Symbolic Politics or Rational Choice? Testing Theories of Extreme Ethnic Violence Author(s): Stuart J. Kaufman Source: International Security , Spring, 2006, Vol. 30, No. 4 (Spring, 2006), pp. 45-86 Published by: The MIT Press Stable URL: https://www.jstor.org/stable/4137529 REFERENCES Linked references are available on JSTOR for this article: https://www.jstor.org/stable/4137529?seq=1&cid=pdf- reference#references_tab_contents You may need to log in to JSTOR to access the linked references. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at https://about.jstor.org/terms The MIT Press is collaborating with JSTOR to digitize, preserve and extend access to International Security This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? Testing Theories of Extreme Ethnic Violence Stuart J. Kaufman Understanding the roots of extreme ethnic violence has long bedeviled students of politics. What causes some ethnic groups to engage in bitter wars or even to commit genocide, whereas others in apparently similar circumstances do not? The issue remains a key one in international security affairs: of twenty major armed conflicts ongoing in 2005, for example, one study identifies fifteen as ethnic or communal.' Additionally, ethnic conflict has been a major source of terrorism: many prominent terrorist groups, from the Liberation Tigers of Tamil Eelam to Palestinian Islamic Jihad and the Irish Republican Army, are motivated by ethnic concerns. Efforts to explain such violence have generated a huge literature. Early attempts tended to assume the presence of mutual hostility among contending ethnic groups in what eventually came to be labeled "ancient hatreds" argu ments.2 Later study disproved this view, showing that attitudes and even the identities in question have varied widely over time. The theory of "conflictual modernization," for example, represented an advance, but it failed to explain why modernization leads to more violent ethnic conflicts in some times and places than others.3 The most popular approach today is based on rational Stuart J. Kaufman is Professor of Political Science and International Relations at the University of Delaware. The author thanks Aaron Karnell, Chaim Kaufmann, Marc Ross, Donald Rothchild, and Crawford Young for their insightful comments. 1. Monty G. Marshall and Ted Robert Gurr, Peace and Conflict, 2005: A Global Survey of Armed Conflicts, Self-Determination Movements, and Democracy (College Park: Center for International Development and Conflict Management, University of Maryland, May 2005). 2. See, for example, J.S. Furnivall, Colonial Policy and Practice: A Comparative Study of Burma and Netherlands India (Cambridge: Cambridge University Press, 1948); and M.G. Smith, The Plural Society in the British West Indies (Berkeley: University of California Press, 1965). The related "primordialist theory" was first articulated in Clifford Geertz, "The Integrative Revolution," in Geertz, ed., Old Societies and New States: The Quest for Modernity in Asia and Africa (New York: Free Press, 1963). See also Harold R. Isaacs, The Idols of the Tribe: Group Identity and Political Change (New York: Harper and Row, 1975). A more recent "ancient hatreds" view appears in Robert D. Kaplan, Balkan Ghosts: A Journey through History (New York: Random House, 1993). For a rational choice account using similar assumptions, see Alvin Rabushka and Kenneth A. Shepsle, Politics in Plural Societies: A Theory of Democratic Instability (Columbus: Charles E. Merrill, 1972). 3. For an early observation that identities in conflicts change over time, see Crawford Young, The Politics of Cultural Pluralism (Madison: University of Wisconsin Press, 1976). The now-consensus constructivist view appears in Crawford Young, "The Dialectics of Cultural Pluralism: Concept and Reality," in Young, ed., The Rising Tide of Cultural Pluralism: The Nation-State at Bay? (Madison: University of Wisconsin Press, 1993), which cites works such as Leroy Vail, ed., The Creation of Tribalism in Southern Africa (Berkeley: University of California Press, 1989). For a critique of the logic of the ancient hatreds account of ethnic war, see Stuart J. Kaufman, Modern Hatreds: The Symbolic PoliInternational Security, Vol. 30, No. 4 (Spring 2006), pp. 45-86 ? 2006 by the President and Fellows of Harvard College and the Massachusetts Institute of Technology. 45 This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 46 choice theory, which incorporates some realist ideas of the security dilemma into an explanation that insists on the rationality even of such extreme behavior as genocide.4 Opposing this approach is the social-psychological school, which asserts the critical importance of intangible concerns such as status and of emotional motives in explaining ethnic violence.5 Most of these works address extreme ethnic violence-that is, ethnic war and genocide-only in passing, but a few do offer complete theories to explain these outcomes. The majority agree on defining a war as ethnic "if the contending actors or parties identify themselves or one another using ethnic criteria."6 A leading rational choice theory proposed by David Lake and Donald Rothchild builds on James Fearon's understanding of the security dilemma to assert that ethnic war occurs primarily because information failures and commitment problems prevent competing groups from reaching a negotiated bargain that all would prefer.7 Rui de Figueiredo and Barry Weingast propose an alternative model that purports to explain not only ethnic war but also genocide.8 They argue that predatory elites are the key cause of ethnic war and tics of Ethnic War (Ithaca, N.Y.: Cornell University Press, 2001), pp. 3-5. The seminal work on conflictual modernization is collected in Walker Connor, Ethnonationalism: The Quest for Understanding (Princeton, N.J.: Princeton University Press, 1994). For a critique, see Donald Horowitz, Ethnic Groups in Conflict (Berkeley: University of California Press, 1985), pp. 99-104. 4. For a prominent early example of this approach, see Robert H. Bates, "Modernization, Ethnic Competition, and the Rationality of Politics in Contemporary Africa," in Donald Rothchild and Victor Olorunsola, eds., State versus Ethnic Claims: African Policy Dilemmas (Boulder, Colo.: Westview, 1983). More recent examples include James D. Fearon and David D. Laitin, "Explaining Interethnic Cooperation," American Political Science Review, Vol. 90, No. 4 (December 1996), pp. 715-735; Russell Hardin, One for All: The Logic of Group Conflict (Princeton, N.J.: Princeton University Press, 1995); and David A. Lake and Donald Rothchild, eds., The International Spread of Ethnic Conflict: Fear, Diffusion, and Escalation (Princeton, N.J.: Princeton University Press, 1998). The seminal application of security dilemma theory is Barry R. Posen, "The Security Dilemma and Ethnic Conflict," in Michael E. Brown, ed., Ethnic Conflict and International Security (Princeton, N.J.: Princeton University Press, 1993), pp. 102-124. 5. See, for example, Horowitz, Ethnic Groups in Conflict; and Roger D. Petersen, Understanding Ethnic War: Fear, Hatred, Resentment in Twentieth-Century Eastern Europe (Cambridge: Cambridge University Press, 2002). 6. The definition is Rodolfo Stavenhagen's, cited in Ted Robert Gurr, Peoples versus States: Minorities at Risk in the New Century (Washington, D.C.: United States Institute of Peace, 2000), p. 65. Criticisms of the usefulness of the ethnic war concept include John Mueller, "The Banality of 'Ethnic War,"' International Security, Vol. 25, No. 1 (Summer 2000), pp. 42-70; and Charles King, "The Myth of Ethnic Warfare," Foreign Affairs, Vol. 80, No. 6 (November-December 2001), pp. 165-172. 7. Lake and Rothchild, The International Spread of Ethnic Conflict; David A. Lake and Donald Rothchild, "Containing Fear: The Origins and Management of Ethnic Conflict," International Security, Vol. 21, No. 2 (Fall 1996), pp. 41-75; and James D. Fearon, "Rationalist Explanations for War," International Organization, Vol. 49, No. 3 (Summer 1995), pp. 379-414. 8. Rui J.P. de Figueiredo and Barry Weingast, "The Rationality of Fear: Political Opportunism and Ethnic Conflict," in Barbara F Walter and Jack Snyder, eds., Civil Wars, Insecurity, and Intervention (New York: Columbia University Press, 1999), pp. 262-263. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? 147 genocide, because they provoke violence as a way of maintaining power and misleading their supporters into thinking the other side is to blame for the violence. Among social-psychological theories, the main contender is my symbolic politics theory.9 According to this model, the critical causes of extreme ethnic violence are group myths that justify hostility, fears of group extinction, and a symbolic politics of chauvinist mobilization. The hostile myths, in this view, produce emotion-laden symbols that make mass hostility easy for chauvinist elites to provoke and make extremist policies popular. Rationalists disagree: in a prominent review article, Fearon and David Laitin assert that myths or discourses cannot explain the different outcomes of ethnic politics.10 This dispute is important not only for scholars' understanding of why extreme ethnic violence occurs but also for what to do about it. If the LakeRothchild pure-uncertainty model is right, the best policy for ending ethnic war is one based on the liberal rationalist idea of third-party monitoring and enforcement to ameliorate the commitment problem."1 If the de FigueiredoWeingast elite-predation model is accurate, the correct prescription is to insist on leadership change. If, however, the symbolic politics theory is accurate, the pivotal problem is the tougher one of changing the intragroup symbolic politics that makes predatory elites and policies popular. This article presents a head-to-head test of the rationalist and symbolist theories of ethnic war by reexamining two of the cases-Sudan and Rwandadiscussed by Fearon and Laitin. Rwanda is, additionally, one of the cases de Figueiredo and Weingast use to develop their model. The questions are these: Why did Sudanese President Jaafar al-Nimeiri, who had signed a peace agreement in 1972, abrogate that agreement in 1983 and restart Sudan's north-south civil war? And why did hard-liners in Rwanda resort to war and genocide in 1994 in the aftermath of President Juvenal Habyarimana's death? A focus on extreme cases such as these is useful for comparing theories because all-out wars are a critical test of theories of ethnic violence: explaining ethnic peace is easier than explaining ethnic war. My findings are clear: both rationalist models fail in both cases. The pure- 9. Kaufman, Modern Hatreds. See also Tamara Cofman Wittes, "Symbols and Security in Ethnic Conflict: Confidence-Building in the Palestinian-Israeli Peace Process, 1993-95," Ph.D. dissertation, Georgetown University, 2000. 10. James D. Fearon and David D. Laitin, "Violence and the Social Construction of Ethnic Identity," International Organization, Vol. 54, No. 4 (Autumn 2000), p. 864. 11. Barbara F. Walter, "The Critical Barrier to Civil War Settlement," International Organization, Vol. 51, No. 3 (Spring 1997), pp. 335-364; and Barbara E Walter, Committing to Peace: The Successful Settlement of Civil Wars (Princeton, N.J.: Princeton University Press, 2002). This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 148 uncertainty model is incorrect because uncertainties such as information failures and commitment problems were irrelevant in the Sudan case and are insufficient to explain the Rwanda case. The elite-predation model rightly assumes that both conflicts were the result of elite predation-not uncertaintybut it identifies the wrong mechanism. In neither case was the predatory strategy the best option for leaders seeking to maintain power; in fact, in both cases their violent strategies resulted, predictably, in their loss of power. And in neither case was mass uncertainty about these leaders' violent intentions an important factor in explaining their popular support. The lone insight provided by the rationalist approaches is that the timing of the violence in both cases can be explained in part by shifts in leaders' incentives caused by economic downturns. The symbolic politics theory is much more successful in explaining the extreme violence that occurred in Sudan and Rwanda. In both cases, group fears and myths that justified hostility were strong, leading to powerfully hostile mass attitudes. These attitudes created a context for leaders in which predatory policy was more popular than moderate policy; and the hostile narratives provided a symbolic vocabulary that the leaders used as tools to mobilize support. In the Sudan case, Nimeiri was forced to form a coalition with his strongest rivals-who soon replaced him-because their aggressive policy was more popular than his previous peaceful one. In the Rwanda case, the extremists' genocidal strategy is explicable only in terms of their devotion to an ideology that led them to prioritize genocide over their own survival in power. Furthermore, symbolist theory explains a range of phenomena that rationalist theory cannot-why the contending parties were defined as they were, why some ideas were more popular than others, and why politicians put so many resources into making symbolic appeals empty of real information. Although no two cases can provide a definitive test of alternative theories, rationalist theory emerges from this test significantly weakened, and symbolist theory significantly strengthened, as explanations for extreme ethnic violence. The rest of the article proceeds as follows. The first section specifies the assumptions and logic of each of the three models of ethnic war that are to be tested. The next section explains the process-tracing method being used and specifies the causal chain that must have occurred if each theory is correct. This is followed by two case studies, first Sudan and then Rwanda. Each case section begins with a brief narrative of the events under study, then presents the evidence for and against each theory, including whether the hypothesized process occurred in each case. The conclusion summarizes the results and considers the implications. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? 