Journal of World-Systems Ressearch - Festschrift for Immanuel Wallerstein Part I 308 Nationalism and Rationality Michael Hechter* Michael Hechter Department of Sociology University of Washington Box 353340 Seattle, WA 98195-3340 http://www.soc.washington.edu/ hechter@u.washington.edu Recent years have witnessed a virtual epidemic of nationalist violence in the world. In 1994, for example, eighteen of the twenty-three wars being fought were based on nationalist or ethnic challenges to states. About three quarters of the world’s refugees were fl eeing from, or were displaced by, ethnic or nationalist confl icts. And eight of the thirteen United Nations peacekeeping operations were designed to separate the protagonists in eth- nopolitical confl icts. As a result, much attention of late has been devoted to two questions. Can this violence be contained—say, by adopting particular institutions and policies? And if so, then what might such institutions and policies consist of? The answers to these vitally important questions are unclear. In part, they are unclear because there are fundamental disagreements about the motivations of the participants in nationalist movements, and more specifi - cally, about the rationality of nationalism. For some observers, nationalism is anything but rational—it results not from benefi t/cost calculation but from deep-seated sentiments and emo- tions that are virtually immutable. If nationalism basically springs from the old Adam, then little, if anything, can be done to contain it. We would have about as much luck containing the destructive force of nationalism as in dealing with El Niño. If, however, nationalism springs from rational roots, then its course con- ceivably might be affected by institutions that decrease individuals’ incentives journal of world-systems research, vi, 2, summer/fall 2000, 308-329 Special Issue: Festchrift for Immanuel Wallerstein – Part I http://jwsr.ucr.edu issn 1076-156x © 2000 Michael Hechter http://www.soc.washington.edu/ http://jwsr.ucr.edu mailto:hechter@u.washington.edu Michael Hechter309 Nationalism and Rationality 310 • The dyed-in-the-wool nationalist is a romantic, not a rationalist. He is a communitarian, not an individualist. He thinks in terms of the spirit and culture of his people, not in terms of bargains and calculations. He will fi ght for his cause despite any number of rational arguments show- ing it to be unjustifi ed (Birch 1989: 67). • There remain ‘irrational’ elements of explosive power and tenacity in the structure of nations and the outlook and myth of nationalism… The confl icts that embitter the geo-politics of our planet often betray deeper roots than a clash of economic interests and political calcula- tions would suggest, and many of these confl icts, and perhaps the most bitter and protracted, stem from just these underlying non-rational ele- ments (Smith 1986: 363). • The passions evoked by ethnic confl ict far exceed what might be expected to fl ow from any fair reckoning of ‘confl ict of interest’ (Horowitz 1985: 134-35). • As Chateaubriand expressed it nearly 200 years ago: “Men don’t allow themselves to be killed for their interests; they allow themselves to be killed for their passions.” To phrase it differently: people do not volun- tarily die for things that are rational (Connor 1993: 206). Before rushing to judgment, however, it must be appreciated that the claim that nationalism is irrational can mean two quite different things. It is obvious that nationalism can be collectively irrational, for it is often associ- ated with undesirable social outcomes like economic decline and civil war. It is questionable whether such outcomes are the consequence of individual irrationality, however. After all, it is well-known that many undesirable out- comes—like rush hour traffi c, overfi shing and environmental pollution— are by-products of rational action (Kollock 1998). For this reason, the claim that nationalism may be the product of indi- vidual irrationality is more provocative. For anthropologist Clifford Geertz (Geertz 1994: 31), nationalists regard congruities of blood, language and custom as having “an ineffable, and at times overpowering, coerciveness in and of themselves. One is bound to one’s kinsman, one’s neighbor, one’s fellow believer, ipso facto; as a result not merely of personal affection, practical necessity, common interest, or incurred obligation, but at least in great part by virtue of some unaccountable absolute import attributed to the very tie itself.” to participate in it. The prospects for containing nationalist violence there- fore hinge on the extent to which nationalists are rational—and thus might respond to institutional incentives. Yet is nationalist violence rational? Consider three examples of nationalist violence culled from different parts of the world. Our itinerary begins in Sri Lanka. In the midst of a light- hearted celebration of May Day taking place in the city of Colombo in 1993, a man rushed through the parade toward the marching 68-year-old Presi- dent Premadasa of Sri Lanka and set off explosives that were attached to his body. The President and his assassin were instantly killed, as were at least ten other people. The government of Sri Lanka blamed the Tamil Tigers, a rebel group that had waged a ten year war of secession in the country’s north and east and had used suicide bombers in the past to kill government and army offi cials (Gargan 1993). Whereas such self-sacrifi ce in the name of a nationalist cause is relatively rare, it is far from unique. For example, Irish Republican Army hunger strikers in Northern Irish detention centers were willing to pay the ultimate price on behalf of their own national movement. And Hamas suicide bombers in Israel effectively set the stage for the victory of the hard-line Likud Party in the 1996 Israeli elections. Now move on to the metropolitan heartland—England. Consider the movie Patriot Games, in which two members of the Irish Republican Army target a prominent British politician for assassination. (Luckily for the distinguished British target, Harrison Ford was lurking in the neighborhood to make sure that the assassin’s best laid plans would go awry.) Although this particular story is, of course, a fi ction its main outlines are all too real (Feldman 1991). This little tour ends—where else?