Separatist Conflict In Eurasia and Beyond

Separatist Conflict In Eurasia and Beyond: Exploring Communist Bloc Exceptionalism Benjamin Smith Associate Professor Department of Political Science University of Florida Box 117325 Gainesville, FL 32611 USA [email protected]

Abstract In the two decades since the collapse of the Soviet Union, the study of separatist conflict has been dominated both geographically and theoretically by post-communist cases (see for example Hale 2008; Jenne 2007; Roeder 2007; Toft 2003). Frequently, however, the efforts to extend arguments based on the post-communist experience to the post-colonial world come up short. I argue in this essay that the lack of fit is due largely to the mechanisms by which coherent, durable ethnic identities emerged and were contested inside and outside the communist bloc. Inside the communist bloc, ethno-federalism combined with heavy-handed direct rule generated or reinforced ethnic identities at the titular republican level. On the contrary, ethno-federalism in the post-colonial world (i.e. India and Ethiopia) has generally been a post-colonial state response to early separatist mobilization rather than a cause of it. Moreover, the same early post-colonial rebellions against direct rule have catalyzed long-term trajectories of contentious ethnic politics, but did so only in the context of vertical ethnic ties that enabled collective action in the absence of political organizations. Drawing on two different global datasets and a number of original variables, I show a modest cross-regional difference in the role of past rebellion in shaping separatism and a strong cross-regional difference in the effect of ethno-federalism on nation-state crisis inside and outside the former communist bloc. Where prominent post-Soviet studies find ethno-federal institutions to be at the “front” of their explanations, I suggest that, in the post-colonial world, such institutions have nearly always been endogenous to nationalist mobilization. I owe thanks to Michael Bernhard, Cynthia Horne, Bryon Moraski and Conor O’Dwyer for helpful comments on this essay. NOTE: This is a draft. Please do not cite or quote with permission of the author. Comments and critiques are enthusiastically welcome, however.

Prepared for presentation at the annual meetings of the American Political Science Association, Washington DC, September 2-5, 2010.

Smith

Separatist Conflict In Eurasia and Beyond: APSA 2010

Introduction It is understandable how central analysis of the collapse of the Soviet Union and Yugoslavia have become to the empirical study of separatist violence. Yugoslavia alone has seen more independent states emerge from its ashes since 1995 than in the rest of the world, and the 1990s saw the emergence of fifteen independent states from the former Soviet Union. However, scholars have yet to reach consensus on the value of the comparative lessons of Eurasian separatism for the rest of the world. On the one hand, scholars have concluded that different determinants of secession appear to be at work inside and outside Eurasia (see for example Hale 2008, 250). On the other, a theory of “segment-states” leading to nation-states derived from the Soviet experience arguably explains “where nation-states come from” around the globe (Roeder 2007). It appears that the reasons for Soviet and Yugoslav state disintegration do not necessarily hold universal lessons for those of us interested in separatist violence elsewhere. This paper takes some steps to disentangle the study of separatism from Eurasia. It does so by analyzing aggregate data at the ethnic minority group level between 1946 and 2005, employing a new dataset on Ethnic Power Relations and one originally used by Roeder (2007). In doing so it advances two main arguments. First, the historical determinants of separatist mobilization in the communist bloc are not solely a function of ethno-federalism. My analysis of pre-incorporation rebellions against Russia or the Soviet Union and of the role of brief periods of independence indicates that a legacy of resistance is as important for explaining today’s Eurasian separatism as it is for the post-colonial world. Indeed, the causal force of those past rebellions appears to have durably survived seven decades of Soviet rule. Seen in that light, ethno-federalism during the communist era is a key factor, but not the only one, in generating coherent and durable ethnic identities at the titular republican level. Those identities— 1

Smith


shaped in part by historical experience—were reinforced by Soviet nationalities policy, combined with less actual autonomy than that policy promised, and reemerged in a serious way in the context of central collapse. Second, what Suny (1993) terms “pseudofederalism” marks the post-communist cases off in some important ways from much but not all of the rest of the world. Sub-state ethnic autonomy in Eurasia was largely a cause of later nationalist mobilization: in the post-colonial world, by contrast, it was a result of it. These reversed sequential trajectories both confound a global theory-building ambition and suggest that a more nuanced approach to studying separatism might provide reciprocal insight both for scholars of Eurasia and for those of us focused on the post-colonial world. The paper proceeds as follows. First, I discuss the extant literature on Eurasian separatism and on the historical origins of nationalist mobilization. As I outline, history matters to separatism in Eurasia but in a way that differs from the post-colonial world. In the second section of the paper I develop a theory of historically generated ethnic identity based on non-Eurasian cases and contrast it with the communist bloc. In the third section, I employ two different sets of data to assess the long-term impacts of earlier rebellions and of ethno-federal institutions in shaping contemporary ethnic politics in the former communist bloc and in the post-colonial world. First, I employ data from the Ethnic Power Relations project to explore the roles of historical rebellion legacies and central state policies. I then replicate and elaborate the findings of Philip Roeder’s Where Nation-States Come From using original measures to parse out the effects of ethno-federalism inside and outside Eurasia. The results demonstrate a strong rebellion legacy effect in the post-colonial world and the lack of one in the communist bloc. They also suggest something different about the effects of ethno-federalism inside and outside the former communist bloc. Finally, I conclude by emphasizing the differing dynamics that have shaped the success and failure of ethno-nationalist movements inside and outside Eurasia, and how a

2

Smith


long-range historical framework reveals how institutions are often endogenous to contentious ethnic politics. I. Separatist Conflict and the Collapse of the Communist Bloc: the State of the Art In significant part because of the numerical predominance of the Soviet bloc’s disintegration— more than 20 new states from the USSR and Yugoslavia alone versus two in the rest of the world since 1990—scholars of the region’s conflicts have written the lion’s share of books and articles on comparative separatism (see for example Ayres and Saideman 2008, Beissinger 2002, Bunce 1999, Hale 2008, Jenne 2007, and Roeder 2007). In this section I synthesize their major conclusions and theoretical insights, highlighting as I go where those findings differ in important ways from the study of separatism elsewhere. As we shall see, there appear to be differences between the former communist bloc and the rest of the world, across particular questions, samples and dataset delimitations. Moreover, those differences do not yet gel into a consensus: we simply don’t yet have much agreement on why separatist conflict looks so different from the rest of the world. But what do scholars generally think explains why some ethnic groups rebelled against central states in the Soviet block and others did not? In particular here I point to the independent effect of communist bloc institutions (especially in Czechoslovakia, the Soviet Union and Yugoslavia) in reinforcing powerful ethnic identities and interests. A corollary to this point is a general trend to deemphasize preexisting ethnic identities, and more concretely to suggest that pre-socialist history was less important in catalyzing ethnic identity formation than what took place under socialist rule.

