Secessionism: Identity, Interest, and Strategy 9780773538962, 9780773539303, 0773538968, 9780773587502, 0773587500

Table of contents :
Cover......Page 0
Title......Page 4
Copyright......Page 5
Contents......Page 6
Acknowledgments......Page 8
Introduction: Understanding Secession......Page 12
1 The Positive Theory of Secession......Page 28
2 Explaining the Causes of Secessionism Worldwide......Page 61
3 Secessionism in Advanced Democracies......Page 83
4 Secessionism and Rebellion......Page 121
5 Secessionism and Autonomy......Page 148
Conclusion: The Future of Secessionist Politics......Page 162
Appendix: Supporting Data......Page 172
Notes......Page 192
Bibliography......Page 210
B......Page 224
D......Page 225
E......Page 226
H......Page 227
K......Page 228
N......Page 229
R......Page 230
S......Page 231
U......Page 232
Z......Page 233

Citation preview

SECESSIONISM

This page intentionally left blank

Secessionism Identity, Interest, and Strategy JASON SORENS

McGill-Queen’s University Press Montreal & Kingston • London • Ithaca

© McGill-Queen’s University Press 2012 ISBN 978-0-7735-3896-2 (cloth) ISBN 978-0-7735-3930-3 (paper) Legal deposit first quarter 2012 Bibliothèque nationale du Québec Printed in Canada on acid-free paper that is 100% ancient forest free (100% post-consumer recycled), processed chlorine free This book has been published with the help of a grant from the Julian Park Publication Fund at the University at Buffalo – The State University of New York. McGill-Queen’s University Press acknowledges the support of the Canada Council for the Arts for our publishing program. We also acknowledge the financial support of the Government of Canada through the Canada Book Fund for our publishing activities.

Library and Archives Canada Cataloguing in Publication Sorens, Jason, 1976–   Secessionism : identity, interest, and strategy / Jason Sorens. Includes bibliographical references and index. ISBN 978-0-7735-3896-2 (bound). – ISBN 978-0-7735-3930-3 (pbk.) 1. Secession.  2. Separatist movements.  3. Autonomy and independence movements.  4. Nationalism.  I. Title.  

JC311.S67 2011  320.54  C2011–903286–4 Typeset by Jay Tee Graphics Ltd. in 10.5/13 Sabon

Contents

Acknowledgments vii Introduction: Understanding Secession  3 1 The Positive Theory of Secession  19 2 Explaining the Causes of Secessionism Worldwide  52 3 Secessionism in Advanced Democracies  74 4 Secessionism and Rebellion  112 5 Secessionism and Autonomy  139 Conclusion: The Future of Secessionist Politics  153 Appendix: Supporting Data  163 Notes 183 Bibliography 201 Index 215

This page intentionally left blank

Acknowledgments

On reflection, it is fortunate that, when I first decided to write a book expanding my dissertation’s focus on secessionist parties in advanced democracies to secessionist movements and violent conflicts worldwide, I had little idea what I was getting into; otherwise, this book might not exist today. Through the years of writing and rewriting, I have enjoyed the unfailing support and encouragement of my wife, Mary, a true partner in this enterprise. Chapters 3 and 5 draw in part on my dissertation research, and I am indebted to the members of my committee for their useful criticisms and guidance through that process: Geoffrey Garrett, Frances Rosenbluth, Jose Cheibub, and the much missed Fiona McGillivray. Steve Saideman and Allen Buchanan very generously gave their time to reading entire, early drafts of this manuscript. I am also indebted to Hanna Birnír, Nicholas Sambanis, Jan Erk, Larry M. Anderson, Gregg B. Johnson, Joshua Dyck, Dinissa Duvanova, and many others for useful comments and conversations. The Baldy Center for Law and Social Policy, Julian Park Publication Fund, and United University Professions provided valuable financial assistance.

This page intentionally left blank

SECESSIONISM

This page intentionally left blank

INTRODUCTION

Understanding Secession I would remind you that we once had a civil war in our country, in which we lost on a per capita basis far more people than we lost in any of the wars of the twentieth century, over the proposition that Abraham Lincoln gave his life for, that no state had a right to withdrawal from our Union. Bill Clinton on Boris Yeltsin’s Chechen campaign, Moscow, 21 April 1996

Bill Clinton’s comments quoted above and the mostly critical media response that followed illustrate the contested nature of secession. Secession is the withdrawal of territory and the people living on that territory from the sovereignty of an existing state and the establishment of a newly independent state with sovereignty over that territory and its people.1 Secession is a rare phenomenon. One can count the number of countries affected by internationally recognized secession since the end of World War II on two hands: the Soviet Union, Czechoslovakia, Pakistan-Bangladesh, Yugoslavia, Indonesia, Sudan, and Ethiopia-Eritrea.2 At the same time, secessionist political parties and secessionist insurgencies are common, and they have had profound effects on the countries in which they occur. Armed self­determination movements are the primary cause of ethnic violence in the world today, and, since the 1980s, at least half of all ongoing civil wars in any given year have been secessionist.3 As the Soviet Union was breaking up, the idea of secession accomplished through the democratic process fired imaginations around the world. As a 4 September 1991 Guardian story noted, Nationalist fervour in the Soviet Union and Yugoslavia has stirred the separatist and hardline regional movements that underpin and fragment Italian politics. German-speaking separatists in the Alto Adige region, bordering Austria, have rekindled

4 Secessionism

a long-dormant cause by demanding to be cut loose from Italy. Union Valdôtaine, the usually tame nationalist party governing the Aosta Valley, bordering France and Switzerland, yesterday set in motion procedures for seceding. The calls follow renewed activism by militant separatists in Sardinia, and the increasing power of the Lombard League.4 Minority nationalists and secessionists from Indonesia to Ethiopia like­wise took heart in the fall of the Soviet Union. The extreme violence that shortly thereafter accompanied Yugoslavia’s breakup caused some scholars and policymakers alike to reconsider a liberal attitude toward secessionism. Others argued that it was precisely Yugoslavia’s attempt to hold itself together by force, along with Croatian and Bosnian refusal to allow Serb secession, that was responsible for the violence. Since the early 1990s, fears of a worldwide secessionist contagion have not materialized, but a steady drip-drip of secessions, recognized and unrecognized, has continued: East Timor in 1999, Montenegro in 2006, Kosovo in 2008, and South Sudan in 2011. Secession and autonomy movements continue to possess remarkable staying power throughout the world, and, as we shall see, change the course of politics even when they do not reach their ultimate goal of independence. Chechnya is an example of the dilemmas posed by ­secessionist politics. Boris Yeltsin’s intervention in Chechnya, which ­occasioned Clinton’s quotation, led to a series of further interventions as the region remained unstable. Under Vladimir Putin, the Russian ­military successfully targeted moderate Chechen nationalists such as Aslan Maskhadov for assassination. As the moderate nationalist leadership died out, a new strain of virulent Islamic extremism took hold in Chechnya, and new leaders emerged who had no qualms about targeting civilians in bloody incidents like the Beslan school hostage-taking.5 The continuing restiveness in Chechnya seems to be an indication of the failure of Russian policy. On the other hand, other Russian regions with aspirations to wide-ranging autonomy, if not independence, have not rebelled, and Russia’s harsh treatment of Chechnya has arguably deterred them from doing so. Should Russia allow Chechnya to secede, and, if so, by what procedure? The issue is complicated by the fact that a majority of Chechens may prefer to remain part of Russia. The Dudayev ­government