149 Rational Choice and Symbolic Politics Models of Extreme Ethnic Violence Rational choice and symbolic politics theories differ in their assumptions about the nature of ethnicity and the nature of ethnic conflict, as well as about the causes of ethnic war. Additionally, rational choice theorists differ on some issues, offering two contradictory models of extreme ethnic violence. Below are summaries of the assumptions and logic of each model. RATIONALIST MODELS OF ETHNIC WAR AND GENOCIDE Rational choice theory assumes that ethnic conflict, like all is the result of individuals' rational pursuit of univers wealth, power, and security. Rationalist logic is based on t understanding of what ethnic identity is: "The instrum understands ethnicity as a tool used by individuals, group some larger, typically material end.... [E]thnicity is prima used for political advantage."12 Ethnic war, in this unders lectively suboptimal) result of the rational pursuit of i self-interest. Two different models explain how this proc places the security dilemma at the heart of the process by tention leads to war; the other emphasizes elite motivation THE PURE-UNCERTAINTY SECURITY DILEMMA MODEL. Th model builds on Fearon's pioneering work about internatio argues that even if neither side in a conflict wants war, un rity can cause it to erupt. One reason is informatio Rothchild argue that because ethnic groups are uncertain ent intentions, they may overestimate the adversary's conflict in response. They are also uncertain about likely conflict, so the likely loser may not know to avoid catastr early.'3 Another problem is credible commitment: even if good faith, they may not be able to credibly commit to im future. Finally, incentives for preemptive military action m dilemma, spurring violence neither side may want. In this "risk-aversion is enough to motivate murderous violence." 12. Lake and Rothchild, The International Spread of Ethnic Conflict, pp. 5 13. Lake and Rothchild, "Containing Fear"; and Posen, "The Securit Conflict." 14. Hardin, One for All. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 150 THE ELITE-PREDATION MODEL. De Figueiredo and Weingast disagree with this conclusion: "The security dilemma is not sufficient in and of itself to explain the conditions under which hawkish leaders or subgroups succeed in garnering the support of an often reticent public audience that typically prefers peace to violence."15 They propose instead an elite-predation model, which assumes that masses do not want violence but elites do, adding to the pure-uncertainty logic two more elements: predatory elites and mass uncertainty. Leaders who fear losing power, they argue, may "gamble for resurrection" by resorting to predation-provoking ethnic conflict to try to change the agenda toward issues that favor their remaining in power. Mass publics notice the violence, so even if they are unsure about which side provoked it, they can rationally increase their concern that the other group might be dangerous. The public may therefore rationally support policies leading to war or even genocide, calculating that the costs of violence are lower than the costs of facing threatened violence unprepared. THE SYMBOLIST MODEL OF ETHNIC WAR AND GENOCIDE Instead of using economists' assumptions about the nature human motivation, symbolic politics theory bases its assu findings of social science on both issues. Because this theory is to most readers, I explain it in more detail. THE NATURE OF ETHNICITY. The symbolist approach recogn identity is more than a social category manipulated by eli findings of Anthony Smith, symbolist theory notes that eac defined by a "myth-symbol complex" that identifies which ele culture and what interpretation of history bind the group tog guish it from others.16 These definitions of identity are alway example, in some places (e.g., Ireland and Bosnia), myths divid ligious tradition into different nations, whereas in other cases shared language and presumed common descent trump religio This constructivist understanding of ethnicity notes tha myth-symbol complex is malleable to some degree, it typical in history and culture that cannot be easily ignored. The cultu who promotes a given interpretation of ethnic identity is ty same person as the politician who exploits it; politicians' abili 15. De Figueiredo and Weingast, "The Rationality of Fear," pp. 262-263. 16. Anthony D. Smith, The Ethnic Origins of Nations (New York: Blackw See also Anthony D. Smith, Myths and Memories of the Nation (Oxford: Oxf 1999). This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? ]51 ity instrumentally is, therefore, limited by the cultural context in which they operate. For example, the Serbian myths and symbols related to Kosovo were popularized by the nineteenth-century linguist Vuk Karadzid and promoted by nineteenth-century Serbian governments. Slobodan Milo'evik did not create them: he used them but was also constrained by them.17 Although all group mythologies are likely to display some level of ethnocentrism, myths do vary in the degree to which they portray others as the group enemy. For example, textbooks in France and Germany portrayed the other country as its "hereditary enemy" before World War II, but did not do so after the 1950s.'8 This variation generates symbolist theory's first key hypothesis: the more a group's myth-symbol complex focuses group hostility on a particular adversary, the greater the probability of a violent clash with that adversary, and the greater the likely intensity of the violence. I operationalize myths justifying hostility as those that explicitly identify the other group as an enemy, as inferior, or both. THE SYMBOLIST THEORY OF CHOICE. Symbolic politics theory builds on the findings of neuroscience, which show that emotions, not rational calculations, motivate people to act. One study, for example, found that in a gambling game, individuals with brain damage who did not feel the emotional costs of losses learned to understand the game, but did not adjust their behavior to play better.'19 "Physiologically," William Long and Peter Brecke observe, "emotions are dispositions to action."20 Emotions also help people set priorities among competing goals: fear, for example, causes people to prioritize security over other values such as wealth. Studies have shown that attitudes that originally formed emotionally are more responsive to emotional than to intellectual appeals.21 17. Kaufman, Modern Hatreds, pp. 170-173. 18. Alice Ackermann, "Reconciliation as a Peace-Building Process in Postwar Europe: The FrancoGerman Case," Peace and Change, Vol. 19, No. 3 (July 1994), pp. 229-250. 19. Cited in George E. Marcus, W. Russell Neuman, and Michael MacKuen, Affective Intelligence and Political Judgment (Chicago: University of Chicago Press, 2000), especially p. 32. See also Irving L. Janis and Leon Mann, Decision Making: A Psychological Analysis of Conflict, Choice, and Commitment (New York: Free Press, 1977), especially pp. 7-17. The most sophisticated application of psychological theory to the problem of ethnic violence is Petersen, Understanding Ethnic War. The standard work on emotion is Nico H. Frijda, The Emotions (Cambridge: Cambridge University Press, 1986). Also incorporating recent findings in neuroscience are William J. Long and Peter Brecke, War and Reconciliation: Reason and Emotion in Conflict Resolution (Cambridge, Mass.: MIT Press, 2003). See also Neta C. Crawford, "The Passion of World Politics: Propositions on Emotion and Emotional Relationships," International Security, Vol. 24, No. 4 (Spring 2000), pp. 116-156. 20. Long and Brecke, War and Reconciliation, p. 124. 21. Kari Edwards, "The Interplay of Affect and Cognition in Attitude Formation and Change," Journal of Personality and Social Psychology, Vol. 59, No. 2 (August 1990), pp. 202-216; and Kari Edwards and William von Hippel, "Hearts and Minds: The Priority of Affective versus Cognitive This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 152 The symbolic politics argument asserts that ordinary people choose emotionally among competing values and leaders by responding to the most evocative symbol presented to them.22 Symbols are powerful because they simultaneously refer to an interest and to an emotionally laden myth, often framing a conflict of interest as a struggle against hostile, evil, or subhuman forces. Ethnic or national symbols are immensely powerful in this context, enabling a politician to reinterpret a conflict of interest as a struggle for security, status, and the future of the group. Using these symbols to evoke emotions such as resentment, fear, and hatred is how politicians motivate supporters to act. Politicians with logical arguments are often at a severe disadvantage when competing with such emotive symbolic appeals. For example, Armenian nationalist orators drew a million Armenians to rallies on the Karabakh issue in 1988-in spite of a hostile media and government-by tapping national pride, referring to national symbols such as Mount Ararat, and directing hostility toward Azerbaijani Turks by referring to the 1915 genocide. Incumbent leaders urged caution to avoid violence and emphasized the costs of ethnic confrontation. As a result, the incumbents were replaced by nationalists riding on a wave of popular support for the "sacred cause" who quickly provoked a war.23 THE NATURE OF ETHNIC CONFLICT. According to symbolic politics theory, individual or group interests are not the main consideration in ethnic conflict. Rather, as Donald Horowitz shows, "the sources of ethnic conflict reside, above all, in the struggle for relative group worth."24 In Yugoslavia, for example, voters responded to the emotional appeal of nationalist firebrands such as Milo'evi6 and Franjo Tudjman instead of the interest-based appeals of economic reformers supporting the popular Prime Minister Ante Markovi6. The result was a war in the interests of almost no one, but comprehensible if understood as a competition for status. "Why should I be a minority in your state," a Croat is said to have asked a local Serb, "if you can be a minority in mine?" PRECONDITIONS FOR ETHNIC WAR AND GENOCIDE. A group mythology that justifies hostility is a precondition for violent ethnic conflict; it must exist before a politician can manipulate it. Such myths are not, however, a sufficient Factors in Person Perception," Personality and Social Psychology Bulletin, Vol. 21, No. 10 (October 1995), pp. 996-1011. 22. Murray Edelman, Politics as Symbolic Action: Mass Arousal and Quiescence (New York: Academic Press, 1971); Kaufman, Modern Hatreds; Marc Howard Ross, "Psychocultural Interpretations and Dramas: Identity Dynamics in Ethnic Conflict," Political Psychology, Vol. 22, No. 1 (Spring 2001), pp. 157-178; and Marc Howard Ross, The Culture of Conflict: Interpretations and Interests in Comparative Perspective (New Haven, Conn.: Yale University Press, 1993). 23. Kaufman, Modern Hatreds, pp. 60-62. 24. Horowitz, Ethnic Groups in Conflict, p. 143. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? 153 explanation for violence: hostile myths are common, but ethnic violence is rare. The centrality of emotion in symbolist theory suggests a seco precondition: ethnic appeals are successful in producing extreme if the group also fears that its existence is threatened. In some Israelis, there may be a literal fear of genocide. In other cases, threats can be exaggerated so they seem to be existential ones. E the threat is often part of the point: what matters is not a sober mographic or military balances, but the ability to evoke the vis "our group is in danger." Because the fear is subjective, a probab standing of its effect is appropriate: the greater a group's fear of greater the likelihood, and likely intensity, of violence against A third precondition for ethnic war or genocide is political which consists of two elements. First, there must be enough (whether the result of political freedom, state breakdown, or f to mobilize without facing effective repression; access to state i viously increases the opportunity to act. Second, as Monica Toft tail, is a territorial base: ethnic rebels cannot mobilize unless th territorially concentrated in some region or have a territorial b boring country.25 Again, a probabilistic formulation is appropr ing equal, the greater the opportunity for both sides in an ethn mobilize for violence, the greater the probability of war between Genocide requires opportunity only on the side of the perpetrat PROCESSES LEADING TO ETHNIC WAR. Symbolic politics leads or genocide through a process involving three dynamics-mass h vinist political mobilization, and a security dilemma-that intera spiral of escalation, if the preconditions are present.26 Two causa tinguished, the more common of which, and the one relevan Rwanda, is the elite-led path. In elite-led mobilization, incum leaders appeal to symbols that cast another group as an enemy s hostile emotions, and they provoke violence to stoke ethnic fear background of hostility and fear, predatory policies-those ai domination over rival groups-become popular. This is the ope ure of mass hostility: evidence that predatory policies are popul of the process can be found in Serbia, where Miloievik used the the battle of Kosovo Field in 1389 both to conflate Serbs' modern Albanian ri- 25. Monica Duffy Toft, The Geography of Ethnic Violence: Identity, Interests, and Indivisibility of Terri tory (Princeton, N.J.: Princeton University Press, 2003). 26. Kaufman, Modern Hatreds, chap. 2. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 154 vals with their historical Ottoman enemies (promoting feelings of hostility and fear), and to justify on mytho-historical grounds his policy of domination over the Kosovo Albanians. Such a policy creates a threat to other groups, increasing fear and hostility and leading to the emergence of symbol-manipulating extremist leaders among other groups. The result is a spiral of hostility and ethnic war. Symbolist theory therefore specifies the many filters between ethnic myths and the outbreak of war or genocide. The myths must justify hostility toward the out-group and be strengthened by existential fear; the groups must have the opportunity to mobilize; and hostile attitudes must grow strong. Chauvinist mobilization is also necessary: even given these other factors, extreme violence can be avoided if ethnic elites skillfully define group needs in moderate ways and collaborate across group lines to prevent violence: this is consociationalism.27 Extreme violence occurs only if all of these factors work to promote it. TYPES OF ETHNIC SECURITY DILEMMA. Symbolic politics theory argues that the spiral of escalating conflict in these cases is a sort of security dilemma. A conflict driven only by predatory motives is not a security dilemma, but as Jack Snyder and Robert Jervis note, often "the security fears of the parties to civil conflict [are] intertwined with their predatory goals." Snyder and Jervis therefore distinguish two different security dilemma patterns: "In some cases predatory motives may be the primary cause, and security fears derivative.... Conversely, in some cases security fears may be primary."28 Theoretically, the difference is whether the analyst understands the "predatory" goals as security driven or simply "greedy."29 If expansionist goals are motivated at least in part by security fears, then there is a security dilemma. The symbolist argument is that in cases of ethnic civil war, "greed" and "security motives" are largely indistinguishable. Chauvinist leaders always claim to be driven by security motives, but what makes them chauvinists is that they define their group's security as requiring dominance over rival groups-which is, naturally, threatening to the others. If two or more competing groups feel they need dominance over the same territory, the result is a security dilemma: neither group feels secure unless its status needs are met, but both sets of de- 27. Arend Lijphart, Democracy in Plural Societies: A Comparative Exploration (New Haven, Conn.: Yale University Press, 1977). 