—in that territory now known as The Former Yugoslavia. In late October 1992, some of the Croatians who had fl ed towns in Krajina that had been taken by Serb forces returned to their former homes in the company of United Nations troops. They soon discovered that Serb gunmen had desecrated the graves of their ancestors. Serbs had pulled the covers off Croatian tombs and machine gunned their remains. Surely this bizarre event must reveal great irrationality; it must take some very odd passions to make these Serbs waste valuable ammuni- tion on Croats who, after all, were already long dead (Fearon 1994). On the basis of stories like these, it is no wonder that many observers regard nationalism as irrational. Thus, Michael Hechter311 Nationalism and Rationality 312 The hallmark of rational individual action is its instrumentality. People are rational to the extent they pursue the most effi cient means available to attain their most preferred ends. These ends may be material or non-mate- rial. Thus people are irrational when they pursue a course of action regard- less of its consequences for their personal welfare. This need not condemn altruistic actions to the realm of the irrational if the altruist’s own personal welfare is enhanced by her giving. In some circumstances, committing sui- cide may even be rational, for death can be preferable to a life that promises little but extreme and unremitting physical suffering. Not so, however, for political suicide: it cannot be rational to consider social ties as binding regardless of their consequences for one’s own welfare. Now let’s return to the three stories. How can it be rational to know- ingly die for a cause, engage in terrorism, or waste ammunition on corpses in a cemetary? Of the three, the Sri Lankan story is the only one that qualifi es as irrational. This is because suicide bombers know with certainty that they will die in carrying out their mission.1 By contrast, the Irish Republican Army took pains to minimize risks for its snipers. Even ostensibly individual events like sniper attacks involve elaborate planning and the coordination of many different people—from the gunman, to support staff providing weapons, ammunition, and vehicles, to sympathetic bystanders. Typically, paramilitary snipers employed a typical “runback”, or escape route, against British occupational forces in Northern Ireland. The point of the runback was to provide maximum security for the sniper and his team. Security must be provided, in turn, to induce rational members to under- take the risky business of attacking a British Army troop carrier. Unlike in the Sri Lankan example, the participants here are presumed to be at least somewhat self-interested.2 Fair enough, but why would a rational person ever join a treacherous organization like the Irish Republican Army? Whereas it is easy to appre- ciate that, in certain circumstances, nationalist groups might strategically adopt violent means to attain their goals, is it a mystery to understand why individuals might bear very high risks of injury, punishment and even death to help bring the collective good of sovereignty to their nation? Not at all. One might surmise that the use of violent means will tend to attract members who are skilled in violence, and will discourage others who are neither skilled nor interested in it. But there is a deeper reason, as well. To the degree that members are dependent on a solidary group that adopts vio- lent means, they may be willing to take great risks. (Much the same can be said of membership in inner-city gangs in North America, or in the Sicilian mafi a). Risk-taking is not irrational: people have always engaged in risky occupations. Although it is dangerous to build sky scrapers, or to be a policeman or a fi reman, that people can be found to fi ll these positions is no cause for consternation. The riskiness of membership in a violent national- ist group is not so very different. What of the Yugoslav example? An event that on its face seems wildly irrational can also have an instrumental explanation. Fearon (1994) argues that the Serb gunners’ behavior was consciously designed to heighten the salience of the boundary between Serbs and Croats. Under Tito’s regime, this boundary had been downplayed, and there was considerable social inte- gration (indicated by relatively high rates of exogamy between these two 1. Can the political suicide carried out by young Tamil Tigers be considered to be a rational act? There is no hard and fast answer because there is an ongoing debate among social scientists about the status of beliefs in rational action. Some writers insist, with Pareto, that an agent must have scientifi cally valid beliefs to act rationally (Elster, 1989). On this view, all action that owes to faith or religious belief is irrational. Since scientifi c research is based on metaphysical premises about the existence of an ordered universe, it too would have to be considered as irrational by a Paretian. Others hew to a more subjective conception of rational action (Boudon, 1996). On this view, political suicide is rational if agents have a fi rm conviction that they will be adequately compensated in the hereafter. This raises other questions, however. Should we consider paranoid schizophrenics who dress like Napoleon to be rational, as well? Hardly, but why not? Popper’s (Popper 1994) idea that rational people must show a readiness to revise their beliefs in the face of much contrary evidence provides one possible answer. Since it is the most intellectually conservative option, in this paper I choose to regard political suicide as irrational. 2. Further, detailed ethnographic analyses of intergroup riots in Sri Lanka, India and other South Asian countries reveal that the participants were ususally organized and acting purposively, often with the complicity of the police and other authorities (Tambiah 1996). Michael Hechter313 Nationalism and Rationality 314 communities.) In such a context, desecrating Croat cemeteries had a pre- dictable effect: it instantaneously heightened the salience of the Serb/Croat boundary: In both Serbian and Croatian culture ancestral graves are endowed with great signifi cance. For example, ceremonies are held and offerings made regularly at the graves of important family members. Serb gunners knew this, of course, knew that the Croats knew it, and knew that the Croats knew that they knew it. Desecrating cemeteries is part of a calculated plan by Serb extremists to make ethnic cohabitation impossible by spreading and deepening hatred across groups. Likewise for the Serb policy of systematic rape in the Bosnian war. Such measures do more than just make Bosnian Muslims or Croats too angry to live with Serbs in the future. They are also calculated to make it more dif- fi cult for less virulently nationalist Serbs to live with