Introduction 5

that originally declared independence did not take power in free and fair elections. What caused Chechnya to develop a strong secessionist movement in the first place, and why did Chechens take up arms, while other unhappy regions turned away from the option of independence? Finally, why do some governments, such as those of Canada and Great Britain, rule out military suppression of democratic secession attempts, while other governments, such as those of Russia and Turkey, pledge to use overwhelming force against all secessionists? These are the sorts of questions this book seeks to answer. This is a book about secession, but more particularly secessionism. I define “secessionism” broadly to include movements that aim at substantial territorial autonomy for a minority group and do not rule out independence in the future. The evidence shows that most secessionist movements try to foster some ambiguity about their aims. The Tamil rebels of Sri Lanka – Liberation Tigers of Tamil Eelam (LTTE) – insisted on total independence in the 1980s and 1990s but in the early 2000s dropped that aim in favour of “federalism,” only to return to uncompromising independentism when peace negotiations broke down. The Kurdish parties of Iraq clearly would support independence if they thought they could achieve it (more than two million Kurds signed a petition demanding a referendum on independence6), but at the moment they are content with far-reaching autonomy, because a move to full independence could provoke a foreign-policy crisis with Turkey and the United States. The methods of this book are those of institutionalist comparative politics and international relations. For the most part, I do not consider the individual-level psychology of ethnic hostility itself, nor do I examine the historical emergence of nationalism over the longue durée. Rather, this book assumes that there are ethnically and nationally distinctive groups, and that there are leaders within those groups who have to decide whether to make demands for institutional change and whether to organize violence to achieve their ends. Ethnic differences and hostilities do not automatically create a demand for secession or a serious risk of violence. Contextual factors such as geography, economics, and political institutions will explain why some group leaders choose the “exit” option rather than “voice” or “loyalty” within the existing politico-territorial settlement.7 Despite its timeliness, the topic of secession has received short shrift in empirical political science. A vast literature on ethnic c­onflict

6 Secessionism

and its regulation exists, but little of it treats secession as a phenomenon distinct from “ethnic politics,” “ethnic violence,” or “minority nationalism” in general. Indeed, most existing empirical work on secessionism has treated the phenomenon as merely a particularly intense form of ethnic conflict. Both in the abstract and in practice, however, secessionism has distinct origins. The argument of the book comes in three stages, addressed under the headings “identity,” “interest,” and “strategy.” Within the historical context of the modern state, ethnicity and nationality – distinct concepts serving similar functions – emerge as super-tribal forms of collective, political identity, deeply rooted in the circumstances of human evolution. Ethnic and national identification often trumps other categories of political identity, such as class, because it constitutes a potential unifying and distinguishing characteristic for an autonomous political community. Within every ethnie lies the kernel of an independent state. Every territorially concentrated people with its own myth of common origins, usually based on language, independent history, or even political ideology, is potentially secessionist. Very few secessionists in history have lacked either strong cultural distinctiveness or territorial concentration. Most potentially secessionist peoples do not ever spawn movements for independence. Only when identity is coupled with interest does a popular desire for independence come about. If a people can be better off economically, culturally, and politically as the core of an autonomous political jurisdiction, then secessionism emerges; otherwise, even a politically aware ethnie will pursue other ­political goals such as recognition and representation within the existing political arrangement. Speaking of collectives as if they were unitary actors does, of course, distort reality. Within every ethnic group, there are individuals who identify more and less with the group and who support and oppose secession. But the more the group can expect to gain, as a group, from independence, the more individuals within the group will support it. At this stage of the argument, the “costs and benefits” of independence are an empty vessel into which we must ladle substantive theory. The real contribution of this book is to argue that the costs and benefits of independence for a community depend on the central government’s ability to commit credibly to adopting policies beneficial to that community. If the government cannot commit in this way, autonomy or independence becomes a desirable solution to at least some members of the minority community.



Introduction 7

However, even when popular desires for autonomy or independence emerge, they do not automatically translate into political action. Individuals must overcome the collective action problem to organize secessionist movements. Governments can use various techniques, from electoral systems to repression, to discourage such organization, or they can try to reduce latent secessionism by making concessions. Furthermore, even when a secessionist organization exists, it faces a choice between peaceful, democratic means of pursuing their goals and rebellion. Moving from latent to active secessionism and from active secessionism to insurgency lies within the domain of strategy. The main focus of this book is positive (explaining why things are the way they are) rather than normative (arguing why things ought to be a certain way). However, the empirical results are relevant to policy­makers’ attempts to reduce ethnic violence. The conclusion dis­cusses some of these implications. Much of the recent literature on ethnic conflict, both scholarly and popular, has emphasized the downsides of democracy or democratization in the context of ethnically diverse societies.8 The literature on civil wars has emphasized state strength and the lure of natural resources as explanations for violence but has generally downplayed any ameliorative effect of democracy.9 Additionally, some scholars have argued that government accommodation of separatists simply encourages more of them to rebel.10 This book’s argument for a more concessionary approach to secessionist movements seems to fly in the face of that evidence. I do not so much challenge those earlier findings as to argue that context matters. Secessionist movements are different from other kinds of ethnic and nationalist movements in their sources and demands, and secessionist civil wars are different from other civil wars in their solutions. This book finds that governments that have explicitly ruled out military suppression of democratic secession have suffered far less ethnic rebellion than governments that have declared their eternal indivisibility. I call this solution “quasi-legal secession,” because it does not necessarily entail a unilateral right of secession in the constitution, which is extremely rare. In addition, decentralization reduces rebellion in the short run but has no further impact in the long run. Another interesting fact is that newly independent countries have less ethnic violence than older countries. In most places, a majority of the local population opposes secession, because they

8 Secessionism

consider the risks of independence to be too great. With few exceptions, only groups that have suffered severe repression, such as Kurds in Iraq and Kosovo Albanians, display near-unanimous support for independence.11 On the basis of this evidence, I infer that a constitutional right of secession would substantially decrease ethnic violence around the world without significantly increasing the risks of actual state breakup in most countries – and, when countries do break up, further ethnic conflict is the exception rather than the rule. A constitutional right of secession would instead result in widespread devolution of power, allowing minorities to obtain rights of self-government in the areas most important to them. However, most governments are unlikely to pursue the solution of legalizing secession, because they are willing to accept substantial violence in order to decrease the risks of state breakup even slightly. Even so, there are reforms that even countries that prohibit secession could adopt to reduce secessionist violence, such as increasing checks and balances in the central government.