28. Jack Snyder and Robert Jervis, "Civil War and the Security Dilemma," in Walter and Snyder, Civil Wars, Insecurity, and Intervention, pp. 16, 21. 29. This terminology is suggested in Charles L. Glaser, "Political Consequences of Military Strategy: Expanding and Refining the Spiral and Deterrence Models," World Politics, Vol. 44, No. 4 (July 1992), pp. 497-538. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? [55 mands cannot be satisfied at once.30 The situation is therefore best understood as involving both predation (some degree of "greed") and a security dilemma (in which moves on each side motivated partly by insecurity create insecurity on the other side). Method and Case Selection This article tests three models of ethnic war and genocide by tracing the processes that led to ethnic wars in Sudan and Rwanda, and to genocide in Rwanda. As Alexander George and Andrew Bennett argue, process tracing goes beyond looking for observable implications of theory.31 The analysis focuses on asking whether the sequence of historical events in these two cases is compatible with the causal path posited by the theories. Two main questions are asked. First, is each theory accurate-do the processes posited by each theory occur in the cases, and do they plausibly account for the outcome? Second, which theory or model explains the most about each case? The primary criterion for case selection is unambiguous compatibility with the scope conditions of both theories. To ensure that the tests are fair to the rationalist models, I choose cases that rationalist theorists explicitly claim to explain. Fearon and Laitin assert that the rationalist approach better explains both the Sudan and Rwanda cases and reject arguments based on mythsymbol complexes. "Discursive or cultural systems at best create a disposition for large-scale violence," they argue, but "this discourse is too widespread to explain variation in levels of violence across cases."32 De Figueiredo and Weingast proposed the elite-predation model specifically to explain ethnic war and genocide in the Rwandan case. These are therefore critical cases for rationalist theory: if rationalists cannot explain their own favorite cases, the empirical basis for their theories is significantly infirmed. At the same time, because the symbolic politics theory claims to be a general theory that explains ethnic war and atrocities within such wars, it should also be able to explain the outbreak of both wars and the Rwandan genocide. The two cases together provide a strong test of both theories because Rwanda represents a relatively easy test and Sudan a tougher test. A theory 30. This argument parallels the one in international security theory regarding the role of predatory motives in driving the interstate security dilemma. See, for example, Randall L. Schweller, "Bandwagoning for Profit: Bringing the Revisionist State Back In," International Security, Vol. 19, No. 1 (Summer 1994), pp. 72-107. 31. Alexander L. George and Andrew Bennett, Case Studies and Theory Development in the Social Sciences (Cambridge, Mass.: MIT Press, 2005), pp. 13, 205-232. 32. Fearon and Laitin, "Violence and the Social Construction of Ethnic Identity," pp. 863-864. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 156 that passes both tests is therefore strongly supported; a theory that fails both tests is severely weakened. The Rwandan case is easy because all of the causal factors identified by both theories strongly tended toward the renewal of severe violence in 1994. If the factors identified by any model turn out not to have been important, the model is badly undermined. The Sudan case, in contrast, is more difficult because key causal factors identified by each theory are questionable. For the pure-uncertainty model, a peace settlement that had already lasted a decade created a context in which commitment problems were not obviously important. Similarly, for the symbolist and elite-predation models, the leader who signed a peace deal in 1972 was an unlikely candidate for predatory leader a decade later. The more plausible any model's explanation for that case, the more strongly supported that theory will be. Although this case selection does involve selecting on the dependent variable-both cases are examples of extreme ethnic violence-that selection is appropriate for this study. As George and Bennett argue, such case selection is not problematic when a case poses a tough test for theory, as the Sudan case does.33 Furthermore, in a head-to-head test of different models, the key issue is the relative ability of each to explain the outcome of interest-severe ethnic violence. The models' ability to distinguish peaceful from violent cases is a relevant question, but a different one. Also, ethnic peace is the empirical norm, as Fearon and Laitin show in another study,34 so any plausible theory has to set stringent conditions for the outbreak of violence, as the theories considered here do. The result is that peace tends to be theoretically overdetermined, so it is hard to identify tough tests involving cases of ethnic peace that "should" have been wars. Furthermore, the one case of relative ethnic peace considered by Fearon and Laitin-northern India-does not represent a tough test for symbolist theory because the Muslims, as a dispersed minority, clearly lack a territorial base, and thus opportunity (except in Kashmir where war did erupt).35 The challenge in the current study is to specify the causal chain posited by each model, and so identify the factors that must be present if each theory is to explain the key dependent variables: outbreak of ethnic war and occurrence of genocide. 33. George and Bennett, Case Studies and Theory Development in the Social Sciences, p. 76. 34. Fearon and Laitin, "Explaining Interethnic Cooperation," pp. 715-735. 35. Territorial dispersion of Muslims in India is coded in Ted Robert Gurr, Minorities at Risk: A Global Viezo of Ethnopolitical Conflicts (Washington, D.C.: United States Institute of Peace, 1993), p. 330. Of the other three cases Fearon and Laitin consider, Yugoslavia is omitted here because a symbolist account of it is included in Kaufman, Modern Hatreds. The other two, Sri Lanka and Northern Ireland, are worthy of future examination. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? 57 RATIONAL CHOICE THEORY This study tests two different rationalist models of ethnic cified separately. PURE-UNCERTAINTY MODEL. The pure-uncertainty secur posits the following logical chain: R1. Leaders recognize the costs of war and so prefer a ne Their efforts to avert war, however, founder on one or lowing obstacles: R2. Information failures cause ethnic leaders to overestimate their rivals' hostility. R3. Information failures result in ethnic leaders miscalculating the likely outcome of fighting. R4. Problems of credible commitment obstruct leaders' efforts to avert violent conflict. R5. War begins when one side launches a preemptive attack in pursuit of military advantage. If this model is correct, there should be evidence that R1 is true, and that some combination of R2-R5 caused the outbreak of war in each case. ELITE-PREDATION MODEL. The model proposed by de Figueiredo and Weingast posits a different causal logic, which goes as follows: R6: Predatory elites choose ethnic war or genocide as a strategy to maintain power. A related point, logically required by this account but made more clearly by Fearon and Laitin, is that the leaders' incentive structures must disfavor alternative strategies, therefore: R7. The violent strategy is the best one available for maintaining power, in part because less violent options are obstructed (e.g., economic difficulties may impede efforts to buy support). R8. Mass publics support their leaders' aggressive policy because they fear possible attack by the other group. R9. Mass publics (which prefer peace) are uncertain about the sources of violence, attributing the violence to the other side even if their own leaders are to blame. These propositions also define a proposed causal mechanism. For the model to be correct, all four propositions must be true for each case of ethnic war or genocide. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 158 SYMBOLIC POLITICS THEORY Symbolist theory posits the following causal chain, but firs tions for ethnic war are necessary: S1. Widespread group myths exist on both sides that expli tility toward, or the need to dominate, the ethnic adv S2. Fear of group extinction is strong on both sides at th breaks out. S3. Both sides have a territorial base and the opportunity to mobilize. Next, mobilization for violent conflict is driven by three processes: S4. Extreme mass hostility is expressed in the media and in popular support for the goal of political domination over ethnic rivals (on at least one side) or resistance to such domination. S5. Chauvinist elites use symbolic appeals to group myths, tapping into and promoting fear and mass hostility, to mobilize their groups for conflict. S6. A predation-driven security dilemma arises, in which the growing extremism of the leadership on at least one side results in radicalization of the leadership on the other. These propositions together identify a causal mechanism. For the model to be correct, there must be evidence that in each case of ethnic war, the necessary preconditions not only were present but were causally important in enabling chauvinist mobilization and in causing a predation-driven security dilemma. Additionally, the logic requires that an explicitly genocidal policy must be based on explicitly eliminationist myths. The case studies below test whether the hypotheses derived from these three models accurately explain the outbreak of war in Sudan and war and genocide in Rwanda. In each case, the critical question is not just whether hypothesized factors were present, but whether there is evidence that they were causally important.36 Case 1: Ethnic War in Sudan There have been three major rounds of ethnic violence in Sudan since World War II. The first north-south war began in 1955, a few months before Sudan's 36. Since I lack the language skills of an area specialist, the main evidence used in this study is the judgments of English-speaking area specialists. Wherever possible, this evidence is supplemented by references to primary sources such as leaders' speeches and public documents. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? 159 independence, and ended in a negotiated settlement, the Addis Ababa agreement, in 1972. Then, in 1983 the same Sudanese leader who negotiated the accord, Jaafar al-Nimeiri, abrogated its main provisions, provoking a new round of north-south warfare that continued until a peace agreement was signed in January 2005. A third major conflict erupted in February 2003 in the western region of Darfur. I focus in this analysis on explaining the outbreak of the most violent of these conflicts, the second north-south war that began in 1983 and resulted in the deaths of some 2 million people. Sudan is divided between a Muslim, Arabized north and a Christian and animist black African south; northerners and southerners are in turn broken up into a large array of groups, tribes, and clans. In the north, Muslims are divided into a number of sects, with the Ansar and Khatmiyya Sufi orders the most important. A cross-cutting cleavage is ethnolinguistic: some northern groups are less Arabized than others. This fact was made prominent by the fighting in Darfur, which has generally pitted sedentary Muslim Zaghawa, Fur, and other groups against the Sudanese government and their Arabized (also Muslim) nomadic neighbors organized in so-called Janjaweed militias.37 Additionally, there is an economic cleavage, with the Arabs of the Nile River valley traditionally providing most of the national elite and excluding others.38 Southern groups speak an array of languages and share no common religion: some are Christians, while most retain their traditional spiritual beliefs. The largest of the southern peoples are the Dinka, who have formed the backbone of the rebel groups, and the Nuer, who have typically been divided between the rebel and government camps.39 Neither Dinka nor Nuer have an overarching ethnic leadership, and their subgroups have often fought each other as well as other southern peoples. Thus while Sudan's conflict was an ethnic war, it was also a coalition war, with the Arabs of the north and the Dinkas of the south leading shifting coalitions of ethnic groups and subgroups against each other. The critical questions in explaining the outbreak of fighting in Sudan are: Why did President Nimeiri abrogate the Addis Ababa agreement in 1983, and why did that action provoke a massive rebellion? Nimeiri came to power in a military coup in 1969 in alliance with leftist elements. The Ansar sect at- 37. For details on the people of Darfur, see Alex de Waal, "Who Are the Darfurians?" African Affairs, Vol. 104, No. 415 (April 2005), pp. 181-205. On the conflict, see Scott Anderson, "How Did Darfur Happen?" New York Times Magazine, October 17, 2004, p. 52. 38. For background, see Amir H. Idris, Sudan's Civil War-Slavery, Race, and Formational Identities (Lewiston, Maine: Edwin Mellen, 2001), pp. 5-6. 