Muslims or Croats, due to fear of reprisal or discrimination (Fearon 1994). Whereas in the 1980s Serbs carrying symbols of Serbian iconography were treated with contempt even by many fellow Serbs, a decade later, Serbs were punished for their failure to make Serbian their primary social iden- tity. Serbs who clung their Yugoslav identity were subjected to harsh punish- ments (Glenny 1993). Serb paramilitary units that swept into multiethnic Bosnian villages fi rst killed those Serbian residents who were in favor of ethnic integration. Only later did they turn their attention to non-Serbian residents (Mozjes 1994). Spending time and effort to desecrate Croat cemeteries is strategic given the knowledge that Croats are bound to regard this behavior as highly threatening—akin to cross-burning in the American South. Extrem- ist Serbs, in turn, counted on their ability to predict how Croats would act given this provocation. So a good deal about the kind of nationalist violence that, on the face of it, seems to be irrational has, at least, a plausible rational account. If nationalist violence is largely, if not wholly, the result of rational action, this suggests that, under certain conditions, it indeed can be contained because rational actors will respond to institutional incentives. If so, then what kinds of institutions can contain it? Now many people might think that theory in general, or rational choice theory in particular, provides an answer to this question. For example, I was recently asked to write an encyclopedia article about ‘the rational choice theory of nationalism.’ After some refl ection I turned this offer down because, so far as I am aware, no such creature exists. To be sure, a rational choice theorist can always say that nationalism can be contained by adopt- ing institutions that raise the costs of nationalist violence and/or decrease its benefi ts. But this answer is practically vacuous: in particular, it does not tell us what kinds of institutions provide the requisite incentives. To illustrate this point, I consider an institution that has long been asso- ciated with nationalism. One of the most venerable ideas in social theory, harking back at least to the 16th century German political theorist Althusius (Althusius, 1964), is that political decentralization—or federation—is best suited for the governance of multinational polities. This idea has also been forcefully advocated by the political scientist William Riker (Riker, 1964). According to Riker, federation is the outcome of a bargain between rulers of the central state and leaders of its territorial sub-units. As in all bargains, this one only works because it appeals to both parties. Federation enables the territorial sub-units to attain some degree of political self-control while profi ting from access to the greater resources and military protection that is afforded by membership in large polities. At the same time, the federal bargain offers central rulers a relatively-cost effective means of maintaining their state’s territorial integrity. However, recent events—including the collapse of federations in the Soviet Union, Yugoslavia, and Czechoslovakia, and the continued thirst for secession among the cultural minorities in federations like Canada and Spain—have led many scholars to question the ameliorative effects of polit- ical decentralization. These events suggest the very real possibility that, far from inhibiting nationalist confl ict, federation instead exacerbates it. Still other observers argue that federation has no determinate effects on nation- alist confl ict at all. Evidently, the nature of the relationship between federa- tion and nationalist confl ict is highly contentious. Does federation reinforce nationalism by empowering national leaders, and whetting their appetites for even greater powers or privileges, does it erode nationalism by enabling nations to satisfy their demands within the existing state, or does it have no determinate effects at all? There are three views. Michael Hechter315 Nationalism and Rationality 316 why federation could intensify nationalist conflict The causal mechanism responsible for this effect owes to the very nature of federation. Federation diverts some government functions—and hence resources—from the center to territorial sub-units. Federation may stimu- late nationalist confl ict because it provides potential nationalist leaders with patronage and other resources that can be mobilized for nationalist ends. Federation also tends to provide institutional supports for nationalism: Federalism…is an important source of institutional capacity because it pro- vides a set of political levers and access to resources that make group mobilisa- tion more likely. While often put in place as a means of accommodation and cooptation, federal institutions can be quickly turned to new agendas when a coopted leadership is replaced or changes its preferences (Meadwell 1993: 200 ; Roeder 1991) In addition to the material incentives to nationalist mobilization that it may provide, federation also may have cognitive implications. When nations are given many of the accoutrements of real states, this also encourages people to think and act according to national categories (Brubaker 1996).3 Moreover, federation may be better suited to resolving material differences between units than cultural ones (King 1982: 47-8). All told, federation may provide both material and cognitive supports for nationalist confl ict. This view has at least one clear policy implication: to contain nationalist confl ict, local leaders should be offered meaning ful, substantial careers in the central government (Laitin, 1998). By this means, nationalist leaders will be transformed from peripheral magnates anxious to drain power from the state into stakeholders committed to upholding it. The historical record provides ample evidence linking federation to nationalist confl ict. The United States civil war broke out in a federation; Pakistan—another federation—lost Bangladesh. The only socialist states that dissolved following the climactic year of 1989—the Soviet Union, Czechoslovakia and Yugoslavia—also just happened to be federations. Moreover, Soviet policies aiming to defuse nationalism by decentral- izing authority backfi red (Kaiser 1990). Federation encouraged congeries of local groups to form nations where none had previously existed. In each union republic, the titular nationality used its position whenever possible to develop its own version of great power chauvinism, limiting the rights of its own minorities (save, of course, for