SECESSIONISM AS A VARIANT OF NATIONALISM As mentioned previously, the book uses the terms “secession” and “seces­sionism” to refer to all movements seeking extensive selfgovernment for their territories, whether or not they explicitly ­ endorse full independence. No country has complete independence in the sense of being free of all foreign political ties, influence, or obligations. Self-government is really a sliding scale, from full centralization (complete lack of self-government) to full independence (complete self-government). Every governing unit falls somewhere between those two extremes. Nevertheless, a key concept is that of sovereignty, the legal right of a governing unit to pursue actions in some area without being overridden by some higher power. Some regions possess internal sovereignty but not external sovereignty. For instance, the Isle of Man controls its own taxes, budget, and local laws, but does not have the ability to make treaties, declare war, or conduct foreign affairs – Great Britain does those things for the island. US states are considered sovereign in constitutional law because they enjoy autonomy in certain reserved areas where the federal government cannot overrule or usurp them. To be more



Introduction 9

­ recise about my definition of a secessionist movement, I will say p that a secessionist movement is an organization that: a) supports at least internal sovereignty covering a wide range of policy areas for a territory that does not yet possess it, and b) does not explicitly oppose the eventual attainment of external sovereignty as well. We can think of secessionism as a type of nationalism, where nationalism is an ideology or set of practices emphasizing and promoting the unity, autonomy, and identity of the nation.12 Nationalists promote the nation as a potential political community that is distinct from other political communities, that is unified in the sense that it contains no rival political communities within it, and that should have the autonomy to pursue its own destiny and preferences. In turn, the definition of “nation” might be “a human community with a shared commitment to unique political institutions for the whole community.”13 Nations should not be identified with states, because what makes the issue of secession interesting is the fact that many peoples who consider themselves nations do not have states at all (Basques, Scots, Baloch, and so on). Moreover, nations should not be restricted to “politically conscious ethnic groups,”14 for not all politically conscious ethnic groups desire unique political institutions (African-Americans, for example), and not all nations have an ethnic component or myth of common descent (the “American nation,” for example, is a civic not ethnic nation).15 Not all nations desire an independent state either; sometimes the desire for unique, common political institutions finds satisfaction in some kind of substate autonomy. Because most important political institutions require exclusiveness over a certain territory, only groups with a territorial claim can be nations, because only those groups can aspire to common political institutions. In this book, I use the terms “regional group,” “territorial minority,” and “minority nation” interchangeably. Figure 0.1 presents a typology of nationalist movements and shows where secessionism fits in. Nationalist movements can be based on either an advantaged group (often a majority) or a disadvantaged (often a minority) group. “Advantaged” and ­“disadvantaged” here refer to possibilities of political control of the state, not economic

10 Secessionism

well-being: an advantaged group controls the state, while a disad­ vantaged group does not control the state. Advantaged-group nationalisms (or “statist nationalisms”) are con­­servative in that they seek to maintain the security of the nation against threats external or internal. Examples of externally oriented statist nationalisms are imperialism (where the purpose of imperialism is the glory or enrichment of the nation), irredentism (that seeks to annex territories held by other states), xenophobia (where the concern is to exclude immigrants or foreign influences), and protectionism. The term “economic nationalism” typically refers to protectionist policies aimed at keeping foreign goods and investment out of the country.16 Statist nationalism directed against internal threats generally advocates policies of assimilation or expulsion toward immigrants or national minorities, or both, and of concentration of power in the hands of the central government. Typically, statist nationalist movements combine concerns about both external and internal threats; examples include Kemalism in Turkey, the National Front (FN) in France, Dixiecrats in the Jim Crow South, and the Bharatiya Janata Party (BJP) in India. Statist nationalisms can be either ethnic or civic: arguably, French nationalism in the moderate form of traditional Gaullism represents the interests not of the French “ethnic group” but of the French civic nation as conceived in postRevolutionary terms, while the extreme form of nationalism endorsed by the FN takes on a more ethnic character. Soviet ­nationalism was



Introduction 11

more civic than ethnic, because it was more about loyalty to the Soviet state than subservience to the Russian ethnic group. Disadvantaged-group nationalisms (typically called just “minority nationalisms”)17 focus on one of three demands: reunion, autonomy, or recognition with unity. Minority nationalists seeking unity with another country are minority irredentists, while the advocates of reunion in the retrieving state would be statist irredentists.18 Irre­ den­tism usually requires interstate warfare or threat of warfare to be satisfied. As a result, it is much less likely to be satisfied than secessionism, which can achieve its end peacefully. Recently, movements that would otherwise be irredentist often take on a secessionist ideology instead (e.g., Kosovo, Nagorno-Karabakh). This book rarely deals with classic cases of irredentism where secessionism is not an option (e.g., Northern Ireland), but it does address those cases where leaders can plausibly exchange irredentist claims for secessionist ones.19 Three types of movements make autonomy demands: autonomist, independentist, and anti-imperialist movements. Independentist and anti-imperialist movements demand full inde­ pen­dence, while autonomist movements may settle for wide-ranging control over internal affairs without the external sovereignty that independence implies. Anti-imperialist or decolonization movements enjoy international sanction under Articles 73 and 74 of the United Nations (UN) Charter and the 1960 Declaration on the Granting of Independence to Colonial Countries and Peoples, which together recognize the right of all non-self-governing territories under c­ olonial administration to independence. Because of this internationally recognized right, decolonization conflicts were not so much battles over “whether” but “when” and “how” the territories in question would become independent. Anti-colonial movements thus present a special case and are largely excluded from this book; however, this book does examine both independentist and autonomist movements and considers them both under the general term, “secessionism.” The book also does not directly examine “regionalists,” groups that seek recognition and limited self-government within a state but explicitly oppose wide-ranging sovereignty and seek assistance from the central government in their economic and political development.20 Examples of such groups include the Frisian National Party (FNP) in the Netherlands, the contemporary Dravidian parties of south India, and the Andalucist Party (PA) in Spain.

12 Secessionism

Even where secession movements follow national boundaries and adopt the rhetoric of national self-determination, the motivating factors for secession go beyond mere ethnic or national difference. Some territorially concentrated minority groups opt for the secessionist solution, while others do not, and the reasons have to do with the kind of self-government that would benefit the minority group. The findings presented in this book suggest that secessionism has different origins from other types of ethnic conflict.

GOVERNMENTS VERSUS SECESSIONIST MOVEMENTS Outbreaks of secessionism appear to occur in worldwide cycles, but these cycles reflect political circumstances rather than some natural rhythm. During the period of decolonization, the rhetoric of selfdetermination (a principle never intended to apply to secessionists within integrated states) inspired minority nationalist movements in Western Europe and North America, while the decolonization process itself often involved contested borders, as the abortive secessions of Katanga and Biafra showed. Less well known are the cases of several princely states in the Indian subcontinent that sought to declare their independence but were incorporated into India and Pakistan by force of arms. The only successful secession between 1944 and 1991 was that of Bangladesh from Pakistan.21 The high tide of secessionism in advanced democracies appeared to be the late 1970s and early 1980s, as the Scottish, Welsh, and Quebecois n ­ ationalists thereafter went into short-term decline (only to rise again in the 1990s). However, the fall of the Soviet Empire soon unleashed disintegrative forces in Eastern Europe. Not only did the USSR, Yugoslavia, and Czechoslovakia break up but countries that relied on the Soviet Union for military and financial support found themselves ­suddenly in a position of weakness. Ethiopia’s communist regime fell to a coalition of opponents, including the Eritrean secessionists, who won their independence. Moreover, the United States also lost interest in propping up repressive noncommunist regimes and stepped up a campaign to promote human rights, democracy, and a vaguely defined principle of self-determination internationally. As a result of this policy and of economic crisis, the Suharto regime in Indonesia fell, and East Timor gained its independence shortly thereafter. Some observers have argued that the wave of secessions in Eastern Europe and elsewhere bolstered the older secessionists of Western