39. See, for example, Ann Mosely Lesch, The Sudan: Contested National Identities (Bloomington: Indiana University Press, 1999); and Sharon Elaine Hutchinson, Nuer Dilemmas: Coping with Money, War, and the State (Berkeley: University of California Press, 1996). This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 60 tempted to resist and was crushed in 1970; in 1971 the leftists attempted a coup and were defeated. Nimeiri thus found himself short of allies, while he shared a common enemy-the Islamists-with the southerners. He also seems to have concluded that the war in the south was not winnable.40 The stage was set for the 1972 agreement, after which Nimeiri allied himself with a coalition of centrist technocrats, the army, and leaders from the south, which was granted autonomy. In 1976, however, a coalition of Islamist groups launched a coup attempt with Libyan aid. Although Nimeiri defeated the coup, he decided to reconcile with the Islamists, taking Ansar leader Sadiq al-Mahdi into his government. Nimeiri then moved increasingly toward the Islamists and away from the terms of the Addis Ababa agreement. He merged the previously separate secular and sharia courts in 1980, then imposed full Islamic law in September 1983. Starting in 1981 he began trying to break up the southern region into three less autonomous provinces. When the southern legislature resisted, he imposed the division unilaterally in 1983. These actions came in the context of increasing economic distress-a recession leading to strikes, riots, and a punitive recovery plan imposed by the International Monetary Fund-beginning in 1981. Nimeiri also tried to transfer southern army units to the north, in violation of the Addis Ababa agreement. The southerners mutinied, and when Nimeiri ordered northern troops to attack the mutinous southerners in May 1983, the civil war resumed.41 SUDAN: THE SYMBOLIST EXPLANATION Because Sudan's civil war was initiated by the north, a sym begin with the myths and fears underlying northern h south. Other important factors include the myths and fear southern resistance, opportunity, and the processes of chau of ethnic hostility. These hypotheses are assessed below. S1 AND S2: HOSTILE MYTHS AND EXISTENTIAL FEARS IN THE NORTH. The mainstream northern Sudanese mythology asserts an Arab and Muslim identity for its adherents. Northern families cultivate myths of descent from Arab ancestors as a way of claiming membership in the Arab world and a distinction from the "Africans" of the south. The aim is "to enhance pride and selfesteem," which often shades into "an attitude of racial and cultural superior- 40. Donald Rothchild, Managing Ethnic Conflict in Africa: Pressures and Incentives for Cooperation (Washington, D.C.: Brookings, 1997); and Bona Malwal, The Sudan: A Second Challenge to Nationhood (New York: Thornton, 1985), p. 11. 41. Lesch, The Sudan, pp. 49-55. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? 161 ity."42 G.P. Makris sums up the myth as one of an "imagined Arab community ... defin[ed] as the realm of freedom, humanity, Islamic tradition and heroic history ... [which] constructed its exact opposite; the realm of the pagan African slaves."43 Although the idea of Sudanese nationalism is essentially a twentiethcentury invention, some of the underlying ideas and attitudes extend as far back as the seventh century, when "races and religions were [legally] ranked, with Arabs and Muslims respected as free, superior, and a race of slave masters, while Negroes, blacks, and heathens were viewed as . . . [rightfully] slaves."44 The medieval geographer Ibn Khaldun justified this ranking in racist terms, writing: "Negro nations ... have little that is human and possess attributes that are similar to those of dumb animals."45 Even after the racial pecking order was no longer enforced by law, these attitudes remained common among those who adopted the Muslim Arab culture. In 1955, for example, a government report on disturbances in the south stated that southern Sudanese were "one of the most primitive peoples in the world"; a typical northern policeman at that time called southern children "monkeys."46 But the falsity of the racial distinction needs to be emphasized: the "Negro nations" Ibn Khaldun was writing about were mostly ancestors of today's Sudanese northerners, who later adopted the Muslim Arab identity along with the associated racism. Another element of the northern Sudanese nationalist narrative is a proselytizing brand of Islam. While northerners see their Muslim Arab identity as superior, many also view southerners as "eligible for salvation through Islam." Indeed, some northerners regard salvationism as their duty. Ansar sect leader Sadiq al-Mahdi, for example, reportedly remarked that "Islam has a holy mission in Africa and southern Sudan is the beginning of that mission."47 According to the traditional Muslim concept of the dar al-harb, the "abode of war," such a mission should be pursued by force of arms if necessary. The father of Sudanese nationalism, Muhammed Ahmed al-Mahdi, was an Islamic would-be Messiah who in 1885 established Sudan as an independent 42. Francis M. Deng, War of Visions: Conflict of Identities in the Sudan (Washington, D.C.: Brookings, 1995), especially p. 5. 43. Ahmad Alawad Sikainga, "Northern Sudanese Political Parties and the Civil War," in M.W. Daly and Sikainga, eds., Civil War in the Sudan (London: British Academic Press, 1993), pp. 78-96. Makris is quoted in Idris, Sudan's Civil War, p. 43. 44. Deng, War of Visions, pp. 4-5. 45. Quoted in Idris, Sudan's Civil War. 46. Ibid., pp. 15-16, 113. 47. Quoted in Gabriel Warburg, Islam, Sectarianism, and Politics in Sudan since the Mahdiyya (Madison: University of Wisconsin Press, 2003), p. 167. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 162 state with a Muslim fundamentalist government. The Mahdiyya adhered to a violent jihadist fundamentalist ideology, which some northern intellectuals long continued to justify as central to Sudanese national identity. As M.W. Daly puts it, "In retrospect the Mahdiyya assumed the quality of a nationalist myth,... and its ideology of xenophobia and Islamic rectitude became an exclusive heritage of the North."48 There is, however, a distinct feeling of threat and anxiety underlying northern assertions of superiority, in part because other Arabs often dismiss Sudanese as Africans, not Arabs. The fact that many contemporary "Arab" Sudanese really are the descendants of black "African" slaves may help explain the ferocity of northern Sudanese resistance to the label. This again shows the bogus nature of any "racial" or even "ancient" characterization of Sudanese conflicts: the conflicts are between those Sudanese who adopted an "Arab" identity in recent centuries and those who developed competing identities; all would be considered "black" in the United States. Ansar leader Sadiq al-Mahdi has articulated the link between northerners' anxieties and the push for Islamic rule. "This nation will not have its entity [sic] identified and its prestige and pride preserved," he has stated, "except under an Islamic revival."49 This is, in a way, an existential fear: if southerners' idea of an African identity were to become the official one in Sudan, northerners' Arab identity might be extinguished even if they were physically unharmed. Thus, as al-Mahdi said, the northern Sudanese narrative of an Islamic identity, plus perceived threats to that identity, are the key reasons for the popularity of Islamism in northern Sudan-the key reasons, that is, for the northerners' repeated insistence on a predatory policy of imposing Islamic law on the south. S1 AND S2: HOSTILE MYTHS AND EXISTENTIAL FEARS IN THE SOUTH. Southern Sudanese, especially Dinka, myths also identify slavery as a key element in the identities of both sides.50 In the southern myths, however, the south is the land not of slaves but of those who resisted northern slave raids-raids that became more intense after the south came under the rule of the Arab north. In one prominent case, southern intellectuals note, an Arab slave trader gained such power that he was appointed a regional governor by the Turco-Egyptian government. The Mahdist government that followed was even worse from th southern point of view. Slave raids intensified further. Worse, the Mahdists 48. M.W. Daly, "Broken Bridge and Empty Basket: The Political and Economic Background of the Sudanese Civil War," in Daly and Sikainga, Civil War in the Sudan, p. 4. 49. Quoted in Sikainga, "Northern Sudanese Political Parties and the Civil War," p. 83. 50. Deng, War of Visions, pp. 70-75. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? 163 never succeeded in establishing law and order, resulting in a Hobbesian state of violence so severe that Sudan's population declined by more than 50 percent. Dinka tradition vividly remembers the period as equivalent to "the total destruction of the world.""' According to southern mythology, this "shared experience of slave trading and colonialism unified various ethnic groups in the South."52 Fearon and Laitin try to rebut this view, citing Deng's prominent study, which they assert does not "put any stress on a specific cultural discourse creating a disposition toward ethnic violence."53 This is a bizarre misreading, given that Deng's one-word summary of southern identity discourse is "resistance"-that is, armed hostility to the north motivated by deep-seated fears of destruction and enslavement. Deng writes, "The Dinka refer to the Turk-Egyptian and Mahdist periods as the time when 'the world was spoiled,' an abomination of which they speak with consistency and vividness." He reports "extensive interviews" with Dinka chiefs recounting how the Mahdists "destroyed the country," and "captured our people and sold them."54 Furthermore, this experience of extreme violence was within living memory each time the south revolted. For the people of Sudan's south, myths or narratives hostile to the Arabs are fundamental to their identity, and ethnic fears are an important fact of life. S3: OPPORTUNITY. Given the consistent history of northern dominance of Sudanese governments, northerners always have the opportunity to initiate violence against the south: they simply unleash the army. Given the vast size of Sudan's territory and the difficult terrain, the option of guerrilla resistance is equally omnipresent for Sudanese southerners, especially when any of the neighboring states is willing to help. Thus southern Sudanese rebels typically have two sorts of territorial bases: their home territory in southern Sudan, and neighboring rear base areas (e.g., in the 1980s, in Ethiopia).55 In Sudan, lack of opportunity is rarely a constraint on the outbreak of ethnic war. S4 AND 55: MASS HOSTILITY AND CHAUVINIST MOBILIZATION. Given the prominence of the Islamist narrative in Sudanese society, Islamist policies are perennially popular in Sudanese politics. Even British colonial rule was conducted 51. Ibid., p. 70. 52. Idris, Sudan's Civil War, p. 24. See also Warburg, Islam, Sectarianism, and Politics in Sudan since the Mahdiyya, p. 140; and Catherine Jendia, The Sudanese Civil Conflict, 1969-85 (New York: Peter Lang, 2002), pp. 27, 39. 53. Fearon and Laitin, "Violence and the Social Construction of Ethnic Identity," p. 863. 54. Deng, War of Visions, pp. 70, 73. 55. On the role of Ethiopian aid and bases, see Douglas H. Johnson and Gerard Prunier, "The Foundation and Expansion of the Sudan People's Liberation Army," in Daly and Sikainga, Civil War in the Sudan, pp. 117-141. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 164 in northern Sudan largely according to Islamic law; and as early as 1957, the leaders of the two biggest parties in Sudan-the parties dominated by the Ansar and Khatmiyya sects-called for creation of an Islamic republic. When democracy returned to Sudan in the mid-1960s, the same parties reemerged; indeed they had just agreed on a constitution for an Islamic state in 1969 when Nimeiri seized power.56 Talk of an Islamic state was thus a familiar and widely popular symbol in northern Sudan. This was the context in which Nimeiri decided in the late 1970s on reconciliation with the Islamists, accepting Ansar leader Sadiq al-Mahdi and Muslim Brotherhood leader Hassan al-Turabi into the government. Needing a base of support in the north, Nimeiri discovered, as had his secularist predecessors, that "the only ones who could mobilize people beyond the tribal level were the sectarian figures" and other religious leaders.57 Cooperation with the Islamists and picking a fight with the southerners offered the hope of increasing Nimeiri's political support through chauvinist mobilization, rallying the northern population behind the key symbols of Sudanese nationalism-sharia, jihad against the south, and Nimeiri's proclaimed status as an imam. The southerners would obviously fight back, but taking them on helped Nimeiri portray himself as an Islamic hero. Timothy Niblock summarizes Nimeiri's strategy this way: "The political arena . .. became a theatre where the president could arrange spectacles ... to weld together a shifting basis of proclaimed yet insubstantial support." Regarding the Islamist program, Khalid Duran notes: "Sharia, most of which had long since fallen into oblivion, was turned by [the Muslim Brotherhood] into a chimerical remedy of all ills, a kind of Aladdin's wonderlamp." The nature of the resulting conflict, as the title of Deng's book asserts, was a competition of group identities, Islamic versus African: "The issues involved are essentially those of competing identities and the alternative visions they offer the nation."58 Indeed, this is the standard view among experts on Sudan, and one articulated by northern and southern political leaders alike.59 Nimeiri clearly engaged in such symbol manipulation. Deng writes, "[He] began to dress in Arab garb, with all the outward symbols of an Islamic sheikh or imam."60 He publicized his attendance at Friday sermons, requested that his 56. Warburg, Islam, Sectarianism, and Politics in Sudan since the Mahdiyya, pp. 145-148. 57. Sikainga, "Northern Sudanese Political Parties and the Civil War," p. 80. 58. Quoted in Deng, War of Visions, pp. 124, 178, 184. 59. According to southern rebel leader John Garang, speaking in 1986, "Our major problem is that the Sudan ... is still looking for its soul, for its true identity." Quoted in Lesch, The Sudan, pp. 88- 89. Garang was killed in a helicopter crash on July 30, 2005, after the north-south peace agreement had begun to be implemented. 60. Deng, War of Visions, p. 170. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? 165 ministers refrain from drinking alcohol, and published a book in 1980 entitled Why the Islamic Method.61 After sharia was imposed, Nimeiri's speeches also shifted to include Islamist symbolism, stating for example: "The Islamic avenue is the paramount pillar of the comprehensive political program."62 T. Abdou Maliqalim Simone writes, "The excision of limbs for crimes of thievery, flogging for alcohol consumption, and stoning for adultery were imbued with great symbolic value,"63 as progress toward the ideal Islamic society. Evidence shows that this Islamist tilt was probably necessary for any politician seeking support in northern Sudan. As noted above, Islamism was already popular in Sudan in the 1950s and 1960s. Nimeiri suppressed political discourse while he was in power, but the politics of the year following his ouster, which culminated in free multiparty elections in April 1986, is illuminating. The Muslim Brotherhood was the first to organize political rallies after Nimeiri's ouster, attracting 30,000-40,000 people to two rallies four days apart. One fiery speaker shouted, "We will not make any compromise on the application of the sharia and, if need be, we are prepared to die fighting to defend Islam!" A slogan shouted by the crowd was, "No alternative to God's law!" In an interview, Brotherhood leader Turabi threatened: "The street protests which overthrew Nimeiri can also topple anyone who tries to abolish the Sharia."64 In this period, according to one analysis, "the threat to northern cultural identity posed by the SPLA [southern rebel Sudanese People's Liberation Army], including opposition to Sharia demands, swelled the ranks of the NIF [National Islamic Front, led by the Muslim Brotherhood]."65 In this context, leaders of the interim regime and of the leading political parties all chose to defend the codification of sharia; "moderation" was signaled by calls to implement it "in a correct way."66 By the time votes were cast in April 1986, the secularist parties were 61. Abdel Salam Sidahmed, Politics and Islam in Contemporary Sudan (Richmond, U.K.: Curzon, 1997), p. 120. 