Russians). For example, Georgia became a protected area of privilege for Georgians. They received the bulk of the rewards of the society, the leading positions in the state, and the largest subsidies for cultural projects, while the Armenians, Abkhazians, Ossetians, Ajarians, Kurds, Jews, and others were at a considerable disadvantage in the competition for the budgetary pie. Although Soviet leaders decentralized to coopt indigenous elites, when these elites began tilting towards nationalism, the Soviets abruptly switched gears and centralized from 1934-38.4 Similarly, the recent increase in Quebec’s regional authority does not seem to have dampened the fate of the separatist political party. Despite taking over the Quebec government—and the subsequent passing of exten- sive language legislation protecting French and the Francophones, a near- majority of Quebecois voted for separation in 1995. Nor has Spain’s devolution of power to the Basque region put its separatist party (ETA) out of business. The opposing view is also based on an intuitively appealing causal mechanism. Since federation is a form of indirect rule, it ought to reduce the demand for sovereignty (Hechter 2000). On this account, federation should also serve to mute nationalist confl ict.5 Since nations are by defi nition cul- turally distinctive, individual members’ values refl ect (to some degree) these distinctive national values. Although members of national groups share val- ues—minimally, those relating to the attainment of wealth, power and pres- tige—in common with all the other inhabitants of a given multinational state, they also share a set of values derived from their national culture that are distinctive. Typically, these values include preferences to speak a distinct language, and practice a distinct religion. 4. If Soviet decentralization generated nationalism, so did the subsequent attempt to increase central control. Re-centralization proved impossible in an increasingly complex economy; even when regional leaders were motivated to act in the center’s interest, they lacked the information necessary to do so. Hence, the only policies they could pursue were nationalist in outcome (Kaiser 1994: 331). 5. This advantage is far from costless, however. “The more decentralized a state is, the more the coordination or negotiation that will have to be carried on between the different jurisdictions. Therefore, the total cost of coordination increases as the degree of decentralization increases (Breton and Scott 1980: xvi-xvii).” Michael Hechter317 Nationalism and Rationality 318 Governments provide a range of collective goods. Some of these goods—like defense—are universally valued by inhabitants. Others—like education in a particular language, and state support for a particular reli- gion—appeal to only a portion of the state’s inhabitants. Whereas some universally-valued goods may better be provided centrally,6 goods that are valued only by a segment of the society are better provided locally (Oates 1972). Local provision of these goods is superior because it increases the likelihood that the right mix of goods will be produced—the mix that is most congruent with the distinctive values of the national group.7 As federa- tion involves the devolution of (at least some) decision-making to localities, it increases local self-governance. To the degree that at least some of the units in a federal system constitute nations, then federation should have the effect of inhibiting nationalism. Since sovereignty is neither more nor less than self-governance, it follows that to the degree federation increases a nation’s self-governance, its demand for sovereignty must be correspond- ingly reduced.8 This reasoning implies that the less self-governance a nation has in a multinational state, the greater the possibility of nationalist confl ict. Assum- ing that nations make up at least some of the constituent local or state sub- units in a multinational polity, then the greater the powers of the central government relative to those of state and local governments, the greater the nationalist confl ict. A constitution that minimizes the state’s control over disposable, transferable revenue and rights presents a very small target for nationalists. It stands to reason that local politicians are less likely to play the nationalist card when their constituents see less benefi t in sovereignty. This view has quite a different policy implication: to contain nationalism, the central rulers of multinational states ought to grant political devolution to mobilized national minorities. In spite of the apparent failures of federation alluded to above, this argu- ment also commands ample supportive case-study evidence. Many central rulers have turned to federation as a means of reducing nationalist dis- content, and they continue to do so. Britain’s recent offer of devolution to Scotland and Wales was welcomed by voters in both lands; further, the more thoroughgoing devolution in Scotland was more enthusiastically sup- ported than its relatively anemic Welsh counterpart. Spain and Belgium have recently undergone signifi cant constitutional moves from unity toward federation as a means of resolving national confl icts (Forsythe 1989), and even France—traditionally, the archetypal unitary state—has granted Cor- sica a certain amount of devolution (Savigear 1989). In the Spanish, Belgian and British cases, very signifi cant powers have been granted to the relevant sub-units. Swiss federation has been widely celebrated (Smith 1995: 14; McGarry 1993:31).9 Finally, the federal United States has experienced little in the way of nationalist confl ict since the Civil War (Glazer 1977). This evidence suggests that federation may indeed mute, if not inhibit, national- ist confl ict. There is yet a third view of these matters. Some scholars claim that noth- ing general can be said about the effects of decentralization at all because it 6. There is even some doubt about this. Whether collective goods are optimally produced by central rather than local authorities probably depends on the production functions of these goods— in particular, on scale economies. Defense, for example, probably gains from central provision because it entails large economies of scale. Many other state-provided goods, however, may be optimally provided on a more decentralized basis. The decentralized provision of collective goods, which is a characteristic of federation, is likely to be superior because it sets up a market for governmental rules and regulations giving incentives for citizens to ‘vote with their feet’ (Ostrom 1961; Weingast 1995). 