Soviet Bloc Institutions and the Collapse: the Peril of Ethno-Federalism under Dictatorship Valerie Bunce (1999) has gone perhaps further than any other scholar of the communist world to argue that political institutions spelled the demise, and the shapes it took, of the Soviet bloc. The 3

Smith


disintegration of the Soviet Union, Yugoslavia and Czechoslovaki, she argues, were a result of the tensions between national federalism (sub-state units defined in ethnic republican terms) and socialist dictatorship. Put a bit more simply, the promise of ethnic self-determination coupled with heavy-handed socialist centralism built a self-destruct mechanism into these systems, explaining why they came apart in the early 1990s while their unitary Soviet bloc counterparts remained whole. The mobilizing impetus of national history—by which I mean a past legacy of conquest by imperial powers and/or rebellion against it and whether it plays a part in contemporary nationalist narratives—comes into play in Bunce’s narrative primarily in the cases of the Serbians (93-94). Other republics in these three federal socialist states do not invoke the same reference to an historical trajectory of resistance. Henry Hale (2008) makes a similar argument but endeavors to parse out why it was that some Soviet republics seceded before others rather than why some socialist states disintegrated. Building first a theory of ethnic identity and of ethnic politics (centered on cognitive uncertainty reduction and interest protection, respectively), Hale argues that republics’ respective economic status relative to the Soviet mean posed different incentives. Richer republics had good reason to fear being exploited in a restored union, while poorer ones had strong incentive to support unionism for the advantages it could provide them. In his framework national federalist institutions also mattered greatly because they “effectively resolve the microlevel collective action problems and thus isolate the macrolevel problem” (63). Here again it is ethnofederalism playing the central role in ethnic politics—by casting elite and mass interests at the titular ethnic republican level and thereby making decisions a function of the question: what is best for my (republican) group and therefore for me? The answer drives nationalist mobilization: the “politics of separatism is fundamentally about shaping regional beliefs regarding the nature of the (potential) central government” (72). To take one example that I discuss more below, however, why was Azerbaijan, a much poorer republic than the Soviet average in the 1980s, the first non-Baltic one to 4

Smith


declare independence? Resource wealth is an important part of the answer—Azerbaijanis knew their republic was oil-rich but that Moscow siphoned off most of the revenues from “their” oil—but in Hale’s framework we would predict Azerbaijan to be strongly integrationist. Roeder (2007) in one important sense stands out from Bunce and Hale in his keen focus on where states come from—not where separatist movements come from, but why some succeed in their ambitions to state-ness. The answer is institutions: ethnic groups with segment-state institutions are more likely to become nation-states than ones that lack them. Still, his concept of the “segment-state” hews closely to both national federalism and ethnofederalism as the central explanant. It also effectively limits the analytic gaze to groups that make an explicit claim to independence aspirations. This is understandable in the context of international relations theory, but it does run the risk of truncating the possible sample by excluding groups that have not, or have not primarily, sought independence through nationalist mobilization. And, Roeder too minimizes the prospect of “nations before nationalism”1 as a prior resource for republican elites, contending that their established autonomy from common-state leaders allows them to consolidate control over ethnic politics and identity. Ethnic identity here is largely elite-driven: indeed, “The public at large is likely to be indifferent to the choice between the elite’s projects and the status quo” (2007, 81). The reason that publics follow despite this likely indifference is identity hegemony—i.e. republican identity appears to be the only game in town given the structure of the system. One might reasonably ask why publics ally with elites outside the context of ethno-federalism, as they so often do in the post-colonial world. Dmitry Gorenburg (1999) goes even further, illustrating how in four ethnic republics of Russia ethnic elites consciously used separatist activism to garner institutional and material resources, which

1

This phrase borrowed from John Armstrong, Nations Before Nationalism, (Chapel Hill: University of North Carolina Press, 1982).

5

Smith


were in turn used precisely to expand nation-building efforts. In other words, ethnic leaders in Tatarstan, Bashkortostan, Chuvashia and Khakassia sought substantial federal autonomy for their regions precisely in order to devote the new resources to enhancing (in Tatarstan) or to outright national identity construction efforts (in the other three republics). Gorenburg’s work illustrates just how salient national federalism—in these cases as aspirational examples—has come to be in Eurasia: “these programmes were implemented because titular governing elites were conditioned by Soviet nationalities policy to believe in the importance of ethno-cultural development. Soviet policy was based on a primordialist reification of ethnic groups” (269, emphasis added). These four scholars capture what appears to be the largest vein of argument in post-Soviet bloc separatism studies: namely that socialist institutions tell us much of what we need to know in order to understand who seceded and why during the disintegration of the Soviet bloc.

Ethnic Identity Before Soviet Rule? Contrasted with scholarly voices pointing to Soviet institutions as central to nationalist mobilization in the 1980s and 1990s are a smaller group that suggest more variation: not that all Soviet nationalities were well formed up by the Revolution but that some were. Ronald Suny’s work on national identity in the Soviet space stands in marked contrast to the others. His seminal comparative work on Soviet nationalities, Revenge of the Past, charts out the varying shapes and strengths of national identity at the onset of Soviet rule, then shows how Stalin’s correctives to Lenin’s policies catalyzed still further variation. In Table 1 I present a numerically coded version of Suny’s classifications alongside Hale’s coding of the republics’ separatist activism in the 1990s. Table 1 about here

6

Smith


What is notable about comparing these two sets of measurements is that Suny’s pre-Soviet national identity scores and “downgrade” measures bear no apparent relation to the strength or earliness of a republic’s propensity to declare sovereignty in the late Gorbachev era. In short, past history does not appear to have generated much difference. Once it became evident that the Soviet Union was likely to collapse, republican defection arguably looked less like a systematic parade of sovereignties and rather more like a collective rush for the door. Clearly “history” mattered to the collapse of the Soviet bloc—and in particular the disintegration of Czechoslovakia, the USSR and Yugoslavia. But how it mattered appears to differ from how history matters elsewhere in the world. There is a loose consensus among scholars of the post-communist world that, on balance, national identities in these three states were substantially a function of state policies. Where there were coherent pasts on which to draw—and the brief but much-played interwar independence period for the Baltic states (and Azerbaijan) stands out among these cases—they became touchstone national narrative moments that grounded the national imagination in actual past events. What these historical touchstones did not do, however, was to generate any real difference in the timing, scope or intensity of nationalist mobilization during the process that culminated in the Soviet collapse. Following the next section I explore the impact of a rebellious past on separatist conflict around the world, contrasting global trends with Eurasian ones.