62. Khartoum Radio SUNA (in English), October 16, 1983, in Foreign Broadcast Information Service (FBIS) Daily Report: Middle East and Africa, October 17, 1983. For another example, see Khartoum Radio SUNA (in Arabic), October 12, 1983, in FBIS, October 13, 1983. 63. T. Abdou Maliqalim Simone, In Whose Image? Political Islam and Urban Practices in Sudan (Chicago: University of Chicago Press, 1994), p. 26. 64. Eric Rouleau, "Muslim Brotherhood Demands Maintenance of 'Islamic Law,"' Le Monde, April 18, 1985, p. 4, in FBIS, April 19, 1985; Turabi's threat was reported by Hong Kong Radio Agence France-Presse, April 15, 1985, in FBIS, April 15, 1985. 65. Abdelwahab el-Affendi, "'Discovering the South': Sudanese Dilemmas for Islam in Africa," African Affairs Vol. 89, No. 35 (July 1990), p. 384. 66. See statements by interim leader Siwar al-Dhahab, Khartoum Radio SUNA (in Arabic), April 17, 1985, in FBIS, April 17, 1985; Sadiq al-Mahdi, Cairo Radio MENA (in Arabic), April 17, 1985, in FBIS, April 18, 1985; and Mohammed al-Mirghani, Cairo Radio al-Musawwar (in Arabic), April 26, 1985, in FBIS, May 2, 1985. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 166 marginalized, and the Islamist parties dominated Sudanese politics. Sadiq alMahdi's "modern" Islamist appeal, combined with his base in Sudan's largest religious sect, earned his Umma Party the biggest share of the vote, gaining it 39 percent of the seats in parliament. Next were Mohammed al-Mirghani's Democratic Unionist Party, based on the Khatmiyya sect, with 24 percent of the seats; and the NIF with 20 percent. Regional parties won the rest, except for the Communist Party's 1 percent take.67 Two conclusions are clear: virtually the only victors were those who made identity appeals; and in most of the north, that meant Islamist, usually sectarian, appeals. Further, it was the extremists who most often controlled the streets: when Mirghani negotiated a tentative peace deal with SPLA leader John Garang in November 1988, for example, the NIF mobilized some 100,000 people to protest Mirghani's agreement to freeze the application of sharia.68 Another illustration of the outsized importance of symbolic politics in the north is the attention given to the revelation that Nimeiri's government had cooperated with Israel in evacuating Ethiopia's Jews, the Falashas, from Ethiopia. The revelation provoked a political firestorm in Sudan. The transitional government launched a legal investigation, and politicians competed with each other to denounce this "pan-Arab treason."69 Considering the complex issues of regime transition facing Sudan at the time, it is striking that so much attention was devoted to this purely symbolic historical issue. Symbolic mobilization, centered on the theme of resistance to northern pretensions to superiority and domination, was important in the south as well. As one southern leader, Peter Adwok Nyaba, phrases it, "The minds and hearts of the people of South Sudan ... will only be stirred into action by fundamental ideas, images and feelings rooted deeply in their past."70 One key symbolic issue was slavery: Sharon Hutchinson quotes a representative Nuer warrior as saying, "The peace of the slave is not worth it." Another complained, "The Arabs called us dogs and slaves and said that we were no better than the dirt under their feet."71 Another key symbolic issue was sharia, as its repeal was repeatedly a major demand of southern politicians.72 Rebel leader Garang fol- 67. Lesch, The Sudan, pp. 42-43, 66-72. 68. Kamal Osman Salih, "The Sudan, 1985-89: The Fading Democracy," Journal of Modern African Studies, Vol. 28, No. 2 (June 1990), p. 218. 69. The phrase is a former foreign minister's, Khartoum Radio SUNA, May 12, 1985, in FBIS, May 13, 1985. See also al-Mahdi's comments, Khartoum Radio SUNA (in Arabic), April 18, 1985, in FBIS, April 19, 1985; and a Mirghani spokesman, Khartoum Radio SUNA, May 8, 1985, in FBIS, May 9, 1985. 70. Peter Adwok Nyaba, The Politics of Liberation in South Sudan: An Insider's View (Kampala: Fountain, 1997), p. 32. 71. Hutchinson, Nuer Dilemmas, pp. 9, 133. 72. See Khartoum Radio SUNA (in Arabic), April 10, 1985, in FBIS, April 11, 1985; and Carolyn This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? 67 lowed a sophisticated strategy, using words such as "liberation," "equality," and "freedom" as code for anti-Arab sentiment in the south, but also as a basis for building a broader coalition for a "New Sudan" in other regions.73 SUDAN: RATIONALIST EXPLANATIONS Rationalist analyses push aside these considerations. The model hypothesizes that a security dilemma in Sudan gr ment problems and uncertainty about competing groups' i predation model assumes that predatory leaders deceive th their policies. These hypotheses are assessed below. R1-R5: A PURE-UNCERTAINTY SECURITY DILEMMA? Fran tion of one north-south conference neatly summarizes the sized by the pure-uncertainty understanding of the securi northerners suspected the southerners of desiring separatio phrasing the conference chairman, "and the southerners su erners of wishing to dominate the south."74 The confe however, occurred in 1947, and its result was to continue p independent, united Sudan. Uncertainty alone did not le The evidence above points to predation rather than uncer factor driving the security dilemma in Sudan in the 1980s the war by attempting to subjugate the south politicall southern autonomous region, imposing sharia, and trying to fer southern military units. Unwilling to surrender to the and to the loss of most political power-and, for the southe of their homes-many southerners felt compelled to def returned to guerrilla war. Given that Nimeiri had been trying for two years to bre mous southern region and had been moving toward sharia is little reason to believe that southerners were uncertain about his intentions by 1983. Neither is there evidence to suggest that they expected a quick victory; Garang's speeches instead implied determination to persevere in a long struggle.75 Similarly, after his experience in the first war, and after a series of small mutinies by southern troops in the years that followed, Nimeiri can have Fluehr-Lobban, "Islamization in Sudan: A Critical Assessment," in John O. Voll, ed., Sudan: State and Society in Crisis (Bloomington: Indiana University Press, 1991), p. 71. 73. John Garang, John Garang Speaks (London: KPI, 1987), pp. 24-27. It is notable that the Koka Dam Declaration (in ibid., p. 145), an abortive agreement between Garang and Khatmiyya leader al-Mirghani, incorporates most of Garang's language and themes. 74. Deng, War of Visions, p. 89. 75. See Garang, John Garang Speaks, p. 27. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 168 been in little doubt about the near certainty of a prolonged, violent southern response. Once his predatory motives became clear, the security dilemma went into operation for the southerners-anything they did to protect themselves merely provoked Nimeiri more. Southern soldiers began refusing orders; Nimeiri sent northern troops to suppress them; and the southerners returned to guerrilla war.76 R2 and R3 are not supported: information failures were not important in causing this war. Nor is there evidence either of a commitment problem (R4) or of military incentives for preemption (R5). The institutional framework established in the 1972 Addis Ababa agreement remained functional until Nimeiri dismantled it. The problem was not uncertainty about Nimeiri's future intentions toward the south; it was certainty about his hostile current intentions. Neither did either side attack preemptively. Rather, Nimeiri moved piecemeal to respond to southern mutinies as they occurred, and rebellious troops responded piecemeal.77 These considerations point to the conclusion that Nimeiri did not prefer a negotiated bargain to war (R1). Rather, he abrogated the existing bargain and knowingly started a war. R6-R9: ELITE PREDATION AS RATIONAL DECEPTION? The arguments above support R6, the proposition that Nimeiri chose to start a war as part of a rational strategy to maintain power. But for the elite-predation model to make sense, ethnic war must have been the best available strategy for Nimeiri (R7) on rationalist grounds. This is more dubious. Nimeiri abandoned his southern allies, who had every reason to want to help him stay in power and had saved him in 1976. Instead he invited his Islamist rivals to return to the country, rebuild their power bases, and share power in the government until they eventually replaced him. He gave the Muslim Brotherhood control of the Islamic banks, which they used to bankroll their rise to power, and gave them free rein to proselytize in the army, ultimately turning it against him. He imposed sharia, though no one had a material interest in it. This does not make sense as a power-conserving strategy. In rationalist terms, maintaining the alliance with the south would have been safer and smarter, while sharia, which offered material benefits to almost no one, should not have been useful in garnering support. Nimeiri's strategy-clearly flawed, since in hindsight it led to his oustermakes sense only if one acknowledges that the pervasive Islamist discourse of the Sudanese north influenced Nimeiri's thinking while presenting him with a 76. Lesch, The Sudan, p. 49. 77. For a detailed account, see Johnson and Prunier, "The Foundation and Expansion of the Sudan People's Liberation Army," pp. 120-125. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? 169 tool for appealing to the mass public. It makes sense, in short, only if one recognizes that he was engaged in symbolic politics. More clearly supported is the proposition that Nimeiri's alternatives were limited in part by economic difficulties. In the early 1980s, as Sudan's economy turned sour, Nimeiri was no longer able to rely on appeals to northerners' tangible interests by promising economic development, so he had to turn further to symbolic appeals to Islam. Thus rationalist theory helps explain why Nimeiri attempted that dangerous expedient when he did. The core propositions of the elite-predation model's mechanism, however, are not supported. Nimeiri's audience in the north did not fear potential attack by southerners (R8). They feared for their identity, not their safety. Relatedly, I found no evidence that northerners supported the war-to the extent they did so-because they were misled into believing the southerners had started it (R9) and that their safety therefore required war. In sum, war was not popular in the north due to misplaced beliefs about safety. Rather, repression of the south-predation-was popular in the north because northern identity narratives prescribed a quest for domination; war was the acceptable result. Case 2: Ethnic War and Genocide in Rwanda Rwanda faced several rounds of ethnic violence in the post-World War II period-though significantly, no specifically Tutsi-Hutu violence before that time. The first round of such violence came in 1959-64, the period of the Rwandan revolution that shifted power from a king and Tutsi elite under Belgian colonial control to an independent government led by the Hutu majority. A new round of ethnic violence, sponsored by the government, erupted in 1972 and ended after a 1973 coup led by Hutu army chief Juvenal Habyarimana. Habyarimana's regime prevented ethnic violence for almost two decades, but in 1990 the Tutsi-led Rwandan Popular Front (RPF), a Uganda-based exile group, invaded Rwanda, inaugurating a new war. The Arusha peace process of 1993-94 yielded a yearlong cease-fire and a settlement deal. But in April 1994 Hutu hard-liners seized control, restarted the war, and slaughtered more than half a million Tutsi civilians before being driven out of Rwanda.78 This analysis concentrates on explaining the fighting and genocide in 1994. 78. The standard work on pre-1970 Rwandan history is Rend Lemarchand, Rwanda and Burundi (New York: Praeger, 1970). Two excellent works on the genocide and after are Gerard Prunier, The Rwanda Crisis: History of a Genocide (New York: Columbia University Press, 1995); and Johan Pottier, Re-imagining Rwanda: Conflict, Survival, and Disinformation in the Late Twentieth Century (Cambridge: Cambridge University Press, 2002). This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 70 Rwanda's population is divided into three main groups: the Hutu majority, the Tutsi minority, and the Twa, who make up 1 percent of the population. All three groups speak the same language, profess the same religious beliefs, and share the same culture in most other respects. There is a stereotypical physical difference, with Hutu seen as typically shorter and stockier, and with broader noses, than the Tutsi. To account for this difference, some theorize that the Tutsis' ancestors migrated to the area from the north some centuries ago, settling among, and coming to dominate, the agricultural Hutu already in the region while adopting the Hutus' language and culture. In the nineteenth century, the Hutu-Tutsi distinction was apparently based primarily on cattle and land ownership: the rich and powerful were codified as Tutsi, most others as Hutu.79 In addition to this ethnic or racial division, regional and clan ties are also important in Rwanda. President Habyarimana was from northern Rwanda, whereas the regime of President Gregoire Kayibanda, which he had replaced, was based in the south; both favored their home region in governing. Relatedly, Habyarimana himself did not have an extensive network of clanbased support for his regime and therefore had to rely heavily on relatives of his wife. RWANDA: THE SYMBOLIST EXPLANATION According to symbolist logic, Rwanda's genocide must ha by an exceptionally hostile, eliminationist Hutu mythology Tutsi. There must also have been extreme ethnic fears, amp treme mass hostility against Tutsi, and chauvinist mob manipulating ethnic symbols-all resulting in a predation-dr lemma. These hypotheses are explored below. st: TUTSI AND HUTU MYTHS. What gives the Hutu-Tutsi dis character is both groups' adoption of narratives claiming an difference between them. This distinction, based on the rac pothesis" proposed by European colonialists, asserted tha tall Tutsi had come from the northeast of Africa and had c and the Twa because of their superior civilization akin to Eur Lemarchand noted in 1970, "Rwanda is unique in the sheer a ditions purporting to show the superiority of the Tutsi over cording to one of these traditional stories, Rwanda's firs 79. Pottier, Re-imagining Rwanda, pp. 12-14, 112. 