7. This is one reason why political theorists have traditionally insisted that all true democracies must be small societies (Sale 1980). Plato, for instance, argued that the ideal number of citizens was 5,040. 8. In this respect, it should be noted that surveys indicate that federation is even popular among the inhabitants of regions which are culturally similar to state cores. Thus, popular support for subnational governments (Länder) in Germany has risen steeply since their introduction in 1949, and is on the rise in Italy (Putnam 1994: 59). Despite a sharp decline in survey measures of Americans’ trust in their central government in the past thirty years, trust in state and local governments has remained at high levels ( Jennings 1998). 9. Of course, Swiss cantonalism also entails some diffi culties. Drawing up of appropriate levels of government and winning consent for them is problematic. Securing such consent has much to do with the way in which local identities overlap in complex ways, so as to reduce the prospects of ethnic tension. The success of cantonalism for the Swiss federation is largely due to the fact that “the overlapping boundaries of language and religion…have weakened by language and religion as divisive forces, for each linguistic group contains representatives of both faiths and…vice versa (Dikshit 1975: 234).” Michael Hechter319 Nationalism and Rationality 320 can occur on a practically infi nite number of dimensions (King, 1982). Centralization of expenditure may be the key factor, rather than political decentralization. Much depends on the precise nature of the governing institutions (Habermas 1994), especially the party system.10 Whereas these caveats derive from the historical record, there are also theoretical reasons why decentralization may not have a determinate effect on nationalism. Instead, nationalism may result from path-dependent contingencies that cumulate into “reputational cascades” (Kuran 1998). On this third view, therefore, neither systematic co-optation of local leaders nor devolution ought to have determinate effects on the containment of nationalist confl ict.11 Evidently, the nature of the relationship between federation and nation- alism is contentious. Each view is grounded in a plausible causal mechanism and consistent with at least some of the relevant empirical evidence. Finally, these rival views cannot be distinguished on purely theoretical grounds, for each can be derived from the same instrumental motivational assumptions. Is there any relationship at all between political federation and national- ist confl ict? If so, what might its nature be? some new evidence Until recently, there was no means of assessing the proposition that local decision-making decreases nationalist confl ict. Now, however, two different sources of evidence can be merged to shed some light on the question. The fi rst, Minorities at Risk, consists of a large cross national data set based on newspaper reports of nationalist confl ict since 1970 (Gurr 1993). These data contain variables describing various types of collective action carried out by national groups, as well as measures of the various conditions these groups face in their host states. The second consists of a cross-national data set on Government Finance Statistics collected by the International Monetary Fund that documents the degree to which government revenues and expen- ditures are centralized in a large number of countries. Nationalist confl ict is measured by two indicators of anti-regime activ- ity—rebellion (including political banditry, terrorist campaigns, guerrilla activity and protracted civil war), and protest (including expressions of verbal opposition, symbolic rebellion, and demonstrations). These indicators char- acterize the mean level of rebellion and protest events for all the ethnically distinct groups in a given state. Although there are many different types of centralization, fi scal central- ization is key, for any decentralization that occurs without granting budget- ary power to a sub-unit is well-nigh hollow. The Government Finance Statistics data set contains four variables that indicate the degree of fi scal centraliza- tion in each country. These indicate the revenue collected and expenditures made by each level of government in every country by year. Using these measures, new light is cast on the relationship between decentralization and nationalism (Hechter & Takahashi 1999). First, cen- tralization does indeed have a signifi cant effect on nationalist collective action. Second, protest and rebellion events behave quite differently in these data. Although centralization is positively associated with rebellion events, it is negatively associated with protest. This suggests a possible reconcilia- tion of the two opposing arguments in the literature. Whereas decentralization may provide cultural minorities with greater resources to engage in collective action, leading to a rise in protest events, at the same time it may erode the demand for sover- eignty. Since secession is always an uncertain prospect, and groups tend to be more averse to uncertainty than individuals, this decrease in the demand for sovereignty ought to reduce the incidence of nationalist rebellion.12 10. Suppose the parties are highly centralized. Then, it would seem that all the constitutional and institutional prohibitions guaranteeing constituent governments against revision of the federal bargain would be ineffectual. If, on the other hand, the offi cials of the central government do not have partisan supporters operating the constituent governments, they may expect some opposition to their breaking of the guarantees (Riker 1964). Where there is a political party symmetry between the central government and the sub-units, this should help integrate the federation. Where there is a notable asymmetry between regional (provincial) parties and central parties, where the latter often lack viable affi liates in the sub-units, then this is likely to promote intergroup confl ict (Smith 1995: 9). 11. ”In establishing that small differences between two populations may produce large variations in their observed behaviors, this essay suggests that statistical relationships between aggregate ethnic activity and its determinants are bound to exhibit high standard variation (Kuran 1998: 651).” 