II. History and Separatism: The Origins of Nationalist Rebellion Inside and Outside Eurasia In the post-colonial world in the 20th century, state policy has more often than not followed national identity and nationalist mobilization. In other words, where scholars of the post-communist world see ethno-federalism as constitutive of national identity, the post-colonial world has seen more ethno-federalism arise as a result of nationalist mobilization. In short, there exist two main trajectories 7

Smith


of secessionist or separatist conflict in the modern world—one tied to resistance to direct rule (state building) and the other stemming from the collapse of communism.2 The second trajectory is most recently illustrated by the collapse of socialism and proliferation of new states in what had been Czechoslovakia, The Soviet Union and Yugoslavia.3 As I have outlined above, it took the shape of opportunistic moves toward independence by titular ethnic groups—led by republican elites—who had inherited the “tools” of nation building from the socialist-ethnofederalist experience. History, in this trajectory, was largely institutional since these ethnic minorities had not had to build from scratch the elements of a national narrative. That narrative and the tools to build it were instead provided by the center. The other trajectory—applying to much of the rest of the world—was not characterized by a Marxist take on the national question. Rather, it was a Weberian one: a result of ethnic minorities’ early independence-era resistance to first-generation state builders. Struggles for nation-state status began as uprisings against efforts by state leaders to impose direct rule on their territories. Central to the viability of today’s separatist movements,4 subsequently, is the history of the group they speak for. Whether contemporary separatists have such a legacy depends substantially on the conditions that determined whether their predecessors could successfully mobilize a mass following in the early

2

There is debate, of course, over whether the Soviet collapse ought to be called “imperial collapse.” See for example Suny (2001). For discussions of imperial collapse and direct rule, see Wimmer 2002, 2008; Wimmer, Cederman and Min 2009, 2010; Wimmer and Min 2006; and Hechter 1992, 2000. 3 See Hale 2008, Roeder 2007 and Toft 2003. Under conditions of imperial collapse, secession is a virtual certainty; the real question is how boundaries get formed in the aftermath and in what form the secessions take place. This is especially the case in ethnofederal empires such as the Soviet Union—what took historically lengthy rebellions elsewhere to accomplish in identity formation was inadvertently accomplished in the Stalinist and post-Stalinist periods. For Stalin’s own prescient take on institutions and identity formation, see “Marxism and the National Question.” 4 I use this term to refer to movements representing ethnic groups that make a range of demands for selfdetermination, ranging from language and cultural autonomy to greater representation in local and national governance to outright independence. To a large degree my conception of mobilization tracks with Treisman’s (1997) conceptualization of “separatist activism.”

8

Smith


independence period.5 That, I argue, is largely a question of whether vertical social structures in minority regions are co-ethnic: are peasants and landed nobility or religious elites from the same ethnic group? When the answer is yes, minority regions like Kurdistan, Aceh, Punjab and Eritrea were much more likely to have rebelled against first-generation state builders who sought to turn Weberian ideals about territorial control into realities of state power. The key dynamic here was that traditional elites could employ the social structures atop which they stood to mobilize peasants to fight central state authorities. When such rebellions took place, state leaders typically responded with force. Paradoxically, their success at doing so frequently created a trajectory leading to future conflict by catalyzing social change, namely urban migration. Where early state builders responded aggressively, first by conquering rural rebellions and weakening local elites, and second by promoting development in minority regions, they both undermined the mode of mobilization and induced urban migration in those regions. The defeat of rural elites and transformation of agricultural production from social-coercive and local to market-driven catalyzed urban growth in minority regions. When such regions had large and minoritydominated urban centers (such as Sulaymaniya in Kurdish Iraq), it was to those cities that peasants migrated, creating new and larger minority populations in those cities. When they did, secondgeneration nationalist elites did as much recruiting in cities as they did in the countryside, effectively sidestepping state coercion. In essence modernization produced in such cases a cross-generational bridge linking early rebels to later ones. And, over time, they became the locus of second-generation mobilization by new elites, who ironically sought to distinguish themselves from their conservative

5

That period varied, of course, depending on the post-colonial region. Generally, it took place in the 1930s, 1940s and 1950s in the Middle East, in the 1950s and 1960s in Asia, and in the 1960s and 1970s in Africa.

9

Smith


predecessors but who nonetheless drew on those past rebellions to construct historical narratives of resistance. Anderson’s (1991) study of nationalism, to my mind, often takes unwarranted flack for proposing a concept—the imagined community—that is argued to be insufficiently concrete as an explanatory force. However, it is probably reasonable to criticize the use of this concept at the macrolevel. Conceptually speaking, shared memory has to manifest at the individual level—where individuals actually share such memories. A demonstrable history of past rebellion marks off a group politically and culturally and gives it, to tweak Anderson’s language, both a tangible past and an imaginable future. It both strengthens the appeals of ethnic minority elites and lowers the perceived costs for potential recruits. In short, I argue here that the “belief in common descent” that has been central to defining ethnic identity since Weber (1978, 389) postulated it has concrete causal force.6 This way of looking at history as a mobilizing resource for ethnic elites also helps to move us beyond a narrower dichotomous focus on greed and grievance, to encompass perceived viability. That is, an ability to imagine a community as encompassing shared history and therefore a shared future is substantially more feasible—and therefore more rational as a potentially costly choice for individuals—when it can be historically demonstrable.7 In short, prior rebellions by organizations speaking for an ethnic group give the group’s members not an imagined community but an actual one, and one with a history of demanding political recognition.

6

Confirming this individual-level facet of the argument is beyond the scope of this essay and I simply seek to establish its plausibility here. But the implication, and I draw on a lengthy Weberian tradition in making it, is that this macrohistorical argument has its causal roots in individual beliefs. The book project of which this paper is a part will also incorporate several chapters on the construction of shared history in Aceh, Indonesia, drawing on interviews and then a survey of former Aceh Independence Movement (Gerakan Aceh Merdeka, or GAM) fighters and supporters. 7 I say potentially costly because participation in all rebellions poses likely costs—repression, state attack on one’s village or region, and of course the possibility of death. Kalyvas and Kocher (2007), however, illustrate that the costs of participation in rebellion do not always necessarily exceed those of free-riding.

10

Smith


Ethnonationalist rebellion in the modern era generally emerged against post-colonial or postimperial states, during the immediate post-independence period in which early state builders sought to impose direct rule on the territories they claimed to control (Hechter 2000, 28, 73-78). At many postcolonial states’ founding moments, institutions were relatively weak, the economies were largely agricultural, and social power lay mostly in the hands of rural elites. When states attempted to impose direct rule, rural elites rebelled by mobilizing peasants over whom they had substantial social and economic power. The state reactions to those first-generation rebellions shaped what came later. Where states first defeated these rebellions and then set about transforming minority regions by force, the frequent result was massive migration into cities. While this had the effect of weakening the base of peasants that rural elites could mobilize, it also created a second minority population base, this time an urban one. Hence it is unsurprising that, where Sheikh Said and other Kurdish elites in early republican Turkey came from the rural aristocracy, the leaders of later, more radical organizations like the Kurdish Workers’ Party (PKK) recruited at least as much in cities as in the countryside (Marcus 2007). Clearly the framework I developed in the above paragraphs contrasts with the conventional understanding of the roots of separatism in the post-communist world (and especially the former Soviet Union, Czechoslovakia and Yugoslavia). Two very different historical trajectories into the contemporary world are outlined—one by central policy privileging ethnicity as an organizing identity and one by group rebellion. But does this difference hold empirically? Is it the case that past rebellions have a different explanatory role inside and outside Eurasia? It is to this question that I turn in the next two sections.

III. History Matters: Rebellion Legacies in Eurasia and the Rest of the World In this section I test the hypothesis that past rebellions strongly influence the viability of contemporary mobilization by organizations representing ethnic groups. Given the robustness of the 11

Smith


“history” effect across multiple samples, scholars, measurement choices and methods it is appropriate to take its soundness as an empirical starting point. I seek to test one main hypothesis: holding state policy and level of development constant, H1: a legacy of past rebellions makes current rebellions by ethnic minority groups more likely. In the paragraphs that follow I discuss the data and methods used to assess this hypothesis.