Introduction 13

Europe and North America. Quebeckers nearly voted to secede, the Scots overwhelmingly and the Welsh narrowly approved devolution, and Belgium, Spain, and Italy embarked on tortuous paths of decentralization. As chapter 3 will argue, the demonstration effects of secessions elsewhere were actually less important for secessionists in advanced democracies than were newly strengthened international institutions such as the European Union and NAFTA that seemed to promise greater economic security for small sovereign states. In the early twenty-first century, the tide of secessionism seems to have subsided in some places, but revived in others. Support for sovereignty among Quebeckers has declined, although the sponsor­ ship scandal temporarily revived the issue, and the Scottish and Welsh nationalists have had difficulty gaining traction in their devolved assemblies. Russia has curbed secessionist agitation from all its regions except Chechnya. A peace agreement has been signed in Aceh, allowing the rebels of the Free Aceh Movement (GAM) to lay down arms and contest elections so long as they forswear independence in favour of expanded autonomy for the region. A peace agreement for the Southern Sudan includes a provision for a referendum on independence in 2011. A lengthy ceasefire between the Sri Lankan government and the Tamil Tigers (LTTE) broke down in 2006, and after fierce fighting Sri Lanka finally seems to have subdued the LTTE, but many Tamils remain alienated from Sri Lanka’s centralized and arguably discriminatory political system. Montenegro declared independence in 2006 from the Union of Serbia and Montenegro after a successful referendum and was quickly recognized, while Kosovo issued a unilateral declaration of independence from Serbia in 2008 after years of fruitless negotiations. Thus, while secessionism appears to be abating slightly in certain places, it remains at the top of the agenda elsewhere. Ted Gurr argues that ethnopolitical rebellion, whether secessionist or nonsecessionist, has declined since the 1990s because of a number of negotiated autonomy and power-sharing settlements and because of the consolidation of several postcommunist democracies.22 Nevertheless, this book will present evidence that autonomy arrangements work best when a legal path to secession is possible. Since many governments are unlikely to adopt this solution, secessionist rebellion is likely to be with us for a very long time. Most governments have not pursued a consistent, principled, or effective approach with regard to secessionism, either internationally

14 Secessionism

or domestically. The failure of the European Union’s Badinter Commission to forestall warfare in the former Yugoslavia is well known. The United Nations has pursued an inconsistent approach toward secession movements. Two principles of international law appear to be in conflict: territorial integrity of states and self-determination of peoples. Both principles appear in the United Nations Charter, and the principle of self-determination was clearly meant to apply only to colonial cases.23 Nevertheless, it is not clear what, in principle, distinguishes a secession movement from an anticolonial movement. International lawyers have adopted the saltwater criterion, whereby territories separated from the ruling country by sea are entitled to obtain independence if they so desire, while territories adjacent to the ruling country enjoy no such rights.24 This criterion seems arbitrary, lacking a clear justification in conflict prevention.25 Given the ambiguity inherent in the self-determination principle, it is not surprising that secessionists throughout the world have sought to have it applied to themselves. In general, governments have treated independence claims extremely conservatively, refusing to recognize new states until the states from which they have obtained independence consent to the fact. The Badinter Arbitration Committee’s ruling on Yugoslavia deviated from this general rule, although whether the European Union was too hasty or too begrudging with recognition is hotly debated to this day. The United Nations did not intervene in East Timor until invited by the new Indonesian government, and in the case of the Western Sahara the UN has been content to rely solely, and fruitlessly, on diplomatic methods. Governments, of course, are reluctant to countenance any principle that could result in their dismemberment. Almost without exception, governments deny that an unequivocal right to secede exists and at most concede only that they will choose not to use force to crush secession attempts that satisfy democratic criteria.26 Arguably, this position has frequently thwarted the development of clear, generally acceptable, constitutionally consistent standards of assessing and handling secession attempts. Outside the liberal democracies, demands for independence are a significant cause of civil war, riots, terrorism, and other kinds of ethnic bloodshed. The cycle of repression and mobilization frequently obliterates future prospects for ethnic reconciliation and democratization. Governments around the world urgently require new methods of handling their secessionist conflicts. This book is virtually unique in



Introduction 15

bringing together empirical evidence on the causes and consequences of secessionist politics and an evaluation of techniques to resolve and avert secessionist conflict. The evidence shows that minorities’ demands for self-government stem from the distinctive cultural, ideo­logical, and economic policies they wish to pursue and that compromise on autonomy short of independence is almost always possible. However, those compromises are less likely to happen when minorities do not have access to the ultimate sanction: secession. Thus, prohibiting secession increases the risks of violence by making compromise difficult or impossible. Apparently, legalizing secession allows secessionists to commit to electoral methods and reduces governments’ insecurities about offering a secessionist minority the political and economic resources inherent in autonomous institutions. Since we cannot expect governments to take the initiative in adopting policies that will undermine their own power, many of the recommendations made in the conclusion of this book are intended to be consistent with governments’ interests as they perceive them.

PLAN OF THE BOOK Chapter 1 presents the positive theory of secession and secessionism. What causes people to organize collectively to pursue self-government for their territory, and what are the consequences of this choice for state response? Popular demand for self-government is a result of the economic, political, and cultural benefits of self-government. Recent findings from social psychology and experimental economics on the group basis of reciprocal altruism in humans, combined with anthropological and historical research on the origins of the state, explain why ethnic identity and territoriality are such critical factors in demands for collective self-government. Self-government as an institutional solution is relevant to territorial minority groups because of the inability of centralized governments to commit credibly to future policies beneficial to them. Political mobilization for selfgovernment should be a function of popular demand and collective action costs. The outbreak of rebellion and granting of autonomy should both be a function of political mobilization and institutions that either alleviate or exacerbate the inability of both secessionist minorities and central governments to make credible commitments to each other. Quasi-legal secession actually helps governments and

16 Secessionism

­ inorities compromise on autonomy and avoid conflict. Territorm ial autonomy reduces the underlying demand for independence in the short run but can give a secessionist movement access to the levers of power and patronage and increase a population’s willingness to consider independence in the long run. These tendencies make autonomy a difficult compromise under any circumstances, explaining why nego­­tiations over autonomy often seem so intractable. When compro­mise is unattainable, violence is possible, especially when the government is too weak to deter rebellion. Therefore, ethnic groups that make secessionist claims are more likely to rebel than non-secessionist groups. Chapter 2 then tests the determinants of secessionism across a world­wide sample of ethnopolitical groups. The main findings are that territorial concentration in a homeland and lack of dominance over the state are, not surprisingly, prerequisites for secessionism, while, more interestingly, lost autonomy, geographical separation, sea access, secessionist kin, the number of other secessionist groups in the state, mineral resources, and governmental discrimination pro­mote secessionism, and relative size of the group, demographic minority status in the group’s own homeland, and legal prohibition of secessionism reduce it. While lost autonomy, secessionist kin, and presence of other secessionists have received attention in the existing literature, the others are new findings. Chapter 3 focuses on advanced democracies and examines the deter­minants of vote share for secessionist political parties. Looking at secessionist vote shares allows us to get a much finer-grained measure of popular support for secession than we can get with the crude measures used in chapter 2. Accordingly, some findings are generally stronger in chapter 3. Relative affluence is found to encourage secessionism, while irredentist potential reduces it. There are reasons why we would expect these results to obtain in democracies more than non-democracies, which the next chapter explains. The results from chapters 2 and 3 show that political and economic calculations play a critical role in the development of secessionism; mere cultural difference is not enough. The next two chapters deal with the consequences of secessionism. Chapter 4 looks at the effects of secessionism on severity of ethnic rebellion and civil war, conditional on secessionist policy. Secession­ ist and irredentist groups are much more likely to rebel against the state than other groups. Unsurprisingly, rebellion and civil war are



