construction of Ethnic Identity,” International Organization, Vol. 54, No. 4, (Autumn 2000), pp. 845–877.

94. 

Kaufman misstates the numbers as percentages of the total population of Georgians. See Kaufman Modern Hatreds, pp. 94–95. For a more detailed presentation and analysis of the survey, see Lynn D. Nelson and Paata Amonashvili, “Voting and Political Attitudes in Soviet Georgia,” Soviet Studies, Vol. 44, No. 4 (1992), p. 692.

95. 

Ibid.

96. 

Kaufman, Modern Hatreds, p. 93.

97. 

The myths date back to the nineteenth century, when Armenians became economically and socially dominant in Georgia, particularly in Tbilisi. See Suny, The Making of the Georgian Nation, pp. 118–119.

98. 

Robert Jervis, “Cooperation under the Security Dilemma,” World Politics, Vol. 30, No. 2 (January 1978), pp. 167–214.

99. 

Barry R. Posen, “The Security Dilemma and Ethnic Conflict,” Survival, Vol. 35, No. 1 (Spring 1993), pp. 27–47.

100. 

Jack Snyder and Edward D. Mansfield, “Democratization and the Danger of War,” International Security, Vol. 20, No. 1 (Summer 1995), pp. 5–38; Jack Snyder, From Voting to Violence: Democratization and Nationalist Conflict (New York: W.W. Norton, 2000); V.P. Gagnon Jr., “Ethnic Nationalism and International Conflict: The Case of Serbia,” International Security, Vol. 19, No. 3 (Winter 1994/95), pp. 130–166; and V.P. Gagnon Jr., The Myth of Ethnic War: Serbia and Croatia in the 1990s (Ithaca, N.Y.: Cornell University Press, 2006).

101. 

One of the main features of that bargain was the suppression of Serbian nationalism, with its attendant claims on Krajina and the Serbian-populated areas of Bosnia-Herzegovina, as long as Yugoslavia's integrity was not threatened.

102. 

Snyder argues that elite adaptability and the level of socioeconomic development explain this variation, or, more specifically, the variation between “civic” and “ethnic” nationalism. This variation, in turn, explains why some transitions were peaceful, whereas others were violent. The problem is that Snyder's independent variables have fairly similar values across the postcommunist world.

103. 

Cornell, Autonomy and Conflict; and Svante E. Cornell, “Autonomy as a Source of Conflict: Caucasian Conflicts in Theoretical Perspective,” World Politics, Vol. 54, No. 2 (January 2002), pp. 245–276.

104. 

Some of the most prominent studies are Rogers Brubaker, Nationalism Reframed: Nationalism and the National Question in the New Europe (Cambridge: Cambridge University Press, 1996); Valerie Bunce, Subversive Institutions: The Design and the Destruction of Socialism and the State (Cambridge: Cambridge University Press, 1998); and Philip G. Roeder, Where Nation-States Come From: Institutional Change in the Age of Nationalism (Princeton, N.J.: Princeton University Press, 2007). For a review and critique of this literature, see Arman Grigoryan, “Ethnofederalism, Separatism, and Conflict: What Have We Learned from the Soviet and Yugoslav Experiences?” International Political Science Review, Vol. 33, No. 5 (November 2012), pp. 520–538.

105. 

Cornell, “Autonomy as a Source of Conflict,” p. 254.

106. 

Ibid., p. 271.

107. 

Ibid., pp. 269–273.