Data, Methods and Models I employ data from the Ethnic Power Relations (EPR) project,8 first using a sample from the postcolonial world, then in Eurasia. The dataset covers all politically relevant ethnic groups in the world between 1946 and 2005 in 155 states with a population of at least one million and an area of at least 500,000 square kilometers. I use EPR data rather than data from the Minorities at Risk project for a number of reasons. First, the time period is more expansive, allowing for considerably more coverage of post-communist ethnic politics. Second, by including both dominant and excluded groups (especially important in cases where an ethnic group might be dominant in one country but discriminated against in another, such as Russians in Russia vs. Estonia) rather than only those at “at risk,” discrimination becomes a true variable rather than a selection criterion for inclusion in the dataset. Finally, the question of both the effects of ethno-federalism and of post-independence state assaults on ethnic minority regions are ones of state policy: I want to understand, holding state policy constant, what the effect is of rebellion legacies. The Cederman et al EPR data project includes multiple measures that capture with considerable finesse different kinds of state ethnicity policies. As I discuss below, these measures compliment the institutional variables in Roeder’s dataset.

8

The data are available publicly at http://dvn-iq.harvard.edu/dvn.dv/epr.

12

Smith


As I discuss below, I estimate the basic models using two different subsets of the data: Eurasia (all of the former Soviet Union and Soviet bloc) and the post-colonial world (Latin America, Asia minus former Soviet republics, North Africa and the Middle East, and Sub-Saharan Africa). One other point is important: for the non-Eurasia cases (groups), the history effect variable only covers violent conflicts dating back to 1946, the starting year of the dataset. In order to capture possible pre-Soviet dynamics of historical legacies, I extend the time period of relevant conflict legacies back to 1818, an additional 128 years. Given the carefully detailed arguments by Altstedt (198X), XXX and others that republics like Azerbaijan engaged in a “continual pattern of struggle against Russian colonial rule from the original conquest to the late twentieth century,” I wanted to grant maximum leeway to possible historical legacy effects in Eurasia since the start point of the EPR data excludes completely the pre-Soviet era. I discuss my strategies for constructing these variables below.

Dependent Variable: Separatist Rebellion The dependent variable (secethonset) whose determinants I estimate here is the onset of a separatist rebellion that resulted in at least 25 battle deaths per year. As such, this sort of conflict takes place between an organization claiming to represent an ethnic group and a central government, and over ethnic separatist claims. It is drawn from the Armed Conflict Dataset at the Peace Research Institute in Oslo (Gleditsch et al 2002).9

9

The original PRIO ACD data are available at: http://www.prio.no/CSCW/Datasets/Armed-Conflict/UCDP-PRIO/. Here I am using the data as coded in Cederman, Wimmer and Min (2010) to whom I owe thanks for sharing them. At present I am incorporating data from the Minorities at Risk project’s rebellion measures to include an ordered variable for different kinds of ethnic mobilization (from autonomy or rights claims up to violent rebellion). Despite the MaR dataset’s limitations by selection (i.e. only including minorities discriminated against), this more nuanced measure of mobilization should prove useful alongside the EPR dataset’s state policy variables.

13

Smith


Independent Variables: Historical Legacies of Past Rebellion This set of analyses seeks to establish simply 1) whether there is generally an historical legacy effect shaping the likelihood of contemporary separatist rebellions in the world and 2) whether the legacy effect differs inside and outside Eurasia. To that end, the independent variables with which I am most concerned deal with the number of times in the past that an ethnic group has mobilized violently against its central government (or against an imperial center in the Eurasia cases). The measure for the largest sample (Warhist), and for the post-colonial world, simply adds up the onset years from the main dependent variable. However, as mentioned above, the pre-Soviet era is not included in the time period of the EPR data (which begins in 1946). Noting that the analyses of Suny (1993), Altstedt (198X) and others point to resistance and rebellion well before that, and in some cases back into the 19th century as markers of national identity and resistance, I employ an additional set of legacy variables in an attempt to capture theoretically meaningful periods in time. First, I include a variable (pre1848_reb) for antiimperial wars in the Ottoman, Austro-Hungarian and Russian territories of Eurasia before 1848 to capture those that took place before the nationalist revolutions in that year. This Second, I include a variable (natlism_to_indep_reb) that captures rebellions between 1848 and the brief independence periods of a number of Soviet republics following the 1917 revolution. Third, I include a variable (indep_to_ww2_reb) for rebellions between 1918 and 1939. In each of these, rebellions are coded at the group level (i.e. by ethnic groups) at “1” if a rebellion took place during the period in question and “0” otherwise.

Control Variables In addition to standard controls for level of economic development (lgdpcapl), group size(lrsize), and country population(lpopl) (I use the natural logarithm of all three measures), the models I estimate 14

Smith


here data contain a number of key variables from EPR to account for state policy toward ethnic groups. Everywhere, but especially in the context of Eurasia, where state policy was so central to the development and evolution of ethnic identity and politics, it is especially important to capture variation in the ways that central governments treated their ethnic minorities. First, an ethnic group’s powerless status indicates that “Elite representatives hold no political power at either the national or the regional level without being explicitly discriminated against” (Cederman et al 2010, 100). Second, a group’s discriminated status indicates that “Group members are subjected to active, intentional, and targeted discrimination, with the intent of excluding them from both regional and national power. Such active discrimination can be either formal or informal” (101).Third, I start by including variables for political autonomy: autonomy when a group’s status is center-granted and separatist when the group itself declares its independence from the center. As I discuss below, the separatist autonomy variable appears to be causally entangled with its past history of rebellion: for that reason, and because it indicates a group-level move rather than a state-level move, it is included only in the initial models. Finally, I include a variable for ethnic “downgrading” (downgrade), indicating that a group’s status was reduced at the center in the prior year.10 Because the dependent variable here—separatist conflict onset—is dichotomous, I employ logistic regression. I exclude ongoing war years following Cederman et al (2010) because of the likely different logics during wartime, thus the models estimate the likelihood of a separatist war beginning during peacetime. Finally, the results reported here estimate robust standard errors clustered by country to account for within-country dependence.

10

Note: I am at work incorporating additional control variables for ethnic regions’ mountainous territory, for ethnic population concentration and for rural/urban population shares. Including these variables will bring the data very close to including all of the usual suspect factors that other scholars (e.g. Fearon and Laitin 2003, Sambanis 200X) have consistently found to be significant determinants of separatist conflict.