Introduction 17

less common in high-income countries, whether democratic or autocratic, but secessionist rebellion and civil war are particularly rare when secession is quasi-legal. Thus, opening up avenues for future independence dramatically reduces the potential for violence. Short of legalizing secession, a country can also reduce secessionist rebellion by decentralizing and ending discrimination. Qualitative evidence shows that autocracies frequently renege on autonomy agreements and that peace agreements with secessionists in autocratic countries rarely last long. Interestingly, newly independent states have less rebellion than older countries. Thus, permitting secession could also reduce ethnic violence indirectly by increasing the number of newly independent countries, although border changes may affect other forms of ethnic conflict, such as intercommunal riots, differently. Chapter 5 examines the causes of regional autonomy. Almost all advanced democracies are decentralizing, but secessionist electoral success accelerates the process. Countries with fewer executive constraints, mostly autocracies, offer less autonomy than more constrained governments, mostly democracies, and countries with more executive constraints or checks and balances also are less likely to repeal autonomy. The conclusion reviews the evidence on global secessionism and its implications. Culturally distinct, territorially concentrated peoples seek autonomy whenever it is in their interests. Institutional reform (decentralization or secession) is the goal, rather than policy concessions, because policy concessions can be more easily reversed. When secession is forbidden, groups are less confident of their long-run preservation in a united country, and when they do become secessionist, they are likely to engage in violent rebellion because both they and their host states suffer from credible commitment problems: secessionists may promise to abandon their aims of independence but cannot credibly commit not to use institutions of autonomy to cultivate a secessionist constituency, and governments may promise not to revoke these autonomous institutions later on, but their promise is not always credible. In autocracies, the commitment problem is most severe because the central government can repeal or vitiate autonomy at any time. Secessionist rebellions will be ­particularly intractable in autocracies, except totalitarian regimes that are able to use massive repression against ethnic dissenters. Despite all these problems, less developed countries are unlikely to legalize secession, because the risks of actual disintegration are much higher in these

18 Secessionism

countries than in the advanced democracies. Attempting to suppress secession may cause a violent backlash in one or two regions, but this strategy prevents secessionists from organizing in other minority regions that are not as deeply disaffected, thus forestalling a more widespread disintegration of the country. Figure 0.2 graphically represents the development of seces­sionism from ethnic difference. First, territorially concentrated ­ethnic groups desire policies favourable to them (secession for groups that would benefit from sovereignty, limited self-government for groups that would not). Then, the costs of collective action determine whether the popular demand for institutional change translates into political mobilization (secessionism). Then, the severity of commitment problems for secessionists and central governments – determined  by whether secession is permitted – and the ability of the c­ entral gov­­ ern­ ment to deter rebellion through military strength affect whether secessionists choose to rebel or instead work through the political process.

1 The Positive Theory of Secession

This chapter builds a positive theory of secessionism: why it emerges and why it is so closely associated with organized rebellion against the state. One point this discussion stresses is that secessionism must draw on more than cultural differences like language – sovereignty must be economically and politically appealing for the citizens of a region to consider this radical step. In this respect national secession has something in common with municipal secession in the United States, where comparatively wealthy areas sometimes try to withdraw from larger cities in order to retain more control over their tax base.1 Unlike municipal secession, national secession almost always has an ethnic character: both identity and interest are essential for national secessionism. In that sense, then, secessionism is a rational strategy, selected by ethnic leaders when it is the best available means to the given end of collective interest. Intragroup competition among would-be leaders helps drive this process; ethnic minority leaders can­not ignore the interests of their constituents, even when they are not democratically elected, because lacking control of a state with taxation powers, they have to elicit resources and cooperation mostly voluntarily. Once a nationalist organization emerges, the central government must figure out how to handle it: repression or negotiation. Modifying Max Weber’s famous definition slightly, we might define the state as an organization with a monopoly of the legiti­ma­tion of the use of force within a certain territory. Governance is ineluctably territorial for most human societies. Secessionists therefore must secure control over a certain territory in order to establish a new governance unit for themselves. The units of potential secession are territorially concentrated populations. Culturally or e­ thnically

20 Secessionism

distinctive groups that lack territorial concentration never develop secessionist movements, because they lack one of the fundamental prerequisites of statehood. Furthermore, groups that are territorially concentrated but do not reside in their traditional homelands are less likely to develop secessionist movements, because they lack a strong claim of entitlement to the territory on which they reside.2 The argument proceeds in these steps: social identity determines the bounds of political decision-making, affecting how costs and bene­ fits are evaluated sociotropically; group interests determine whether some members of a group come to favour independence or autonomy for the group; strategic interaction among group members determines whether a nationalist movement organizes, and strategic interaction between nationalist organizations and the central government determines whether violence or compromise, such as decentralization, will occur.

IDENTITY: THE ORIGINS OF CULTURALLY DISTINCTIVE GROUPS A culturally distinctive group, for the purposes of this book, is a group sharing a common ethnic or civic-national identity. Ethnic identities are based on myths of common genetic origins or ancestry, while civic nationality refers to a myth of shared political origins and fundamental ideology.3 The Germans were long thought of as an ethnic nation, while Americans are largely today a civic nation.4 This distinction is more conceptual than actual. In practice, civic nations have generally reflected the interests and power of a dominant ethnic group. For instance, post-Revolutionary French civic nationalism demanded assimilation of Occitans, Bretons, Basques, and others into Parisian-French language and their loyalty to a centralized state; black Americans were originally excluded from citizenship (and even the human race) in many states; and so on. In recent years, liberal democracies have increasingly addressed immigration and growing ethnic heterogeneity by stressing civic rather than ethnic bonds of loyalty and fraternity, even in states with clearly ethnic origins such as Germany. Stateless cultures, too, can lean toward either ethnic (e.g., the Kurds) or civic (e.g., the Scots) bases of identity. Why is it that most independent states promote a common cultural identity for their people and that most secessionist movements seek a state for a culturally defined collectivity? After all, human beings



The Positive Theory of Secession

21

categorize each other into groups defined along many other dimensions: gender, age or birth cohort, class, party affiliation, family, even personality. What should be immediately clear is that none of these other group identities can form the basis of a whole political society. (While orthodox Marxists believe that a single economic class can build the foundations of a new, classless society, they also reject the division of the human race into separate nations or ethnic groups.) Only ethnicity and civic nationality can serve as common identities for distinct, enduring polities.5 This answer only yields new questions. After all, cultural identity can have many different bases: language, race, ideology, common political history, even religion. People generally have multiple possible cultural identities. What explains which of these identities constitutes the basic marker of their collective, political loyalties? In answering this question, I will bypass the stale debate among the traditional paradigms of identity theory: primordialism, constructiv­ ism, and instrumentalism.6 Political leaders are sometimes able to use nationalism to pursue power, as instrumentalists and constructivists claim, but their ability to manipulate their supporters is highly constrained by the cultural material they have to work with and its salience to the masses, as primordialists maintain. Instead, I will focus on what we know about the psychology and rationality of cultural group identity. Horowitz argues that ethnicity represents the extension of kinship networks under the modern state. Literal kinship is too narrow a basis to compete for political power in the modern state, and ethnic entrepreneurs have therefore created myths to bring together tribes of similar characteristics (such as language) into the extended strangergroups we call “ethnic groups.”7 To explain the ability of ethnic identities to provoke hostility, he draws on the work of psychologists on “minimal groups.”8 Experiments have shown that when individuals are classified into groups, even on the basis of arbitrary criteria such as tastes for abstract art, their behaviour changes. They cooperate more with members of their own group and less with members of other groups than they do when no group identities are assigned. Horowitz’s arguments show that human beings are willing to act in a self-denying manner for the sake of the group, but they do not explain why group affiliations along ethnic lines would be especially prone to political conflict.9 Why do political entrepreneurs not organize people more successfully along class, gender, or generational