80. Aimable Twagilimana, The Debris of Ham: Ethnicity, Regionalism, and the (Lanham, Md.: University Press of America, 2003), p. xxiv. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? 71 three sons: Gatwa, Gahutu, and Gatutsi. He decided to test them by assigning them each to watch over a pot of milk for a night. Gatwa drank his milk, and Gahutu spilled his; only Gatutsi carried out his father's instructions. As a result, the story goes, the king made Gatutsi his successor, Gahutu the serf, and Gatwa a pariah.81 An alternative Tutsi myth denied even this asymmetrical kinship, asserting instead that Kigwa's Tutsi ancestors "conquered the lands of the Bahutu by slaying their kings and have thus enslaved the Bahutu."82 These racist ideals were taught in Rwandan schools in the 1920s and 1930s.83 The less savory aspects of Tutsi domination, and its origins, were illustrated by a key symbol of the king's power, and by extension that of the Tutsi: the kalinga, a ceremonial drum decorated with the testicles of defeated Hutu princes.84 For those Tutsi who fled into exile after the Hutu takeover, their Rwandan Tutsi identity became even stronger, as social barriers distinguishing different Tutsi social groups evaporated. Their mythology mellowed into a reinterpretation of precolonial Rwanda as a harmonious society with substantial social mobility between Hutu and Tutsi, into which ethnic divisions were inserted by the colonialists.85 On the Hutu side, leaders of the 1959 revolution did not reject the Tutsi ideology; instead, they "merely inverted its sign." Thus they agreed that the Tutsi were invaders from afar but took that to mean the Tutsi were aliens whose presence was illegitimate. "Traditional" Tutsi rule, in this mythology, was "cruel and ... oppressive tyranny" that had enslaved the Hutu. A group in cluding the future first president of Rwanda, Gr6goire Kayibanda, articulate this narrative in a 1957 document called the "Bahutu Manifesto," saying: "Th problem is basically that of the political monopoly of one race, the Mututsi which condemns the desperate Bahutu to be for ever subaltern workers."86 T the Hutu, the kalinga drum, with its decoration of Hutu testicles, was a partic ularly effective symbol of what Tutsi rule meant: bloody emasculation. Reversing the logic of the "natural" Tutsi right to rule, the new Hutu elite considered its own rule democratic and legitimate because it consisted members of the indigenous majority, based on a "myth of Hutu solidarity."" Soon after the Hutu takeover, the denigration of the Tutsi was expanded int dehumanization: Tutsi guerrilla fighters in the 1960s were labeled "coc 81. Lemarchand, Rwanda and Burundi, pp. 33, 43. 82. Quoted in Twagilimana, The Debris of Ham, p. 65. 83. Mahmood Mamdani, When Victims Become Killers: Colonialism, Nativism, and the Genocide in Rwanda (Princeton, N.J.: Princeton University Press, 2001), p. 90. 84. Prunier, The Rwanda Crisis, p. 47. 85. Pottier, Re-imagining Rwanda, p. 110. 86. Prunier, The Rwanda Crisis, pp. 45, 80. 87. Lemarchand, Rwanda and Burundi, p. 285. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 72 roaches," a term that was revived in the 1990s. Although regime rhetoric moderated after Habyarimana took power, school curricula in the Habyarimana period still "painted the Tutsi as natural enemies of the Hutu.... This systematic preaching of racial ideology served to keep alive racial hatred."88 S2 AND R8: TUTSI AND HUTU FEARS. Tutsi fears of extinction were significant even before the 1990s. The seeds were sown in the extreme violence of the independence period, when some 30,000 Tutsi were killed and more t 300,000 driven into exile.89 Indeed, some Hutu officials were already prom ing acts of genocide in the 1960s: one local prefect is reported to have said: " are expected to defend ourselves. The only way to go about it is to paralyse Tutsi. How? They must be killed."90 The result, in response to Tutsi guerri attacks, was the massacre of some 10,000 Tutsi between December 1963 and January 1964 alone, including the murder of a number of moderate Tutsi cabinet members. Kayibanda protected the guilty officials from punishment, and indeed the murderous prefect quoted above was elected to the National Assembly in 1965.91 Tutsi refugees in Uganda found new cause for fear in 1989-90, as Ugandan President Yoweri Museveni increasingly sided with Ugandan citizens trying to dispossess them of lands they had acquired. By 1990 the viability of the Tutsis' Ugandan exile community was in question. In the Hutu case, fear of enslavement was at the core of the ethnic mythology. As Peter Uvin has put it, Hutu ideology claimed that the Hutu-led government was "the sole defense against the Tutsi's evil attempts to enslave the people again."92 This fear was made concrete by the continuation of Tutsi rule in neighboring Burundi, and even more by the waves of violence that swept Burundi in 1965 and 1972, each time culminating in massacres of Burundian Hutu. Events in the early 1990s increased Hutu fears, giving extremist politicians' symbolic appeals more traction. The RPF's initial invasion in 1990 made Hutu fear of its Tutsi leadership an active concern. Later, after observing a cease-fire for half a year, the RPF launched a brief military offensive in February 1993, largely in retaliation for massacres of Tutsi the month before. During their attack, RPF troops committed atrocities of their own, giving even moderate 88. Joan Kakwenzire and Dixon Kamukama, "The Development and Consolidation of Extremist Forces in Rwanda, 1990-94," in Howard Adelman and Astri Suhrke, eds., The Path of Genocide: The Rwanda Crisis from Uganda to Zaire (New Brunswick, N.J.: Transaction, 1999), p. 72. 89. Peter Uvin, "Prejudice, Crisis, and Genocide in Rwanda," African Studies Review, Vol. 40, No. 2 (September 1997), p. 96. 90. Quoted in Lemarchand, Rwanda and Burundi, pp. 223-234. 91. Prunier, The Rwanda Crisis, pp. 56-62; and Lemarchand, Rwanda and Burundi. 92. Uvin, "Prejudice, Crisis, and Genocide in Rwanda," p. 98. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? ? 73 Hutu qualms about the Arusha peace process that was aimed at providing the RPF a share of power. Then, in October 1993 moderate Hutu President Melchior Ndadaye of Burundi was murdered by extremist Tutsi military officers, setting off an orgy of killings that left 50,000 dead on Rwanda's doorstep and another 150,000 Hutu seeking refuge in Rwanda.93 The shock of that event radicalized Rwandan politics, driving hard-line political factions to support the genocidal ideas that had previously been confined to a few extremists. Against this backdrop, later propaganda aimed at promoting fear could take root. One reluctant participant in the genocide reported that at the time neighbors were "spreading the rumour that the Tutsis in our area had had a meeting and were planning to kill the Hutus."94 As a young Rwandan summed it up, his neighbors "hear over and over again that the Tutsis are out to kill them.... They act not out of hate as fear."95 S3: OPPORTUNITY. The Hutu elite obviously had the opportunity to organize ethnic violence at any time, as long as they held government power. That they used that power to organize the genocide is beyond dispute. Extremist party leaders established the militias that carried out the genocide; extremist economic elites established the "hate radio" station that prepared the ground for it; and extremist military leaders deployed key army units to aid in the killing. Before 1990 Tutsi opportunity to mobilize inside Rwanda was nonexistent, effectively repressed by the government. Indeed, given the repression and the dispersed settlement pattern of Tutsi in Rwanda, only an external territorial base, such as that provided by Uganda, could have enabled the RPF to have a chance to organize an assault. The initiation of war in Rwanda in 1990 was therefore primarily an external invasion, and so was outside the ambit of any theory of civil war. That is why I do not focus on it here. By the end of 1992, in contrast, the RPF had recruited 20,000 mostly Tutsi exiles into a powerful military force with logistical support from Uganda and the Tutsi exile community. It controlled northern Rwanda, and proved its military superiority in its February 1993 offensive, when it penetrated to within 23 kilometers of Kigali against a crumbling Rwandan army, before it withdrew under international pressure.96 Both sides now had the opportunity to fight. S4 AND S5: MASS HOSTILITY AND CHAUVINIST MOBILIZATION. Symbolic poli- 93. Prunier, The Rwanda Crisis, pp. 174-180, 200. 94. Quoted in African Rights, Rwanda: Death, Despair, and Defiance (London: African Rights, 1994), p. 572. 95. Quoted in Mamdani, When Victims Become Killers, p. 191. See also Bruce D. Jones, Peacemaking in Rwanda: The Dynamics of Failure (Boulder, Colo.: Lynne Rienner, 2001), p. 40. 96. Colin M. Waugh, Paul Kagame and Rwanda: Power, Genocide, and the Rwandan Patriotic Front (Jefferson, N.C.: McFarland, 2004), p. 61; and Jones, Peacemaking in Rwanda, p. 33. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 74 tics played a critical role in causing and sustaining the war, and in enabling the genocide. The RPF's Tutsi members and supporters were in exile because of the exclusionary nature of Hutu ideology, which claimed they did not belong in Rwanda; successive Rwandan governments therefore refused to let them return. Had the Tutsi refugee problem been resolved, even on terms favorable to Habyarimana's government, there could have been no RPE Furthermore, young Tutsi exiles in 1990 thought of Rwanda as the mythical "land of plenty" their parents had described; so they flocked to join the RPF even after its severe initial reverses, and even if they had bright prospects in exile: Gerard Prunier estimates that the RPF was "the best-educated guerrilla force the world had ever seen."97 Older exiles, Tutsi businessmen in Europe and North America, provided the funding for the RPF through donations.98 It is unlikely that successful Tutsi joined the RPF because they calculated their individual prospects to be better with the outnumbered guerrillas than in their places of exile, or that RPF financial backers thought supporting the RPF was their best business opportunity. Given that the RPF had few selective incentives to offer-it permitted its soldiers very little looting, for example-the best explanation for its successful recruiting efforts is symbolic politics: the charisma of RPF leader Paul Kagame and the myth-based identity to which he appealed.99 Habyarimana responded with symbolic politics of his own, aiming to build on the long-standing hostility toward Tutsi. One observer claims that hating Tutsi was at the time the Hutu "lifestyle."100 The initial efforts of progovernment propagandists were summed up in the "Hutu Ten Commandments," a 1990 propaganda document widely seen as quasi-official, which advised its adherents that the Tutsi's "only aim is the supremacy of his ethnic group," and that "Hutus must stop having mercy on the Tutsis."'ol The extremist propaganda campaign invoked all the symbols of Hutu mythology to justify elimination of the Tutsi. First, the extremists argued, "a cockroach cannot give birth to a butterfly"-that is, Tutsi are inherently evil and prone to "unspeakable crimes"; they cannot change. Second, the Tutsi goal was at best to impose a "feudalism" approximating slavery, leading to slogans such as, "Let slavery, servitude, and discord be finished forever!" The kalinga 97. Prunier, The Rwanda Crisis, pp. 116-117. 98. Waugh, Paul Kagame and Rwanda, p. 52. 99. On the absence of looting, see Mamdani, When Victims Become Killers, p. 187. On Kagame's charisma, see Jones, Peacemaking in Rwanda, p. 30. 100. Aaron Phillip Karnell, "The Role of Radio in the Genocide of Rwanda," Ph.D. dissertation, University of Kentucky, 2003, p. 176. 101. Philip Gourevitch, We Wish to Inform You That Tomorrow We Will Be Killed with Our Families: Stories from Rwanda (New York: Picador, 1998), p. 88. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? i75 drum, the old king's symbol of the bloody emasculation of Hutu, was also revived as a symbol in Hutu propaganda: "No more kalinga!" was one slogan,102 though the object in question had been absent from Rwandan politics for thirty years. One shocking cartoon portrayed Burundian President Ndadaye being crucified and his genitals being cut off and hung on the kalinga-all with the encouragement of RPF leader Paul Kagame.103 A key weapon in the extremists' propaganda arsenal was the "hate radio" station Radio Telvision Libre des Mille Collines (RTLMC), the only independent radio outlet in the country. RTLMC, which began broadcasting in July 1993, was fiendishly effective because "it knew how to use street slang, obscene jokes and good music to push its racist message." It was even popular with Tutsi troops.104 Accounts suggest that although the extremists' charges against the Tutsi were not initially believed by ordinary Hutu, they came to be believed after months of repetition.105 Extremist propaganda argued that the necessary reaction was preemptive genocide. In one speech widely circulated on audiotape, a local official called on his audience to "rise up" in self-defense against the alleged genocidal aims of the Tutsi and "exterminate the scum." A 1993 newspaper article similarly asked, "What if someone brought back the Hutu revolution of 1959 to finish off the Tutsi cockroaches?"'06 By reinterpreting Hutu ideology as explicitly eliminationist,'07 the extremists were able to make their genocidal program accepted as part of the normal political landscape.10s8 Given this violent mythology, with Hutu popular fears at a fever pitch, and with the opportunity provided by its access to government institutions, the Akazu-the extremist faction surrounding Habyarimana's wife-was finally positioned for its chauvinist mobilization efforts to pay off. The extremists further fed feelings of insecurity with a campaign of assassinations aimed against Hutu moderates and of massacres against Tutsi in certain outlying regions. Bits of the government apparatus, including the presidential guard, increasingly went over to the extremists, as did factions from most of the Hutu opposition parties. 102. Allison Des Forges for Human Rights Watch, Leave None to Tell the Story: Genocide in Rwanda (New York: Human Rights Watch, 1999), p. 77. 