15

Smith


Table 2 about here

The Post-Colonial World The results are presented in Table 2. Models 1 and 2 in Table 2 present results for the postcolonial world. The results here strongly confirm prior works that point to the importance of past conflicts in shaping the prospects of new ones. The rebellion legacy strongly increases the likelihood of violent mobilization across the post-colonial world, as Cederman et al (2010) found that it did in a global sample.11 And, again a number of state ethnic policy variables are robustly significant—exclusion, downgrading, and discrimination—in predicting separatist rebellion. Junior partner status, unsurprisingly, but more surprisingly autonomy are both insignificant. As I discuss below in reference to employing Roeder’s measures, the significance of “autonomy” may be in part dependent on measurement and disaggregation. Another political variable—separatist autonomy—is highly significant and as I discuss below is so across both Eurasia and the post-colonial world. One plausible reason for this is that it represents not a central state policy but a regional one—an ethnic minority region declaring itself to be independent of the center. If we take “separatist activism” on a continuum at one end of which is actual violent rebellion, and somewhere along which is a declaration of independence or separatist autonomy, the two may be strongly correlated and in fact may both be functions of grievance and opportunity rather than cause and effect. A country’s level of development is also significant and negative in shaping the likelihood of separatist conflict, as is the size of the ethnic minority group as a share of the total population. However,

11

In fact, the post-colonial legacy effect is generally stronger than for the whole world. For comparison, see Cederman et al (2010), 105.

16

Smith


population size and the number of years since the last conflict are not. In general, these results for the post-colonial world alone both confirm the important role of an ethnic group’s history of contentious politics and the role of central state policy in catalyzing separatist conflicts. What, however, of the postcommunist world?

Eurasia Tables 3 through 5 present results for much the same model estimations but this time focused on the post-communist world. One initial trend is that state policy variation—and it is worth reiterating that in this sample it is just within the former communist bloc—is much less consistently significant here. In fact, in one or more of these models nearly all of the state policy variables correspond with a complete absence of separatist conflict onset, leading those variables and a large number of observations to be dropped from the sample. The one political variable, as with the post-colonial sample, that is consistently significant is the separatist autonomy measure. Again, since this is driven by ethnic region action rather than by central state policy it may be a part of a repertoire of separatist activism along with rebellion. What it does suggest, given the large number of group-year observations within the Soviet Union and then Russia, is that macro-level (central government) factors remain important. As I discuss below in analyzing Roeder’s data, those macro-variables continue to be significant even when accounting for more fine-grained policies toward individual ethnic groups. Does a history of nationalist mobilization—even a brief elite-led one between 1918 and 1920— shape an ethnic group’s propensity to mobilize nearly eighty years later? One common answer—that ethno-federalism catalyzed and magnified ethnic identities at the republic level, creating tools for mobilization that did not exist before—is a plausible one: “The nation-state projects that so were so powerful seventy years later played a minor role in shaping patterns of political action prior to the 17

Smith


establishment of segment-states. Without segment-states to coordinate and focus identity and provide resources for mobilization, few of the nation-state projects that reached the bargaining table after 1987 were able to muscle their way to the head of the queue in 1917” (Roeder 2007, 277). The logic here is that groups with segment-state status (or titular status under ethno-federalism) need not have had it before or, even if they did, it paled in causal importance to the institutional reality of Soviet nationalities policy. Where the results diverge somewhat from the post-colonial world is in the role of history and past rebellion. In Eurasia, past rebellions (post-1940) are strongly but negatively tied to future rebellion across all model specifications and with multiple state policy measures as controls.12 However, this is not the case when I relax the time constraint on the role of history and expand the periods of past rebellion that might influence current separatist activism. As discussed above, I include in the Eurasia models dummy variables to capture center-periphery wars during the pre-1848, 1848-1917, and 1918-39 periods. When I expand the “legacy” period back from 1946 to 1818 for Eurasia—an additional 128 years—the role of historical rebellion legacies emerges again. The resistance that many territories raised to Russian and then Soviet conquest during the 19th and early 20th centuries generally fails to predict the onset of violent conflict in the late 20th century. The single, and important exception is in Model 5, in which rebellion during the interwar period following the 1917 Russian Revolution but preceding World War II—is a significant predictor of later separatist conflict. This is likely a function of the brief independence moments of a number of later Soviet republics which few years are underplayed in many studies of post-communist separatism. However brief, those few years of statehood became central 12

There are numerous plausible explanations for this. One is coercion: arguably the Soviet state had greater coercive capacity in the post-World War II period than during any time before, and contained uprisings before they could get out of hand. Another is legitimacy: many scholars of Soviet history note the impact of Stalin’s defeat of Hitler as a huge catalyst for regime legitimacy, even in the non-Russian far reaches of the Soviet Union. See for example Beissinger 2002, XXX.

18

Smith


components of the nationalist narratives seven decades later and, statistically speaking, correlate strongly with the timing of the republics’ separatist activism. While post-World War II rebellion legacies do not appear to influence nationalist mobilization in the post-communist world, what we might call the imperial interregnum between Russian and Soviet rule seems to loom large in shaping late activism during the late Soviet era and beyond. It seems appropriate, given these findings, to think of the interwar period as analogous in the Soviet bloc to the immediate post-colonial periods in Africa, Asia and the Middle East. In the next section, I turn more specifically to differentiating rebellion legacies from the effects of institutions by focusing on ethnofederal and more broadly ethnic segment-states and their influence on separatist crises.

IV. History and Ethno-Federalism With and Without Communism In the previous section, it became clear that the effect of a legacy of past rebellions strongly influenced the onset of separatist conflicts in the post-colonial world, but that that effect did not match up in Eurasia. The question is why. The most common explanation per Bunce (1999), Hale (2008) and Roeder (2007) is that ethno-federalism led to the disintegration of Czechoslovakia, the Soviet Union, and Yugoslavia. Hale and Roeder both argue more broadly that ethno-federalism exerts such an effect outside of Eurasia.13 In this section I first replicate Roeder’s analysis to assess the baseline effects of broad categories of segment-states, then explore whether there are particularities inside Eurasia.

Data, Methods and Models

13

Although note that Hale (2008, 250) notes that the portability of this effect to the post-colonial world is somewhat shaky in the context of his dataset and analyses.

19

Smith


This set of analyses employs the ethnic group/nation-state crisis data originally used in Roeder (2007).14 It is a global sample that includes data in 5-year increments from 1955 to 1995. The unit of analysis is, as with EPR, the ethnic group-year. The original data comprise 8,074 group years and the first model reported is a straight baseline replication using the full sample. To maintain complementarity across the two datasets used here, I also estimate models 2-4 below using a subset of the data from the post-colonial world and Eurasia, for a maximum of 6993 group-year observations.15 I begin by estimating the same models as those presented in Roeder (Chapter 9). As I discuss below, in subsequent models I also included dummy variables for the communist bloc,16 for communist ethno-federations, and for ethno-federations in general.

Dependent Variables The main dependent variable here is the nation-state crisis (crisis). It is defined as “a turning point at which further escalation [of confrontation during center-periphery bargaining] may bring about the failure of the existing state and the creation of new nation-states” (Roeder 2007, 41). It is coded “1” in crisis years, and “0” otherwise.

Independent Variables As with the results reported above, my main interest here is in the effects of historical rebellion legacies in spurring subsequent rebellions. Amidst the institutional variables included in Roeder’s data is

14

The data are available at: http://weber.ucsd.edu/~proeder/EthnicforPosting.xls. I’d extend my thanks here for sharing the data publicly. 15 I estimated the models using Roeder’s entire sample and found no substantive differences. Those results are available from the author. 16 In no model estimations was the communist bloc dummy significant. It is excluded from the results presented here.