22 Secessionism

divides? In fact, the minimal groups experiments suggest that even arbitrary characteristics can be used to generate group competition. Hechter focuses on territorially defined, culturally distinctive communities, that is, “nations.” His ingenious argument is that nations form from voluntary communities (“solidary groups”) tasked with the specific function of providing collective cultural goods.10 The func­tion of inter-group cultural boundaries is to mark off those who contribute to collective endeavours such as “security from predation or confiscation, cooperative production of food, insurance from natural disasters, and greater access to information and mates” from those who do not. Outsiders can then be excluded from enjoying the benefits of these collective goods, while insiders can be monitored for their contribution and sanctioned for free-riding. Under feudalism and imperialism, nations were largely self-governing in the direct-rule system, but with the arrival of the modern, direct-rule state, minority nations were subjected to rule by an alien cultural group. Thus, nationalism arises from disjuncture between the locus of governance and the locus of collective cultural reproduction.11 Hechter’s theory therefore holds that cultural distinctions in primitive human societies were intentionally maintained to promote efficient production of public goods by marking off contributors from non-contributors. An alternative view is that the cultures that developed in disparate human populations were a product of human action but not human design, largely unintended by­products of the territorial clustering of the human population (which ­efficient cooperation presumably did require).12 Linguistic evolution, for instance, remains today largely a spontaneous and gradual process. Apart from a few striking examples, divergent features of languages, whether “primitive” or “modern,” cannot be explained as consequences of intentional processes of cultural differentiation.13 The question remains, then, how apparently utilitarian differences in human languages could come to take on such political significance. Hechter’s theory could be modified to hold that, while cultural differences arise spontaneously, maintaining them might be the principal aim of nationalism. Nevertheless, this modification still could not explain how nationalism emerges, often in its strongest form, in nations that are in no danger of cultural assimilation (e.g., pre1945 Germany, contemporary Kurdistan, Biafra). Indeed, for a phenomenon arising from cultural differences in human social groups, modern nationalism places surprisingly little stress on “boundary



The Positive Theory of Secession

23

maintenance” and much more on perceived political and economic exploitation or subordination.14 Indeed, since modern states are not voluntary societies and elicit cooperation as much through threat of punishment as through patriotic fellow-feeling, cultural boundary maintenance is probably less functional than it might have been in primitive societies. Melding some of Horowitz’s and Hechter’s arguments with new insights of his own, Hale returns to the social-psychological research and criticizes the inferences that Horowitz draws from the minimalgroups research.15 Hale argues that human sorting into groups has an evolutionary function and that differences in group identities are not inherently conflict-prone. Recent psychological research shows that people categorize others into groups in order to make sense of social situations and make rapid decisions about prospects for cooperation or competition. For instance, in a confusing situation in which a person is confronted with other people who are having an argument, one set of experiments showed that in default of any other option, people tend to use ethnicity as a marker to group the different sides of the argument, even when the actual argument does not break down along ethnic lines.16 However, when another prompt is provided, such as shirt colour, which correlates better with the lines of dispute, respondents opt for the more accurate prompt. Hale argues that ethnicity is an especially useful uncertainty-reducing device compared to other identifiers, allowing individuals quickly and roughly to categorize potential cooperators and competitors. Hale’s theory that ethnicity is a useful uncertainty-reducing device does not explain why, in the example just mentioned, respondents will opt for any marker or prompt that best serves their needs.17 The argument that ethnicity is a more useful way of categorizing people than age or gender is unpersuasive, since these latter modes of categorization are essential to human reproduction. Since ethnicity does not correlate very well with cooperative or competitive personalities, intelligence, or any other aspect of evolutionary fitness, we should actually expect that in everyday life people will tend to see more uncertainty-reducing value in almost all other ways of socially categorizing others. My argument, developed later, is that it is political conflict that especially activates people’s willingness to identify themselves and others on an ethnic basis. Van Den Berghe argues that evolutionary sociobiology explains why ethnic identities are so conflict-prone. Following in the vein

24 Secessionism

of the “selfish gene” arguments of Dawkins, Wright, and others, he argues that human beings have evolved to seek to advance the survival chances of not only ourselves but also those with similar genes.18 This evolved characteristic explains why humans are so solicitous of the interests of their offspring and near kin. Van den Berghe argues that ethnic groups are also viewed as kin groups and that humans therefore have a particular attachment to their own ethnic group. Whether they realize it on a conscious level or not, human beings will naturally seek to advance the interests of co-ethnics, as in so doing they will enhance the survival prospects of their genetic material. One of the most important problems with this theory is its incorrect implication that family and very narrow ethnic affiliations will predominate over broader affiliations. The broader the definition of an ethnic category, the less genetic material co-ethnics will share. For instance, the primary bases of ethnic politics in the United States are quite broad: most African-Americans would identify more closely with the broader African-American identity than with a specific African ethnic origin discovered through genetic testing or genealogical research. Thus, theories emphasizing genetic similarity as the foundation of ethnocentrism are at best highly incomplete. In addition, van den Berghe’s theory does not explain how civic forms of national identification could come to predominate over ethnic affiliations. Recent research in social psychology and game theory has gone beyond the selfish gene hypothesis to show that group-oriented behaviour can be beneficial to the survival chances of the individual organism, not just its genetic material. “Reciprocal altruism” is the term for beneficent behaviour that continues conditional on future reciprocation from the beneficiaries. The “tit for tat” strategy in the iterated Prisoner’s Dilemma game is an example of reciprocal altruism.19 The strategy requires the player to cooperate with the other player at the beginning and then in subsequent interactions play whatever move the other player did last time (either cooperation or defection). Scientists have observed altruistic behaviour in many species, and in laboratory experiments on human subjects there is considerable evidence of fairness norms and seemingly non-rational cooperation, as in the Ultimatum Game.20 To explain these findings, Frank goes beyond standard game theory to argue that there is an evolutionary advantage to psychologically viewing strategic situations in a selfless, altruistic frame.21 Thus, one is more persuasive about one’s beneficent intentions if one really has a b ­ enevolent