12. Secession is a fundamentally uncertain prospect for at least two reasons. In the fi rst place, it is impossible to predict how third parties in the international system Michael Hechter321 Nationalism and Rationality 322 If nationalist groups engage in violent tactics as a means of pursuing sovereignty, then rebellion should be more likely to occur among groups with the greatest opportunity to attain this end (McAdam 1996). Groups concentrated in territories that already have their own governance struc- tures—such as American states, Canadian provinces, or French départe- ments—can make a more plausible demand for sovereignty than groups concentrated in regions lacking a governance structure.13 To determine if this logic holds, the rebellion indicators from the Minor- ities at Risk data set were reconstructed by excluding all non-spatially-con- centrated groups, as well as those concentrated groups whose territory does not coincide with some intermediate-level political boundary.14 The coun- tries in bold italic have at least one minority group that is concentrated in a region with its own governance structure, while countries in regular font lack such a group. Figure 1 clearly shows that centralization has a strong positive effect on nationalist rebellion. These results should not be overinterpreted. Fiscal centralization is an indirect indicator of local decision-making, and both the meaningfulness and comparability of measures of fi scal centralization have been questioned (Bird, 1986). Further, due to data limitations, the number of countries in the analysis is relatively small. Despite these caveats, the consistency of the results is impressive across three decades of recent history. Yet two questions remain. First, since most of the violent nationalism during the 1980s occurs in less developed countries—such as Uganda, the will react to the new entity. For example, separatists argue that a sovereign Quebec would maintain current economic relations with Canada and would also be included in the North Atlantic Free Trade Association, but they have no means of proving these assurances. In the second place, it is impossible to gauge the stability and effectiveness of any prospective new sovereign government. 13. This contention is supported by empirical analyses of the determinants of secessionism in Russia and Eastern Europe (Beissinger 1996; Treisman 1997). 14. The groups included in the analysis are Amazonian Indians in Brazil, Quebecois in Canada, Indigenous Peoples in Colombia, Slovaks in Czechoslovakia, Basques in France, Papuans, Chinese and East Timorese in Indonesia, Palestinians in Israel, South Tyrolians and Sardinians in Italy, Igorots in the Philippines, Basques and Catalans in Spain, Jurassians in Switzerland, Malay-Muslims in Thailand, Acholi and Baganda in Uganda, Scots in the United Kingdom, Native Americans in the United States, and Hungarians in Yugoslavia. 0 0. 51 1. 52 2. 53 3. 54 4. 55 0 0. 2 0. 4 0. 6 0. 8 1 1. 2 C en tr al iz at io n in t h e 19 80 s Rebellion in the 1980s Y ug os la vi a Sw itz er la nd G er m an y C an ad a A rg en ti na U SA B ra zi l A us tr al iaC ol om bi a Sp ai n Fr an ce C ze ch U K B ol iv ia H un ga ry It al y M al ay si a M ex ic o R om an ia U ga nd a P hi lip pi ne s Is ra el C hi le In do ne sia T ha ila nd K en ya P ar ag ua y Ir an F ig ur e 1: T h e re la ti on sh ip b et w ee n c en tr al iz at io n a n d r eb el li on . (C ou nt ri es in it al ic s ha ve a t le as t on e gr ou p th at is c on ce nt ra te d in o ne r eg io n an d "f or m al ly r ec o gn iz ed ") Michael Hechter323 Nationalism and Rationality 324 Philippines, and Indonesia—is the relationship in Figure 1 merely an arti- fact of the overall level of economic development? On the one hand, people may be less inclined to take action against central governments in rich states because they have more to lose from the resulting disorder. On the other, since democracy may be associated with economic development, so, per- haps, is fi scal decentralization. Further analysis reveals, however, that the relationship holds even when each country’s Gross Domestic Product per capita is controlled (Hechter and Takahashi 1999). Second, how robust is the relationship? There is at least one reason to wonder. Yugoslavia’s placement in the extreme southwestern part of the scat- terplot for the decade of the 1980s would seem to imply that this should be the country that is most immune to nationalist rebellion. Yet in the very next decade the country was plunged into a severe and prolonged civil war and the term “ethnic cleansing” entered the English vocabulary. If, as Figure 1 suggests, nationalism is contained by political decentralization, then how can Yugoslavia’s trajectory be accounted for?15 Whereas decentralization inhibits nationalist rebellion, it stimulates nationalist protest. Herein lies a quandary. Decentralization is a spur to mobilization among minority nations, for it places greater resources (espe- cially government jobs) in the hands of national leaders. As long as these leaders see a benefi t in remaining part of the host state, decentralization ought to contain nationalist rebellion. If the central state implodes, however, then it has little to offer peripheral leaders and fragmentation is the likely consequence. This is what happened in the Soviet Union, which split apart on national grounds in a bloodless revolution. The discussion of federation and nationalist confl ict heretofore has been based on the implicit premise that the key dynamics are endogenous to existing political boundaries. But that premise is questionable: time and again nationalism has been strongly affected by exogenous forces. A country that decentralizes as a means of containing nationalist violence is at risk of fragmenting when its center declines due to exogenous shocks such as mili- tary defeat or fi scal crisis. This was Yugoslavia’s sorry fate. Decentralization and Fragmentation in Yugoslavia From 1948 through 1991 Yugoslavia had managed to contain national- ism, despite the disparate interests of its various republics.16 The country was held together by a constitutional order enforced by the resources of the central (federal) state that was explicitly designed to mitigate confl ict between its constituent nations. This constitution aimed to provide equal- ity among republics, as well as security for national minorities within each republic. Its goal was to prevent any single national group from gaining political dominance over the state. Federal policy depended on cooperation from republican leaders, who had the capacity to veto any decision. All federal activities were required to take the proportional represen- tation of individuals by constituent nationality into account. Nationalities were also guaranteed freedom of cultural