20

Smith


a measure for groups that, prior to incorporation into a larger nation-state, were states themselves but after incorporation had no autonomy (priornoa). This is a close analogue to the rebellion legacy variable in the EPR data but captures a different aspect of a group’s pre-incorporation national cohesiveness— formal independence. As with the separatist autonomy measure in EPR, it is a group action variable, not a central state one. And I would suggest that in cases like the Baltic states and Azerbaijan, which declared their independence in 1918 but were involuntarily incorporated into the Soviet Union in 1920, any effect is more likely political than institutional. The reasons are two: First, less than years is unlikely to have catalyzed substantial state- and institution-building. Second, the subsequent seven decades of Soviet rule are likely to have nearly obliterated any institutional effect, no matter how precocious a three-year effort might have been. This measure is included alongside the original institutional variables for sub-national status: first-order segment state/prior state (autprior), first-order segment-state/no prior state (autnop), second-order segment state (secondary), and no segment-state status (none). As I discuss below, the second-order dummy fails to be statistically significant in Roeder’s full sample but is significant in my truncated Eurasia/post-colonial sample. The significance in the smaller sample appears to be driven substantially by China, whose second-order segment state policies account for close to half of the total outside Eurasia.

Control Variables Included in these data are controls for an ethnic group’s size relative to the total population and for the country’s level of development (taken as the natural logarithm). The dataset also includes regime type variables to capture whether the country in question was democratic or autocratic at the start of the time period, which are derived from Polity IV data. Whether a group considers itself to be a cultural minority—a perceptual and somewhat subjective but politically important factor—is also included, as 21

Smith


well as whether or not it has a non-adjacent ethnic homeland. In short, the Nation-State Crisis data cover some of the same important ground as EPR, albeit with different measurement choices, but also some new ground. Substantively similar results across the two datasets, subsequently, ought to give us confidence in the causal effects.

Results Table 3 presents the results estimating the determinants of nation-state crises. Model 1 is a baseline replication of Roeder’s (2007, chapter 9) analysis. As such, I discuss it only briefly. Among the significant global predictors of nation-state crises are first-order segment-state status (both with and without a nation-state prior to incorporation), the absence of segment-state status (whose effect curiously runs in the same direction as both segment-state variables), and prior nation-state status. This last variable’s significance, as I mentioned above, is to my mind better thought of as an historical effect instead of an institutional one. The reason is that, absent a direct measure of the length of time an ethnic group had its own nation-state, we have no way of knowing whether it was long enough to have actually engaged in state-building.17 And, again, with the 1918-20 experiences of several subsequent Soviet republics as a cue, these periods might be more accurately thought of as independence by default—opportunistic declarations during imperial interregna—rather than hard-fought selfdetermination struggles. Nonetheless, this past history of independence matters strongly and its level of significance remains robust across all model specifications and sample. Again, the implication here is that an historical legacy of resistance via independence may have causal importance that is separate from any institutional effect.

17

I am currently coding a new variable for the length of time an ethnic group had a nation-state prior to incorporation.

22

Smith


Cultural minority status is also a consistently significant positive predictor of nation-state crises, as is group population share. Groups that have non-adjacent homelands are significantly less likely to escalate their bargaining with central governments to nation-state crises. As was the case analyzing the EPR data, a country’s level of development fails to achieve conventional levels of statistical significance when institutional factors are accounted for. Table 3 about here Models 2 through 4 present additional analyses of these data with a smaller sample: I have dropped observations from Western Europe, the United States and Canada and what remains is the post-colonial world (Latin America, Asia, Africa and the Middle East) along with the former Soviet bloc. This smaller sample contains 6993 group-year observations. Model 2 estimates the original Roeder model with the post-colonial/Soviet bloc sample. There is only one substantive difference, and it is that second-order segment-state status is significant at p < .01. Moreover, the effect is negative, rather than positive: all else equal, groups like the Chechens in Russia are less likely to escalate to nation-state crises when they have second-order autonomy. At first this was surprising: not just because the effect is suddenly significant in this smaller sample but because it is not immediately clear why second-order autonomy or titular status would generate significant negative force on the likelihood of nation-state crises. A detailed look at which states in this sample were generating most of the “1” scores for this dummy revealed twelve such ethnic minority groups in China, eight groups in South Africa, three in India, and one in Venezuela. One plausible theoretical explanation of the negative second-order effect here is that, during periods of great uncertainty (regime change, regime collapse, etc.), the demands of first-order segment state elites might “crowd out” those of groups with lower formal status. A simpler one is a macro-level central state explanation: China has been immensely successful at preventing demands by second-order 23

Smith


groups from escalating and its ethnic minorities have not been able to escalate demands with the same efficacy as elsewhere. I estimated the same models in columns 2-4 without the China observations (224 total, leaving 6769 observations) and without them second-order segment-state status is no longer significant.18 Column 3 presents the results with two additional variables added to estimate the effects of central state-level ethno-federalism, coded as per Bunce (1999) with the Soviet Union, Czechoslovakia and Yugoslavia coded as communist ethno-federal states. I also include a dummy variable for ethnofederations more broadly, with these three Soviet bloc states plus Ethiopia and India per Hale (2008). The results here are also interesting: even when state policy or institutions at the group/segment-state level are included, macro-level ethno-federalism is a consistently positive and significant determinant of nation-state crises. However, communist ethno-federation status is not significant in this model, which includes the second-order segment-state measure. Because there appeared to be significant collinearity between the second-order and communist ethno-federation variables, and because the significance of second-order autonomy was so dependent on China, I estimated the model (presented in column 4) without the second-order measure. Specified thusly, both ethno-federation variables are significant predictors, but run in different directions: ethnofederations generally experience more nation-state crises than other states, but communist ethnofederations suffer fewer.

Discussion

18

These results are available from the author.

24

Smith


One thing that seems clear, analyzing two exhaustive but different datasets with different measures for state policy and institutions, is that the state matters in shaping the outbreak of violent separatist conflicts. What is more, the impact of state policy and of central efforts to manage multiethnic societies by institutional means consistently crowds out what we previously thought was a robust effect of economic development. Until we have incorporated other structural determinants— mountainous terrain, the presence and location of natural resource reserves, to take just two—these findings are tentative rather than determinate. However, whether “state policy” is conceptually derived from deep comparative-historical knowledge of non-communist societies (Wimmer 2002) or communist ones (Bunce, 1999; Hale 2008; Roeder 2007), we have good reasons to move forward in elaborating these measures and thinking about them comparatively. This is one reason to think that the lessons from communist bloc cases have broader portability. The bad news, for those of us interested in maximal explanatory leverage, is that ethnofederalism, a concept derived from the Soviet and Yugoslav experiences, appears to have different effects outside Eurasia (as noted here but also in Hale 2008, 250). One inclination—which would run the risk of backward induction from post-Stalinist Soviet politics to the inception—would be to suggest that where India and Ethiopia fell on ethno-federalism out of central weakness that the Soviet and Yugoslav states fell on it through strength, as a conscious strategy to exercise direct rule. However, during the early decades of the post-revolution period, the Soviet state was far from able to exercise such effective Weberian authority across its expanses. During this period central policy was nearly as reactive to regional unrest as later capital elites were in Ethiopia and India. It is definitely true that Stalin’s later collectivization violently undid much of what his earlier nationalities policies had allowed to develop in terms of local authority, and perhaps this catastrophic downgrading of so many ethnic minorities’ status in the Union is a part of the explanation. 25