The Positive Theory of Secession

25

­ isposition. Within standard utility analysis, benevolence can be d under­stood as incorporating the utility of others into one’s own utility function. People with benevolent dispositions will both give and receive more cooperation, enhancing their survival chances. Could ethnic or civic-national identities set boundaries to reciprocal-altruistic behaviour? If so, why? To answer this question, we can refer to historical and social-scientific research on the origins of the state. Early twentieth-century sociologist Franz Oppenheimer propounded an influential theory that remains conventional wisdom in its broad outlines.22 In Oppenheimer’s view, most prehistoric human societies lacked even a proto-state. Voluntary and familial in origin, early hunter-gatherer and farming communities were organized to provide common defense against raids, as well as assistance in necessarily collective economic tasks, such as harvests and the hunt. Levi, Olson, and other political scientists have followed Oppenheimer in seeing the origin of the state in organized banditry, which largely focused on settled agricultural communities that produced an economic surplus.23 In regions such as the Nile, Mesopotamia, and the Levant, where the agricultural surplus was sufficient to support an administrative class, bandits settled and routinized their collection of plunder through taxation. This form of “stationary banditry” enhanced incentives for economic production relative to the uncertainty of random, mobile banditry. Thus, the early states were characterized by the rule of a culturally distinct elite over a hodgepodge of subject communities of various origins. Here Hechter’s account of the indirect-rule state, which lasted up through feudalism and the early modern period, follows the standard interpretation. As Laitin argues, indirect-rule states had little reason to assimilate their subject populations to the elite culture, but over time modernizing monarchies did try to rationalize administration by ensuring that regional elites (the aristocracy) adopted the language of the monarchy.24 The post-Revolutionary French state is often held up as a striking early example of the triumph of the direct-rule state.25 By inculcating strong national loyalties into its citizens, France was able to maintain larger and more disciplined armies than the old empires of Europe. However, Anderson argues that nationalism first developed in Latin America, where Creole administrators adopted American identities in reaction to the discriminatory apparatus of the absolutist Spanish state.26 Indeed, Spanish centralization after 1715 arguably qualifies Spain as the first multinational, direct-rule state.

26 Secessionism

This extremely brief historical sketch is merely intended to show that human political communities were for most of human development divided along cultural lines. This cultural distinctiveness was not a product of human design but was an inevitable consequence of the territorial segmentation of a pre-literate human population. Following the logic of evolutionary sociobiology, we can surmise that cultural boundaries provided an important guide to the prospects of political cooperation and competition in early human societies. Language is an extremely important marker of human political identities because in primitive societies the prospects for useful cooperation between different communities would have been immeasurably higher when they could understand each other.27 Likewise, for most of human existence religion has comprehensively ordered our moral universe. Co-religionists could be trusted better to understand a community’s rules and customs of social interaction and the proper sanctions for breaking them. In the indirect-rule state, divisions between rulers and ruled, exploiters and exploited, tracked cultural differences, and it is not surprising that cultural identities became the most fundamental basis for political conflict in such systems.28 With the emergence of the direct-rule state and an increase in the territorial scale of the state, it is also not surprising that small-scale, tribal loyalties have amalgamated on the basis of mutually intelligible languages, faiths, and customs into what we now know as modern ethnicity. Civic identities, too, can serve as a strong cultural glue in modern states, especially in immigrant societies where for most citizens social ties to more primordial forms of community have been radically disrupted. In territories with a history of slavery, ethnic boundaries inevitably track racial or phenotypic differences above all. In post-Enlightenment societies the culturally binding function of religion has faded but not evaporated. Because religion is now widely viewed as a matter of individual choice, it is more changeable and less ascriptive than other potential markers of cultural identity such as language and race. Perhaps more importantly, religion is no longer so universally viewed as a marker of trustworthiness and subscription to a common moral universe. It is generally impractical to restrict one’s cooperative dealings to co-religionists in such societies. To sum up, ethnic and civic identities can be politically divisive because human beings evolved in culturally distinct political communities, and therefore still tend to view a shared culture as a prerequisite to political freedom (non-domination). Political freedom is



The Positive Theory of Secession

27

important because its lack implies that an alien group can make fundamental decisions about the fate of one’s own group. The human impetus for collective political freedom, which ancient and early modern philosophers generally viewed as being consistent with or even necessary for personal freedoms,29 explains how such fierce conflicts can erupt over symbolic ethnic ownership of the state.30 Contra Hechter, the goal of nationalism is not to maintain cultural boundaries, but to secure the interests of a social group that happens to be bounded culturally. Contra Hale, ethnicity is not necessarily a useful uncertainty-reducing device in modern societies, but most humans’ preference for living in culturally homogeneous polities is in the first instance an evolutionary, psychological relic. In most day-to-day social life, ethnicity should matter much less, and categories such as age and gender much more. Language, race, civic ideology, common political history, and, in societies with less secularized moral systems, religion can all yield collective ­cultural-political identities, depending on the institutional, demographic, and historical context. When a culturally distinct group demographically dominates a particular territory but not the state in which they live, the option of political self-government inevitably arises. The territorial requirement generally means that language and independent history in particular will be associated with secessionism, while racial and religious groups are frequently not territorially concentrated and will seek either equal rights, “personal autonomy,” or control of an existing state. Not all territorially concentrated identity groups develop strong movements for self-government: whether support for independence or robust autonomy becomes widespread in the group depends on group interests.

INTEREST: THE ORIGINS OF DEMANDS FOR COLLECTIVE SELF-GOVERNMENT Self-government is a type of political institution. As an institution, it constrains political actors. Under the classical, Westphalian conception of sovereignty, states are not supposed to interfere in each other’s internal affairs unless they have voluntarily undertaken treaty obligations to permit interference. War merely for the sake of territorial conquest is forbidden. Despite trends toward continental integration in Europe and some degree of global integration through the United Nations, World Trade Organization, and other ­multilateral

28 Secessionism

institutions, sovereignty still is supposed to confer certain rights on all members of the state system, including the right to withdraw from treaties and regain full independence. Under the anarchy of the state system, the threat of force is the ultimate guarantor of the various rights of sovereignty. A state must be willing and able to defend itself, or enlist other states in its defense, to enjoy de facto as well as de jure independence. Some institutions of self-government fall short of full sovereignty. As described in the introduction, self-governing territories may enjoy internal but not external sovereignty. Territories enjoying full internal sovereignty have alienated their rights to conduct foreign policy and military defense to some full-fledged member of the state system. Examples include the Isle of Man and Monaco. In federal systems, the territorial entities enjoy rights to exclusive or concurrent legislation in certain areas; this is a kind of abridged internal sov­ ereignty. These institutions of self-government constrain not only other states but also the central government under whose protection and administration the territories govern themselves. Many federal constitutions give the territories direct representation in the central government to try to prevent the central government from violating their sovereign rights. If the central government has a free hand to alter or abolish the institutions of self-government at any time, then self-government is not fully institutionalized, because it does not necessarily constrain the policies of the central government.31 Why would countries institutionalize self-government at all? The answer is that self-government, like all political institutions, can tie the hands of future governments, preventing them from pursuing undesired policies.32 Suppose that a territorial community within a country wishes to pursue a set of distinctive policies, say, having to do with protecting a local language. Now suppose that the central government is controlled by legislators who represent ­speakers of another language, and they wish to impose this language on the minority region. The citizens of the minority region could try to make threats to prevent the policies from being changed: they could threaten to declare unilateral independence and rebel against the state, for example. Now suppose that the central government, in order to end the protests and threats, declares that it will make the minority language official in the region where it is prevalent. Will the minority citizens be satisfied with the result? It is unlikely that they will: the central government could simply change its mind