103. Karnell, "The Role of Radio in the Genocide of Rwanda," pp. 188-189. 104. Ibid., p. 189. 105. Frank Chalk, "Hate Radio in Rwanda," in Adelman and Suhrke, The Path of Genocide, pp. 95- 99. See also Karnell, "The Role of Radio in the Genocide of Rwanda." 106. Des Forges, Leave None to Tell the Story, pp. 73-85. 107. On eliminationist ideology, see Daniel Jonah Goldhagen, Hitler's Willing Executioners (New York: Vintage, 1997). 108. See Prunier, The Rwanda Crisis, p. 200. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 76 What followed was, on one level, an ordinary coup: a network of Akazu conspirators in Kigali and in the provinces murdered the top leadership (possibly including the president), seized power in the capital, and then co-opted or swept aside nonconspirators lower down in the administrative chain. Because they quickly gained control of the army, they were able to use coercion when necessary. And it was necessary: as Human Rights Watch reports, "Soldiers and national police directed all the major massacres throughout the country," and they coerced those Hutu who tried to resist the slaughter.109 All this begs two questions: Why was war and genocide the plan? And why did the government and regional officials carry out the genocide? Symbolic politics theory provides a comprehensible answer. As a broad array of Rwanda experts emphasizes, the Hutu mythology was central in motivating the genocide.110 The ginocidaires were implementing a program that made sense to them in the context of their ethnic mythology and the fears they convinced themselves and their followers to feel. As Prunier notes, any explanation of the genocide "presupposes one absolutely basic thing: the total dehumanization of the Evil Other.""11 Ultimately, in a process Jack Synder labels "blowback," the genocidaires fell victim to their own propaganda that ordinary Tutsi were as big a threat as the RPF-a myth that also had its roots in the violence thirty years earlier.112 They acted accordingly; and as a result, they lost power. The reason why so many ordinary Hutu participated in the genocide is summed up by an elderly man quoted by Prunier: "I am ashamed, but what would you have done if you had been in my place? Either you took part in the massacre or else you were massacred yourself. So I took weapons and I defended the members of my tribe against the Tutsi." Prunier comments, "Even as the man pleads compulsion,... he agrees with the propaganda view (which he knows to be false) by mythifying [the victims] as aggressive enemies."113 In short, both the ideology and the compulsion were necessary to get most Hutu to participate in the genocide. The compulsion, in turn, was dependent on greater ideological fanaticism in the militia and the army leadership. This evidence demonstrates the inaccuracy of rationalist objections to the 109. Des Forges, Leave None to Tell the Story, pp. 8, 225. 110. These include Josias Semujanga, Origins of Rwandan Genocide (Amherst, N.Y.: Humanity Books, 2003); Twagiliman, The Debris of Ham; and Uvin, "Prejudice, Crisis, and Genocide in Rwanda." 111. Prunier, The Rwanda Crisis, p. 142. 112. On "blowback," see Jack Snyder, Myths of Empire: Domestic Politics and International Ambiti (Ithaca, N.Y.: Cornell University Press, 1990). On Akazu views of their own propaganda Mamdani, When Victims Become Killers, p. 207. For discussion of the earlier roots of the mytholog see Lemarchand, Rwanda and Burundi, p. 256. 113. Prunier, The Rwanda Crisis, p. 247. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? ? 77 symbolist argument. Fearon and Laitin argue that Rwandan "mythology merely made violence a thinkable 'possibility' on the part of Hutus," a possibility not significantly different from that contained in any other discourse of ethnicity.114 But Tutsi mythology, as Lemarchand noted in 1970, was long unique in its emphasis on claims of Tutsi ethnic superiority, and it spawned an unusually vicious Hutu countermythology that was used to justify genocidal violence not only in the 1990s but also in the 1960s. Rwandan ethnic mythology made genocidal violence not just possible, but much more likely than elsewhere. S6: RWANDA'S PREDATION-DRIVEN SECURITY DILEMMA. The nature of the security dilemma in Rwanda is best summed up by Bruce Jones: "Securit fears, real and constructed, were tools more than causes of war," and in many cases, "'fears' were indistinguishable from greed and predatory motives."115 The most important fears were those of the predatory Hutu extremists, who dreaded the possibility of losing power if the Arusha agreement were implemented. To the extent there was a security dilemma in Rwanda, it was manufactured: the Akazu had instigated the January 1993 massacre of Tutsi, and stalled the Arusha negotiations, in the hope of provoking the RPF attack that followed. That is Jones's point: the most predatory group, the Akazu, provoked violenc to force the RPF and some opposition Hutu to line up on opposite sides of the fight. Immediately after Habyarimana was killed, a large portion of the army tried to oppose the coup leaders, but once its commanders received word of renewed fighting with the RPF, they ceased their opposition and united with the extremists. Thus Akazu predation created a security dilemma in which RPF ef forts to protect itself and its Tutsi constituents threatened moderate Hutus (in cluding much of the army). RWANDA: RATIONALIST EXPLANATIONS Any rationalist explanation of the Rwanda conflict would elite's incentives for promoting violence in 1994. The pur would further emphasize the RPF's problem in credibly c by the Arusha accords. The elite-predation model would h and genocide offered a rational regime-saving strategy fo hypotheses are assessed below. R7: ELITE INCENTIVES FOR VIOLENCE. Rwanda's trouble 114. Fearon and Laitin, "Violence and the Social Construction of Ethni 115. Bruce D. Jones, "Military Intervention in Rwanda's 'Two Wars': P ence," in Walter and Snyder, Civil Wars, Insecurity, and Intervention, pp This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 [ 78 nomic crisis driven by a drop in international coffee prices, the closure of Rwanda's previously lucrative tin mine, and a series of droughts.116 As a result, gross domestic product per capita declined 27 percent from 1983 to 1990, while food production declined to starvation levels. Meanwhile, the government became massively dependent on foreign aid-for 22 percent of gross national product by 1991. The regime also increased demands for corvee labor from the hard-pressed peasantry, while being forced to cut its budget 40 percent, with social services absorbing most of the cuts. In this context, political unrest understandably began. Habyarimana had been directing economic largesse primarily to his and his wife's home regions in the north, and politicians from the south began criticizing the government for corruption and other misconduct. The regime was weakened further when the Akazu apparently murdered the president's intended successor. Finally, Habyarimana came under growing international pressure to move his country toward democracy. Incentives for chauvinist mobilization increased in the early 1990s: the RPF invasion further weakened the economy while providing a villain to mobilize against. Part of R7 is therefore correct: it was rational for the Hutu elite to consider high-cost strategies of maintaining power, such as war, because their previous strategy of economic development was failing. R1-R3 VERSUS R6: INFORMATION FAILURES OR PREDATORY ELITES? The violence in Rwanda was rarely driven by uncertainty. Rather, at every stage, actors were motivated by certainty about the other side's hostility. In 1994, as de Figueriedo and Weingast note, the Akazu's concern (and also Habyarimana's) was that, as a result of the Arusha peace accords, they would lose power to a coalition of Tutsis and moderate Hutus.117 Thus R6 is supported: predatory elites, not uncertainty, led to the renewal of violence in 1994. In contrast, R1-R3 are not supported. Nor is there evidence that the Akazu preferred a negotiated bargain to war (R1). Nor is there evidence apart from Akazu propaganda that leaders overestimated their rivals' hostility (R2). Finally, the Hutu leadership had no reason to believe that their military prospects were bright in 1994 (contra R3): even their initial successes in 1990 had come only with help from French and Zairean troops, and the undisciplined Zaireans were quickly sent home. The Rwandan army's discipline had begun crumbling at the time of the first cease-fire with the RPF in mid-1992, and the 116. The next two paragraphs are based on the accounts in Uvin, "Prejudice, Crisis, and Genocide in Rwanda," pp. 106-108; and Prunier, The Rwanda Crisis, pp. 84-97. 117. De Figueiredo and Weingast, "The Rationality of Fear." This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? 79 RPF's successful February 1993 offensive had shown unambiguously its military superiority.118 R4: COMMITMENT PROBLEMS. Commitment problems were real in Rwanda in 1994, but they were not decisive. The RPF, having engaged in ethnic cleansing and some killings of civilians earlier in the war, was poorly positioned to offer assurances of better behavior after joining the government. The Arusha agreement's provisions for dividing the army between Hutu and Tutsi could easily be seen from the Hutu perspective as allowing the fox to help guard the henhouse. And the October 1993 massacres in Burundi illustrated the implications if the RPF were indeed acting in bad faith. Still, rejecting the Arusha accords on these grounds would presuppose a preferable alternative, but in 1994 there were none: the RPF preferred to take power by agreement and in coalition with Hutu moderates, and the Hutu extremists lacked the military might to defend their power against a renewed RPF offensive. In sum, the choice facing extremist Hutus was to cede power peacefully to an RPF-Hutu moderate coalition, and then hope to influence it, or to lose a civil war to the RPE Even if the RPF was not trustworthy, the first alternative was still better than the second. The commitment problem cannot explain the Akazu's decisions in 1994. R5: PREEMPTIVE ATTACK. As noted above, the Rwandan army was militarily inferior to the RPF and had no military incentive for resuming the fighting. Indeed, it did not launch a preemptive attack on the RPF, so no preemptive military incentives motivated its behavior (contra R5). Rather, the Akazu deployed the army's most reliable units to carry out the genocide, while weaker units provoked but failed to contain the RPE.119 R6: WAR AND GENOCIDE AS RATIONAL REGIME-SAVING STRATEGIES. Was Rwanda's genocide a rational, if desperate, regime-saving strategy?120 Th gument goes that, first, ordering genocide was rational for the Akazu re because it would irrevocably commit those who carried out the killings to port the regime, given that the RPF would be highly likely to retaliate s they gain power. Second, the genocide would, if successful, eliminat RPF's base of support, while providing an excuse to eliminate moderate H opponents of the regime. Although these points are valid, they do not suffice to explain the gen 118. Prunier, The Rwanda Crisis, p. 113; Jones, Peacemaking in Rwanda, pp. 30-33; and Mam When Victims Become Killers, pp. 206-207. 119. Jones, Peacemaking in Rwanda, p. 39. 120. De Figueiredo and Weingast, "The Rationality of Fear." This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 180 First, this argument does not explain why genocide-a rarely chosen option that is taboo in most contexts-was the agenda at all. Second, it does not explain why a genocidal program worked in building political support: Why did pivotal actors cooperate with a policy that many knew would lead to the deaths of their friends and (since exogamy was common) relatives? Finally, the rationalist argument overlooks the fact that carrying out the genocide, far from saving the interim Hutu regime, led directly to its ouster. In its successful February 1993 offensive, the RPF had proven its military superiority over the Rwandan army. In the face of that military superiority, the genocide plan called for the Rwandan army to provoke the RPF by attacking it directly, while scattering the army to lead the implementation of the genocide.121 De Figueiredo and Weingast try to save their theory by claiming that the genocidaires recognized this and planned to regroup in exile after completing the genocide.122 There is, however, no evidence for this claim, and most Rwanda experts explicitly reject it. The Akazu did have more reasonable options. A simple coup would have served as well if they wanted only power. If they required war to solidify popular support, war without genocide would have afforded a chance of gaining foreign military assistance, especially from France, to reverse the RPF's military edge, while conserving the regime's limited military forces. Indeed, as was utterly foreseeable, implementing the genocide led to quick and permanent military defeat for the murderous Akazu regime. While the regime's military, logistical, and political attention was focused on committing genocide against unarmed and harmless Tutsi, "there were very few government troops facing the guerrillas and they tended not to put up much of a fight ... [except in] the cities of Kigali and Ruhengeri."123 So the RPF overran the country, and the interim regime was ousted. The genocide is simply not explicable as a rational regime-saving strategy. Rather, the motivation for genocide was in madness and despair. As Mahmood Mamdani puts it, "Faced with a military defeat that seemed to sound the very death knell of Hutu power, the genocidaires chose to embrace death itself as an alternative to life without power."124 The Hutu extremists thus had an incentive for military action and genocide in 1994, but only in the context of insane preferences: if one assumes that they were willing to be driven into exile-losing almost everything they had-as 121. Des Forges, Leave None to Tell the Story, pp. 8, 223. 122. De Figueiredo and Weingast, "The Rationality of Fear." 123. Prunier, The Rwanda Crisis, p. 268. 124. Mamdani, When Victims Become Killers, p. 215. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? 