expression. Individuals retained their national right to self-governance even if they lived outside their home nation’s republic, and the choice of a national identity was voluntary. The manifestation of nationalism, however, was regarded as a threat to the social order and outlawed. As a socialist state, Yugoslavia guaranteed its citizens subsistence, and central and local governments shared responsibility for individual welfare. Public sector employment was the primary source of living standards. The economy was sustained by substantial amounts of foreign aid, largely from the United States, as well as access to foreign credits and capital markets. The basis of this exogenous support was geopolitical, and owed to the regime’s neutrality during the Cold War. During the 1980s, however, a deadly combination of exogenous eco- nomic and political shocks weakened the central government’s ability to maintain this constitutional order. Yugoslavia shared in the worldwide eco- nomic recession of the 1980s. To revive economic growth, the government appealed for assistance from the International Monetary Fund and similar bodies. Some assistance was offered, but only on the basis of commitments that the central government would enact policies promoting economic 15. I owe this question to Adrian Raftery. 16. This section is drawn from (Woodward, 1995), who provides a compelling narrative analysis of the disintegration of Yugoslavia. Michael Hechter325 Nationalism and Rationality 326 privatization and cut public expenditures for welfare, public employment, and social services. The government accepted these conditions, but, as a result, living stan- dards began to decline. Unemployment and infl ation soared. By requiring constitutional revision, the debt-repayment regime turned normal disputes between central and regional governments into constitutional confl icts. The republics best able to adapt to the economic and political reforms of the debt-repayment package—Slovenia and Croatia—sought increasing autonomy from the center. Those that were disadvantaged by these reforms argued for recentralization. The upshot was a constitutional crisis that was carried out between republican leaders seeking to enhance their control over economic and political resources within their territories. Many other countries faced austerity measures in the 1980s but did not suffer Yugoslavia’s fate. Yugoslavia had the misfortune of being the only multinational state that faced another kind of exogenous shock. The cen- tral government’s ability to withstand peripheral nationalism was dealt a severe blow by the abrupt and unanticipated end of the Cold War. Yugo- slavia had profi ted greatly from its neutrality in the Cold War. The demise of the Soviet Union sharply decreased the country’s strategic value to the United States, however. It also ended forty years of American-backed guar- antees of fi nancial assistance and support for Yugoslav independence and integrity. This combination of exogenous economic and political shocks so weak- ened the central government that its ability to contain nationalism was effectively destroyed. Although the center attempted to prevent the seces- sion of Slovenia and Croatia by force, it no longer had suffi cient resources to prevail. Here, too, exogenous forces played a signifi cant role; Germany’s recognition of the sovereignty of Slovenia and Croatia spelled Yugoslavia’s fi nal chapter. Once the constitutional guarantees for minority rights were null and void, there was little to restrain intergroup violence. At least two important lessons can be learned from the Yugoslav case. On the one hand, its complex decentralized constitutional provisions man- aged to contain nationalism for four decades; this is no mean feat. Whereas the constitution was designed to keep the country territorially intact, its extreme decentralization made it diffi cult for the center to adapt to exog- enous economic shifts. On the other, the Yugoslav federation relied too heavily on the country’s strategic position in the Cold War. When—against all expectations—the Cold War ended, the center’s resource base was sub- stantially diminished. Clearly decentralization can proceed so far that it courts fragmenta- tion, which can be another source of intergroup violence. The relationship between decentralization and nationalist violence, therefore, is likely to be U-shaped rather than linear. If too little decentralization causes rebellion, then too much is likely to engender fragmentation. To contain nationalist violence, thus, a balance must be struck between peripheral regions’ dependence on the center for military and economic resources, and the autonomy that allows them to pursue their own interests. Whereas federation is no panacea for nationalist violence in relatively centralized states,17 it does offer substantial hope for mitigating nationalism’s dark side. conclusion Containing nationalist violence is a pressing social issue in the contem- porary world. If this kind of violence largely emerges from irrational roots, then there is little hope of containing it. I have argued, however, that the 17. Some nationalist violence continues to occur in spite of decentralization. A virtual natural experiment has been going on the Spanish Basque and Catalan regions, both of which developed strong nationalism since the death of Franco. Despite the high levels of fi scal and political self-governance granted to these regions by recent Spanish constitutional reforms, Catalan nationalism has been notably peaceful, while ETA in the Basque country is among the most violent nationalist organizations in the world. Differences in social structure between the two regions may help account for the difference in nationalist violence (Diez-Medrano 1995). Whereas both Basque and Catalan capitalists opposed secession because of their economic dependence on the Spanish state, initial economic development in the two regions differed markedly. The Basque region specialized in the production of capital goods (steel, shipbuilding and fi nancial services), whereas Catalonia specialized in the production of consumer goods (textiles). As a result, the Basque bourgeoisie was smaller, more concentrated, and more dependent on the Spanish state and markets than its Catalan counterpart. Because they were less dependent than their Basque counterparts, the Catalan elite was more nationalist. This meant that Catalan nationalism had a much broader base of support than its Basque counterpart. Because the Basque bourgeoisie was anti-nationalist, Basque nationalists were fi ercely