Smith


A second possibility is that it was the strong contradiction of a formal nationalities policy coupled with functionally second-class status—in which titular ethnic elites held power but mostly to enact policies decided on in Moscow, or what Suny calls “pseudofederalism”—that explains why Soviet ethno-federalism had different effects. It is plausible to my mind that Soviet ethno-federalism might have political effects much like of those of empires once dissolution seems possible. If we can extract the analytic utility of the concept of “empire,” much as Suny (2001) has done, from its normative connotations, the hierarchical structure of ethnicized cores and peripheries may help to build productive comparative insight. Now, on the impact of historical legacies. Ethno-federations “act” differently inside and outside the former communist bloc, but it is less apparent that history acts so differently in shaping long-term trends in nationalist and separatist mobilization against central governments. I developed in Section II of this essay a theoretical framework for understanding why it is that nationalist mobilization is sustained across long periods of time in some ethnic regions but not in others.19 Early rebellions by rural elites resisting direct rule—often conservative in outlook who sought to protect their own social power rather than advance more egalitarian minority rights—shape long-range subsequent rebellion by second- or even third-generation urban elites in a systematic way. Given the common weight of historical legacies across Eurasia and the post-colonial world, engaging cross-regional and cross-temporal research may help to synthesize the dynamics of separatist rebellion in a broader way.

V. Conclusions: How Far Should “Separatologists” Travel into (or out of) Eurasia?

19

This is developed more in Smith (2010), in turn part of a broader book project on post-colonial separatist conflict.

26

Smith


It seems appropriate to note in the conclusion here an exchange about fifteen years ago between Terry Karl and Philippe Schmitter, on one hand, and Valerie Bunce, on the other, in the Slavic Review. Karl and Schmitter made a powerful case for broad, cross-regional comparison as a starting point, arguing that we ought to set aside “area studies” limitations in the pursuit of maximally generalizable arguments. Bunce made a contrasting case for more modesty, suggesting that by falsely dichotomizing area studies and social science Karl and Schmitter were both obscuring some important differences between the communist bloc and other regions and dismissing the work of regional scholars who, while focused on one or more post-communist countries, were nonetheless comparative in theoretical terms. To my mind, the study of separatism today is something like transitology 15 years ago, but flipped on its head. Today it is not the case that scholars of former Soviet bloc are poised to write rejoinders to those arguing for the broadest comparative perspective on outcomes of interest. On the contrary, the study of separatism has become a field of globally ambitious inquiry dominated by the study of the former Soviet bloc. However, some of those regionally focused findings seem to travel uneasily to the rest of the world, especially its post-colonial regions. As a first step at disentangling the study of separatist conflict from the influence of communist bloc cases, this study has demonstrated two things. First, it is systematically the case that a legacy of past rebellion matters as much outside the communist bloc as within it. I have developed briefly a theory of historical legacies and tenacious separatism here, but have by no means “proven” it or dismissed alternatives. What I have done is to show that history matters differently inside and outside Eurasia. What is perhaps most interesting is that this is the case across the board in Eurasia, both in ethno-federal socialist states (Czechoslovakia, the Soviet Union and Yugoslavia) and the rest. Where Bunce’s (1999) elegant within-region variation suggests that ethno-federalism spelled the end for the former three states and the lack of it allowed the 27

Smith


rest of Eurasian states to hold together after socialism, a broader look at the world suggests a need to consider whether it was socialism combined with ethno-federalism. Or, somewhat less particularly, heavy-handed centralized dictatorship hand in hand with the functionally false promise of ethnic selfdetermination. This has important implications for the way we interpret the findings of studies such as Roeder (2007), for whom ethno-federalism is a starting point. Taken as an intervening variable that perhaps amplifies the already-strong identity effects of ethnic rebellion, somewhat contradictory findings like those presented in Hale (2008, 250)—that ethno-federalism significantly predicts separatism only in Eurasia—make more sense. But the prevalence of second-order ethno-federalism in the Soviet system may exert more weight on that particular institutional solution to multiethnicity than is due. To my mind, one important implication of the disproportionate role that ethnic group-year data from China exert on second-order autonomy status is that China and the former Soviet Union can fruitfully be compared as ethno-federal socialist states. One key difference, of course, is that only some of China’s first-order segment-states (e.g. Tibet, Xinjiang, and Mongolia) were ethnically driven. Another is that China’s nation-building effort has not been as focused on supra-national ideology as was Soviet nationbuilding (see for example Suny and Martin eds. 2001). Whatever accounts for the variation, I see this as a potentially valuable line of inquiry. What stands out in the end is that a longer-range historical lens is likely to be of much value in building theories to explain why it is that some ethnic groups engage in separatist mobilization and others either fail to or decide not to.

28

Smith


Table 1. Pre-Soviet identity and late-Soviet nationalist mobilization

Republic Armenia Azerbaijan Belarus Estonia Georgia Kazakhstan Kyrgyzstan Latvia Lithuania Moldova Russia Tajikistan Turkmenistan Ukraine Uzbekistan

Hale’s separatist rank (order of declaring sovereignty)* 13 4 10 1 5 15 14 3 2 8 6 12 11 9 7

Hale – number of weeks after 11/16/88 to declare* 92 44 88 0 52 101 101 36 27 83 81 92 92 86 83

*Source: Hale 2008, p. 99. §Source: Suny 1993, p. 30.

29

Suny’s rank of pre-Soviet national identity strength§ 4 1 1 2 3 0 0 3 1 0 0 0 2 0

Post-Soviet Internal Separatist Conflict? 0 1 0 0 1 0 0 0 0 1 1 0 0 0 0

Smith


Table 2: History and Separatist Conflict in Eurasia and the Post-Colonial World: Ethnic Power Relations Post-Colonial World Eurasia Variable Model 1 Model 2 Model 3 Model 4 Model 5 Exclusion 1.217 --§ --(.284)*** Downgraded 1.709 1.760 --§ --(.397)*** (.396)*** Powerless -1.224 ---§ -(.497) Downgrade*exclusion -----§ -Junior partner -.252 .617 --(.429) (1.112) Autonomy -.769 .281 --(.606) (.871) Discrimination -1.702 -.148 .543 .835 (.4778)*** (1.063) (.615) (.435) Separatist Autonomy -2.528 6.224 6.297 7.011 (.643)*** (1.101)*** (1.396)*** (2.0214)*** Group size (share of .257 .244 .275 .840 .880 population) (.089)** (.092)** (.227) (.294)** (.372)* Rebellion Legacy .823 .761 -3.997 -3.146 -3.146 (.189)*** (.187)*** (1.387)** (1.578)* (1.599)* Pre-1848 rebellion ------§ 1848-1918 rebellion ------§ 1918-39 rebellion ----2.659 (1.083)* GDP per capita(ln) -.294 -.289 -.081 .579 .512 (.130)* (.119)* (.233) (.672) (.816) Nat’l population(ln) .021 .024 -.393 .405 .307 (.106) (.103) (.356) (.317) (.415) Peace years -.126 -.112 -.048 -.174 -.197 (.083) (.085) (.238) (.234) (.224) Constant -4.072 -4.364 -1.260 -14.917 -14.298 (1.595)* (1.475)** (4.706) (6.993)* (8.710) Observations Prob. > chi2 Pseudo R2

16895 .000 .081

16895 .000 .092

4063 .000 .360

3378 .000 .455

4526 . .484

Analysis is by logistic regression. Robust (Huber-White) standard errors in parentheses, clustered on country. Splines included in analyses but not reported. *** p < .001 ** p < .01 *p < .05 § Variable predicted failure perfectly in Eurasia sample and was dropped.