81 long as they could slaughter the Tutsi first, then such an incentive existed. In sane military terms, however, they did not, as they could not alter the RPF's military advantage in the context of a genocide. R9: MASS UNCERTAINTY ABOUT THEIR OWN ELITES' ACTIONS. The logic of this hypothesis is that predatory elites provoke violence by their enemies, then conceal the provocation, to convince their constituents that the enemy is aggressive and must be fought. To some extent this did occur, in a fictional 1990 assault on Kigali, and in early 1993 when a series of massacres by Hutu extremists provoked the RPF into answering with a counteroffensive and atrocities against civilians. The political effect benefited the murderers, alienating some moderate Hutus and driving them to support the genocidal extremists.125 The hypothesis of mass uncertainty, however, is wrong in the latter case: the massacres by the Hutu extremists were unconcealed; some moderate Hutus rethought their views in response to the RPF's violence anyway. In 1994, during the genocide, uncertainty about elite intentions played no role. Hutus knew the Akazu's murderous intentions: RTLMC explicitly and repeatedly announced that genocide was under way, saying, for example, "We will ransack and exterminate them,"126 and indeed it even directed specific killings. Elites' deception of their own followers was not a cause of this conflict or of the genocide. Conclusion This article tests the ability of two major theories, the rational choice and symbolic politics theories, to explain the outbreak of ethnic war in Sudan and Rwanda, as well as the Rwanda genocide. A key issue is what cultural discourse can explain. Symbolic politics theory posits that a necessary precondition for ethnic war is ethnic myths justifying hostility, operationalized as ethnic myths justifying predatory policy-that is, the insistence on dominance. Rationalists such as James Fearon and David Laitin, in contrast, deny the importance of such narratives or "incompatible values." The analysis here shows that in these cases, ethnic conflict was driven precisely by "incompatible values" as defined in hostile mythologies. In Sudan, northern myths insist on an Islamic identity that requires application of sharia to the entire country, whereas southern myths cast northerners as would-be enslavers and prescribe resistance to northern domination. Contrary to Fearon 125. Prunier, The Rwanda Crisis, pp. 174-175, 180-183. 126. J. Chretien, J. Dupaquier, M. Kabanda, J. Ngarambe, and Reporters sans Frontihres, Rwanda: Les midias du gdnocide (Paris: Karthala, 1995), p. 195. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 182 and Laitin's claims, the very title of Francis Deng's book-their main sourcenotes that Sudan's ethnic war is all about cultural discourse: it is a "war of visions" of what their country is and ought to be. Similarly, in Rwanda, the Hutu narrative casts Tutsi as alien invaders with no legitimate right to be in the country; in contrast, the Tutsi diaspora story casts the RPF as liberators from Hutu tyranny. Those, too, are "incompatible values," and what happened would not have happened without them. In both cases, evidence shows that myths justifying hostility drove the violence. The first symbolist proposition is supported. A related question is what these values do. The evidence shows that discourses do vary, and they covary with the nature of violence in the two cases examined here. Northern Sudanese discourse justifies discrimination and repression against southerners, but there is no evidence of the sort of rhetoric that became common in Rwanda: southern Sudanese may be seen as "slaves," but not as "cockroaches" that should be exterminated. The point about Rwanda is not that the mythology "merely made mass violence thinkable"; rather, it made open, rapid, and explicitly intended genocide a politically viable option. Sudanese mythology is less extreme, so although war is sanctioned, genocide is not: southern refugees were sheltered in the northern capital of Khartoum; they were not massacred.127 Such mythologies are only one precondition for ethnic violence in the symbolist account; other conditions must also be present. The second and third symbolist hypotheses, about fears and opportunity, are not controversial and are easily supported. In Sudan, northerners and southerners feared the loss of their identity, while southerners also feared violent repression. In Rwanda, fears of physical extermination were prevalent on both sides. More controversial is the fourth symbolist hypothesis, about the role of hostile mass attitudes, but this hypothesis is also supported. It seems clear from election results and media analysis that in Sudan, the imposition of sharia on southerners was highly popular among northerners, and that this was understood on both sides as an assertion of northern dominance. Southerners understandably reacted defensively. In Rwanda the evidence is equally clear: the hard-line elite's goal of Hutu domination was unambiguous and highly popular, while at the mass level, hatred of Tutsi was described as a "lifestyle." The Tutsi fought back when they could. 127. Therefore, although some have labeled the violence in Sudan genocidal, it was not genocidal in the Rwandan sense. For details, see Nyaba, The Politics of Liberation in South Sudan; and Johnson and Prunier, "The Foundation and Expansion of the Sudan People's Liberation Army," pp. 117- 141. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? 183 The fifth symbolist hypothesis, regarding symbolic appeals by elites, is also supported. In Sudan, Nimeiri did engage in such appeals beyond the actual imposition of sharia, but as he was doing it defensively, he did so less than he might have. The symbolic power of the sharia issue in Sudan is best illustrated by the ability of the small but extremist NIF to block any moves by more moderate leaders to repeal it or even suspend its application. In Rwanda, the pervasive racist appeals in the media by Hutu extremists is widely known. These efforts at chauvinist mobilization also demonstrate the nature of the ethnic security dilemma in these cases, in which predatory motives are of primary importance, but security-seeking also plays a role. The rationalist hypotheses fare less well, with the pure-uncertainty model failing entirely. In both cases, information failures between groups play little role: the predatory aims of the northern Sudanese and Rwandan Hutu elites were unambiguous. As a result, they made few efforts to negotiate a bargain with their adversaries, disproving also the assumption that leaders preferred a negotiated bargain to war. Incentives for preemption and miscalculations of power also played little role: the military balance was typically clear to all, and in no case was a war begun as the result of an effort to exploit the adversary's momentary military vulnerability. Commitment problems were irrelevant in the Sudan case; the issue was Nimeiri's hostile current acts, not uncertain future intentions. And in the Rwandan case, while there was a commitment problem facing the sides in the Arusha peace process, the Hutus' weak military position made acceptance of the deal less risky than rejecting it; so the commitment problem cannot account for their behavior. The basic logic of the elite-predation model is also disproved. The hypotheses about predatory elites and mass fear are supported in both cases, but these are also compatible with the symbolist account. There is, however, no evidence that the mass public in either case was significantly fooled about the sources of violence. De Figueiredo and Weingast's oxymoronic title, "The Rationality of Fear," wrongly implies that rational security concerns (based on misinformation) prompt rational security-seeking behavior. What happened in the cases was predation driven by status anxiety in Sudan, and self-destructive genocide motivated by a depraved eliminationist ideology in Rwanda. In neither case was rational concern for physical security the motive, and in neither case was the policy a rational, security-maximizing one. The emotion of fear was present; the rationality of fear was not. Finally, the keystone of the elite-predation model is the assertion about the utility of a violent strategy for maintaining power; but in both cases, the extremist leaders had more promising options. In Sudan, Nimeiri's turn toward the Islamist opposition was clearly riskier than maintaining his alliance with This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 84 the southerners. In Rwanda, using the army to carry out the genocide instead of fighting the RPF predictably led to the gdnocidaires' downfall. These behaviors are explicable only within a particular normative context-the ethnic discourses that made Islamism in Sudan and genocide against Tutsis in Rwanda so popular. The main argument of the elite-predation model is wrong for both cases. It is important to note that the current study encomp Although the theoretical arguments of symbolist theor ble to cases of ethnic war and genocide, and the empiri here are corroborated by findings in studies of the Bal Lanka, and elsewhere,128 there is still the risk of debat findings in different cases. It is possible, for examp violence studied by Paul Brass in North India is a differ non that a different theory applies to it.129 Caution is stage. Still, the symbolist approach is promising for explaining puzzles generated by the rationalist literature. For example, Fearon and Laitin's explanation of ethnic peace is that some ethnic relationships are characterized by "in-group policing" that prevents large-scale violence, whereas others have delicate "spiral equilibria" that can break down into such violence.130 They do not, however, explain the origins of these different patterns. Symbolist theory would suggest that they lie in a difference in mass attitudes shaped by different sorts of ethnic mythologies. Thus even within the model, initial information about attitudes does most of the explanatory work, while the mechanism of symbolic politics better explains the emotionally driven process of ethnic mobilization. The weakness of rational choice theories of ethnic war, in sum, is that they must either explicitly assume or implicitly smuggle in two factors that most need explaining: the political coherence of ethnic groups and their adoption of predatory goals. At a minimum, this means that doing good rational choice analysis requires first doing good symbolic politics analysis to identify those groups most likely to adopt predatory goals. On the other hand, rationalists are right to point to the importance of leaders' incentives. In both Sudan and Rwanda, the timing of the violence was largely driven by economic down- 128. See Kaufman, Modern Hatreds; and Bryan Pfaffenbarger, "Ethnic Conflict and Youth Insurgency in Sri Lanka: The Social Origins of Tamil Separatism," in Joseph V. Montville, ed., Conflict and Peacemaking in Multiethnic Societies (Lexington, Mass.: Lexington Books, 1990), p. 247. 129. Paul R. Brass, Theft of an Idol: Text and Context in the Representation of Collective Violence (Princeton, N.J.: Princeton University Press, 1997). 130. Fearon and Laitin, "Explaining Interethnic Cooperation." This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms Symbolic Politics or Rational Choice? 185 turns that discredited regimes that had promised but failed to deliver economic development. Another advantage of symbolic politics theory is that it explains many additional phenomena that rational choice theory cannot. It explains why ethnic cleavages were so prominent in these cases. It explains the prominence of purely symbolic issues such as northern Sudanese desires to impose sharia on southerners, and why particular options (such as genocide in Rwanda) are prominent in some places but not others. It explains why politicians put so much effort and so many resources into making symbolic appeals, and why successful symbol-manipulators often gain the support of people whose tangible interests are harmed by their policies. This breadth of explanatory power is the theory's chief claim to a status as a paradigm superior to rational choice theory for analyzing elite-mass relations. The policy implications of these findings are also important. If, for example, the pure-uncertainty model were right, then the correct policy prescription for preventing or ending ethnic war would be that proposed by Barbara Walter: institution building and outside guarantees to the parties to ameliorate problems of credible commitment.131 In fact, however, Walter identifies only one case (Rhodesia) in which her prescription was adequate to resolve an ethnic war, and even that case is arguable. More typically, the efforts of outside mediators are undermined by the predatory efforts of extremist elites. As a result, because the rationalist paradigm is the dominant one among practitioners, few ethnic wars are successfully resolved by negotiated agreement.132 The symbolist argument suggests attention to a broader range of options in addition to mediation, outside guarantees, and institution building. A pessimistic interpretation leads to the realist argument of Chaim Kaufmann that reconciliation and institution building are both hopeless; if groups are that hostile, Kaufmann argues, the best that can be done is to separate them by partition or to allow one side forcibly to repress the other.133 An optimistic inter- 131. Barbara F. Walter, "Designing Transitions from Civil War: Demobilization, Democratization, and Commitments to Peace," International Security, Vol. 24, No. 1 (Summer 1999), pp. 127-155; and Walter, "The Critical Barrier to Civil War Settlement," pp. 335-364. 132. On the failure of existing peacemaking techniques, see Stuart J. Kaufman, "Escaping the Symbolic Politics Trap: Reconciliation Initiatives and Conflict Resolution in Ethnic Wars," Journal of Peace Research, Vol. 43, No. 2 (March 2006), pp. 202-203. On the weakness of Walter's evidence, see Chaim Kaufmann, "Rational Choice and Progress in the Study of Ethnic Conflict: A Review Essay," Security Studies, Vol. 14, No. 1 (January-March 1995), pp. 177-178. 133. Chaim Kaufmann, "Possible and Impossible Solutions to Ethnic Civil Wars," International Security, Vol. 20, No. 4 (Spring 1996), pp. 136-175; and Chaim Kaufmann, "When All Else Fails: Ethnic Population Transfers and Partition in the Twentieth Century," International Security, Vol. 23, No. 2 (Fall 1998), pp. 120-156. This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms International Security 30:4 86 pretation yields the conclusion that more effort needs to go into promoting reconciliation between the groups in conflict, roughly according to the model of the French-German reconciliation after World War II.134 The logic is that new institutions can remain effective only if the sides reconcile themselves to the goals they can achieve through cooperation. Both of these optionspartition and reconciliation-require more attention than they have received if conflict resolution practice is to improve its track record in ethnic wars. 134. See Long and Brecke, War and Reconciliation. On French-German reconciliation, see Ackermann, "Reconciliation as a Peace-Building Process in Postwar Europe." This content downloaded from 134.117.10.200 on Mon, 05 Oct 2020 02:32:28 UTC All use subject to https://about.jstor.org/terms