anti-bourgeois. Not so for Catalan nationalists. As a result, Basque nationalism was far more politically extreme than Catalan. Michael Hechter327 Nationalism and Rationality 328 preponderance of nationalist violence seems to have strategic roots, and therefore can be regarded as the outcome of individually rational action. This means that certain kinds of social institutions can provide incentives that should contain nationalist violence. But what kinds of institutions will do the trick? General theory can tell us little about the answer to this key question. Theorists fundamentally disagree about the effects of federation on nationalism: some think that federation will exacerbate it, others think it will inhibit it. The best avail- able evidence shows that whereas federation stimulates nationalist political mobilization, it decreases nationalist violence. This is an optimistic conclu- sion, at least for everyone interested in containing nationalism’s dark side. references Althusius, Johannes. 1964. Politics. Boston: Beacon Press. Beissinger, Mark R. 1996. “How Nationalisms Spread: Eastern Europe Adrift in the Tides and Cycles of Nationalism Contention.” Social Research, 63:97-146. Birch, Anthony H. 1989. Nationalism and National Integration. London: Unwin Hyman. Bird, R. M. 1986. “On Measuring Fiscal Centralization and Fiscal Balance in Federal States.,” Environment and Planning, C: Government and Policy, 4:389-404. Boudon, Raymond. 1996. “The ‘Cognitivist Model’: A Generalized Rational-Choice Model,” Rationality and Society, 8:123-50. Connor, Walker. 1993. Ethnonationalism. Princeton: Princeton University Press. Diez-Medrano, Juan. 1995. Divided Nations: Class, Politics and Nationalism in the Basque Country and Catalonia. Ithaca, NY: Cornell University Press. Dikshit, R. 1975. The Political Geography of Federalism: An Inquiry into its Origins and Stability. Delhi: Macmillan. Elster, Jon. 1989. The Cement of Society. Cambridge: Cambridge University Press. Fearon, James D. 1994. “Ethnic War as a Commitment Problem,” in Annual Meeting of the American Political Science Association. New York. Feldman, Allen. 1991. Formations of Violence. Chicago: University of Chicago Press. Forsythe, Diana. 1989. “German Identity and the Problem of History,” in History and Ethnicity, edited by Elizabeth Tonkin, Malcolm Chapman, and Maryon McDonald. London: Routledge. Gargan, Edward A. 1993. “Suicide Bomber Kills President of Sri Lanka,” in New York Times. Geertz, Clifford. 1994. “Primordial and Civic Ties,” in Nationalism, edited by John Hutchinson and Anthony D. Smith. Oxford: Oxford University Press. Glazer, Nathan. 1977. “Federalism and Ethnicity: The Experience of the United States,” Publius 7:71-87. Glenny, Misha. 1993. The Fall of Yugoslavia. New York: Penguin Books. Gurr, Ted Robert. 1993. Minorities at Risk: A Global View of Ethnopolitical Confl icts. Washington, D. C.: United States Institute of Peace Press. Habermas, Jurgen. 1994. “Citizenship and National Identity,” in The Condition of Citizenship, edited by Bart van Steenbergen. London: Sage. Hechter, Michael. 2000. Containing Nationalism. Oxford and New York: Oxford University Press. Hechter, Michael, and Nobuyuki Takahashi. 1999. “Political Decentralization and Nationalist Confl ict,” Paper presented at the Annual Meetings of the American Sociological Association. Chicago, IL. Horowitz, Donald L. 1985. Ethnic Groups in Confl ict. Berkeley: University of California Press. Jennings, M. Kent. 1998. “Political Trust and the Roots of Devolution,” in Trust and Governance, edited by Valerie Braithwaite and Margaret Levi. New York: Russell Sage Foundation. Kaiser, Robert. 1994. The Geography of Nationalism in Russia and the USSR. Princeton: Princeton University Press. Kaiser, Robert J. 1990. “The Equalization Dilemma in Yugoslavia,” Geoforum, 21:261-76. King, Preston. 1982. Federalism and Federation. London: Croom Helm. Kollock, Peter. 1998. “Social Dilemmas: The Anatomy of Cooperation,” Annual Review of Sociology, 24:183-214. Kuran, Timur. 1998. “Ethnic Norms and Their Transformation through Reputational Cascades,” Journal of Legal Studies, 27.:623-660. Laitin, David D. 1998. Identity in Formation: The Russian-Speaking Populations in the Near Abroad. Ithaca: Cornell University Press. Mamdani, Mahmood. 1996. Citizen and Subject: Contemporary Africa and the Legacy of Colonialism. Princeton: Princeton University Press. McAdam, Doug, John D. McCarthy, and Mayer N. Zald. 1996. Comparative Perspectives on Social Movements: Political Opportunities, Mobilizing Structures, and Cultural Framings. Cambridge [England]; New York: Cambridge University Press. McGarry, J., and Brendan O’Leary (Eds.). 1993. The Politics of Ethnic Confl ict Regulation. London: Routledge. Meadwell, Hudson. 1993. “The Politics of Nationalism in Quebec,” World Politics, 45:203-41. Mozjes, Paul. 1994. Yugoslavian Inferno: Ethnoreligious War in the Balkans. New York: Continuum. Oates, Wallace E. 1972. Fiscal Federalism. New York: Harcourt Brace Jovanovich. Ostrom, Vincent, Charles Tiebout, and Robert Warren. 1961. “The Organization of Government in Metropolitan Areas: A Theoretical Inquiry,” American Political Science Review, 55:831-842. Popper, Karl R. 1994. “Models, Instruments and Truth,” in The Myth of the Framework: In Defence of Science and Rationality, edited by Karl R. Popper. London: Routledge. Riker, William H. 1964. Federalism: Origin, Operation, Signifi cance. Boston: Little-Brown. Roeder, Philip G. 1991. “Soviet Federalism and Ethnic Mobilization,” World Politics 43:196-233. Michael Hechter329 Sale, Kirkpatrick. 1980. Human Scale. New York: Coward, McCann & Geoghegan. Savigear, Peter. 1989. “Autonomy and the Unitary State: The Case of Corsica,” in Federalism and Nationalism, edited by Murray Forsyth. Leicester: Leicester University Press. Smith, Graham. 1995. “Mapping the Federal Condition: Ideology, Political Practice and Social Justice,” in Federalism: The Multiethnic Challenge, edited by Graham Smith. London: Longman. Tambiah, Stanley J. 1996. Leveling Crowds: Ethnonationalist Confl icts and Collective Violence in South Asia. Berkeley: University of California Press. Treisman, Daniel S. 1997. “Russia’s “Ethnic Revival”: The Separatist Activism of Regional Leaders in a Postcommunist Order,” World Politics 49:212-49. Weingast, Barry R. 1995. “The Economic Role of Political Institutions: Market- Preserving Federalism and Economic Growth,” Journal of Law, Economics and Organization 11:1-31. Woodward, Susan L. 1995. Balkan Tragedy : Chaos and Dissolution After the Cold War. Washington, D.C.: Brookings Institution.