30

Smith


Table 3: Ethno-federalism and Nation-State Crisis Inside and Outside Eurasia Roeder’s Nation-State Crisis Dataset Variable Model 1: Models 2-4: Full model & Eurasia & Post-colonial world data (replication) First-order segment & prior state First-order segment but not prior state Prior state, no firstorder segment state Second-order segment state No segment-state status Cultural minority Democracy at start of period Autocracy at start of period Turmoil at start of period Group population as share of total Remote homeland Part of ethnofederation Part of communist ethno-federation GDP per capita

Observations Wald χ2 Pseudo R2

Full model

Full model + ethnofederation variables

Full model w/o secondorder variable

2.411 (.555)*** 1.640 (.342)*** 1.521 (.532)** -.590 (.604) 1.265 (.553)** 1.119 (.237)*** .025 (.286) -.512 (.285) .370 (.307) 4.443 (1.335)** -3.122 (.769)*** --

2.419 (.568)*** 1.688 (.399)*** 1.611 (.551)** -1.095 (.422)** 1.439 (.536)** 1.099 (.260)*** .0563 (.296) -.521 (.282) .414 (.305) 4.624 (1.298)*** -2.653 (.787)** --

2.035 (.919)* 1.406 (.687)* 1.329 (.498)** --

--

--

1.171 (.791)

1.513 (1.005)

2.007 (.853)** 1.369 (.667)** 1.425 (.505)** -.917 (.407)** 1.058 (.532)* 1.134 (.271)*** -.306 (.333) -.489 (.263) .325 (.287) 4.806 (1.255)*** -2.622 (.787)** 1.232 (.368)** -1.083 (.601) 2.236 (1.275)

.773 (.527) 1.114 (.267)*** -.278 (.335) -.494 (.267) .322 (.288) 4.863 (1.242)*** -2.532 (.828)** 1.327 (.374)*** -1.334 (.583)* 2.153 (1.310)

8074 358.81 0.168

6993 475.69 0.159

6993 587.65 0.170

6993 456.42 0.166

Analysis is by logistic regression. Robust (Huber-White) standard errors in parentheses, clustered on country. Coefficients for panel variables estimated in models but not presented. *** p < .001 ** p < .01 * p < .05

31

Smith


References Aspinall, Edward. 2007. “The Construction of Grievance: Natural Resources and Identity in a Separatist Conflict,” Journal of Conflict Resolution 51, 6: 950-72. __________. 2009. Islam and Nation: Separatist Conflict in Aceh, Indonesia. Stanford: Stanford University Press. Aspinall, Edward and Mark T. Berger. 2001. “The Break‐up of Indonesia? Nationalisms After Decolonisation and the Limits of the Nation‐State in post‐cold war Southeast Asia.” Third World Quarterly, 22, 6: 1003‐1024. Ayres, William and Stephen Saideman. 2008. For Kin or Country: Xenophobia, Nationalism and War. New York: Columbia University Press. Beissinger, Mark R. 2002. Nationalist Mobilization and the Collapse of the Soviet State. New York: Cambridge University Press. Braithwaite, John, Valerie Braithwaite, Michael Cookson and Leah Dunn. 2010. Anomie and Violence: Non-truth and reconciliation in Indonesian peacebuilding. Canberra: Australia National University Press. Bunce, Valerie. 1999. Subversive Institutions: The Design and the Destruction of Socialism and the State. New York: Cambridge University Press. Dragadze, Tamara. 1996. “Azerbaijan and the Azerbaijanis,” in Graham Smith ed., The Nationalities Question in the Post Soviet States, 2nd edition, London: Longman: 269-90. Fuller, Elizabeth. 1987. “The Transcaucasian Republics in 1987,” Radio Free Europe–Radio Liberty, Radio Liberty Research (Munich), RL 8/88, 30 December 1987. Gleditsch, Nils Petter; Peter Wallensteen, Mikael Eriksson, Margareta Sollenberg & Håvard Strand, 2002. ‘Armed Conflict 1946–2001: A New Dataset’, Journal of Peace Research 39(5): 615–637. Gorenburg, Dmitry. 1999. “Regional Separatism in Russia: Ethnic Mobilisation or Power Grab?”, EuropeAsia Studies 51, 2: 245-74. Hale, Henry. 2008. The Foundations of Ethnic Politics: Separatism of States and Nations in Eurasia and the World. New York: Cambridge University Press. Laurila, Juhana. 1999. Power and Politics as Determinants of Transition: The Case of Azerbaijan. Helsinki: Bank of Finland Institute for Economies in Transition. Roeder, Philip. 2007. Where Nation-States Come From. Princeton: Princeton University Press. 32

Smith


Saroyan, Mark. 1997a. Beyond the Nation-State: Culture and Ethnic Politics in Soviet Transcaucasia. Berkeley: GAIA Research Series, Global, Area, and International Archive, UC Berkeley. __________. 1997b. The 'Karabakh Syndrome' and Azerbaijani Politics. Berkeley: GAIA Research Series, Global, Area, and International Archive, UC Berkeley. Sjamsuddin, Nazaruddin. 1985. The Republican Revolt: A Study of the Acehnese Rebellion. Singapore: Institute of Southeast Asian Studies. Smith, Benjamin. Forthcoming. “Conclusion,” in Adam White ed. (De)Constructing State Authority: Developments in the State-in-Society Approach. Seattle: University of Washington Press. __________. 2010. “History and Separatism: The Origins of Nationalist Rebellion in the Post-Colonial World,” paper prepared for presentation at the annual meetings of the American Political Science Association, Washington DC. September 2-5. Suny, Ronald Grigor. 1993. The Revenge of the Past: Nationalism, Revolution, and the Collapse of the Soviet Union. Stanford: Stanford University Press. __________. 1999. “Provisional Stabilities: The Politics of Identities in Post-Soviet Eurasia.” International Security, Vol. 24, No. 3 (Winter 1999/2000), pp. 139–178. __________. 2001. “The Empire Strikes Out: Imperial Russia, ‘National’ Identity, and Theories of Empire.” In Ronald Suny and Terry Martin eds. A State of Nations: Empire and Nation-Making in the Age of Lenin and Stalin. New York: Oxford University Press, 23-66. Treisman, Daniel. 1997. “Russia's "Ethnic Revival": The Separatist Activism of Regional Leaders in a Postcommunist Order.” World Politics, 49, 2, pp. 212-249.

33