The Positive Theory of Secession

29

later, when things have died down. It is costly to organize protests or rebellions, and there is no guarantee that they will always deter the central government (because it would be costly to carry them out, the central government may believe that threatening to do so is merely “cheap talk”). The citizens would be much more satisfied if the central government actually forswore any future role in deciding language policy and devolved that policy to regional governments. This decision would ensure that policies protecting minority languages would be safe for the future, not just the present. Self-government is desirable for territorial minorities because it changes the context of majority rule from the whole country to the local region.33 When certain policies are decided by the local government rather than the central government, then the local population knows that, as long as majority rule is in place at the local level, those policy decisions will tend to redound to the benefit of the local majority, i.e., the national minority. The popular demand for territorial self-government thus arises in the credible commitment problem for central governments with respect to their territorial minorities. (Perhaps a parallel logic would explain demand for nonterritorial kinds of personal autonomy, such as separate religious court systems, coming from nonterritorial minorities.) There is no commitment problem for the central government with respect to whatever group dominates policy creation at the central level. In democracies, the median voter, typically the member of a majority ethnic or cultural group, should tend to dominate policymaking because of the institution of majority rule. Thus, in democracies governments face no commitment problem with respect to the country’s majority ethnocultural group, if one exists. This group controls central policy and therefore needs no ­territorial autonomy from the central government. Only minority territorial communities will seek territorial autonomy in democracies. In dictatorships, a minority group might control the state, such as “the Sunnis” in Saddam Hussein’s Iraq (I refer loosely to the whole group as a single entity, although clearly some Sunnis benefited from the Hussein regime far more than others) and the Alawis in Syria. In these cases, most members of the minority group in control will favour centralization, and the other groups will tend to support self-government. According to the logic above, all minority nations should support some kind of self-government. Of course, there are many different kinds of self-government. One limited kind of self-government

30 Secessionism

involves local control of cultural policies and nothing else. At the other extreme is independence. Not all minority nations will support independence. As noted in the introduction, this book deals with all movements seeking at least extensive internal sovereignty and not ruling out external sovereignty (independence). To explain why some minorities would develop full-blooded secessionism, rather than more limited kinds of self-government movements, we need to examine the benefits full internal sovereignty would accord to a minority nation.34 If decision-making is sociotropic (decision-makers make decisions on the basis of what they believe is best for the group, rather than what is best for themselves individually – a plausible assumption for low-cost activities such as political voting), then we might initially expect that the majority of a national group will support secession whenever the expected benefits of independence (or what is the same thing, the expected costs of remaining within the state) exceed the expected costs of independence (or expected benefits of unity) for the whole group.35 This strong result does not hold once we consider that decision-makers may be less than fully informed about expected costs and benefits, and that they may also be risk-averse, in which case the risks of severe negative outcomes will weigh more heavily in their minds. We can still make a weaker claim: all else equal, the higher the expected net benefits of sovereignty for a group, the more members of that group will support sovereignty. “The expected net benefits of independence” include cultural, economic, and political factors. As argued above, once a majority of people in a region have adopted a distinctive culture separate from that of the majority of people in the state governing that region, selfgovernment becomes a possible tool for preserving that culture and enhancing the well-being of people who have adopted that culture.36 Cultural difference alone is usually not enough for secessionism. If a culturally distinctive region can derive economic or political benefits from full internal sovereignty, then secessionism becomes a strong possibility. Sometimes, regions seemingly without stark cultural differences from the rest of the country develop secessionism due solely to economic and political factors; examples include Northern Italy, Alberta, and of course the Confederate States of the nineteenth century. In these cases, political differences were sufficiently deep that some members of the disaffected region felt that they constituted a



The Positive Theory of Secession

31

distinct national group and that their nation’s interests could not be advanced within the existing state.37 Secessionism therefore requires not only cultural difference but also economic and political benefits of sovereignty to become a serious political option. The benefits of sovereignty arise from systematic divergence between the interests of the central government and those of the minority community, a divergence that makes it impossible for the government to guarantee policies favourable to the minority. It will be important to operationalize expectations of this divergence with concrete variables that predict the future ethnic distribution of power. While economic considerations can make secession desirable, the core of the argument is that expectations of being locked out from political power at the centre underlie secessionism. It is important to note that even when a group benefits from sovereignty over its own affairs, it will almost always be willing to compromise on autonomy less than full independence. Only a strong desire to control foreign policy could compel a group to seek full independence and nothing less. Very rarely does a secessionist movement argue that control of foreign policy is essential for the nation’s development. (One exception might be many Montenegrin nationalists, who argued for separation from Serbia on the grounds that an independent Montenegro could obtain EU membership much more quickly than could the union with Serbia.38) It is the inability of secessionists and governments to make believable commitments to respecting an autonomy compromise that causes violence in so many cases. Chapter 2 conducts an empirical analysis of the causes of secessionism among territorially concentrated, politically organized ­eth­nic minorities who are not dominant (that is, who do not control their state). All these groups have the basic prerequisites – terri­ torial concentration, minority status, and cultural differences – for a demand for self-government, but more than half in fact do not have secessionist organizations. We can use economic and political factors to explain why some groups do support secessionist organizations and some do not. Unfortunately, this analysis cannot tell us how important different cultural factors are in promoting demand for self-government among the people of a region, since every group in the analysis is already culturally distinct. Chapter 3 addresses this problem by using legally defined regions as the units of analysis. In that chapter I use cultural, economic, and

32 Secessionism

political factors to explain secessionist vote share across regions and over time. Economic Factors HYPOTHESIS 1 (H1) Regions that benefit economically from internal sovereignty are more secessionist than regions that do not. Relative affluence (in democracies only), mineral resources, population, geographic separation, sea access, and economic globalization should promote popular demands for secession. Territorial grievances are frequently based on perceptions that the region is a net loser from the existing political union. In particular, when a region pays more in taxes than it receives in expenditures it is likely to present demands for fiscal autonomy and possibly even secession if those demands are not met. Regions that receive more in expenditures than they pay in taxes are poor ground for secessionism, because independence would mean the loss of subsidies. It is difficult to test directly the hypothesis that regions that pay more in taxes to the central government than they receive in services from the central government are more secessionist, since figures on fiscal balance are either unavailable or controversial. Scholars that have estimated interregional fiscal flows at particular points in time have found a strong correlation between regional affluence and fiscal outflows in democracies. For example, according to Clark,39 in 1970 Vizcaya (a province of the relatively affluent Basque country) received from the Spanish government only 24 percent of what it paid in taxes, and Payne estimated that in 1967 and 1970 residents of the Basque region paid 14.7 percent and 12.8 percent of total taxes received by the Spanish government yet received only 5.4 percent of the Spanish government’s domestic expenditures in both years.40 Treisman has studied interregional redistribution in Czechoslovakia, Yugoslavia, the Soviet Union, and Russia. He finds that the Croats and Slovenes in Yugoslavia, Czechs in Czechoslovakia, and the Baltic republics, Russia, and Ukraine in the Soviet Union were net losers from interregional transfers, while Serbia, Bosnia, Macedonia, and Kosovo in Yugoslavia, Slovaks in Czechoslovakia, and the Central Asian republics of the USSR were net winners in the fiscal game.41 In Russia, central authorities apparently targeted subsidies to the most nationalistic regions (Tatarstan, Dagestan, Siberia – but



The Positive Theory of Secession

33

not Chechnya, which had already taken an intransigent position on secession), and these subsidies then reduced secessionist agitation in these regions.42 Central governments in the other three states did not attempt this fiscal appeasement, and secessionist agitation rose among most of the net losers until they actually did secede. Emizet and Hesli likewise find that demand for sovereignty among Soviet republics was weakest where per capita income was low and strongest where it was high.43 Fiscal considerations probably partly explain why the Slovenes, Croats, and Baltic republics pushed for secession, while the Kirghiz, Uzbeks, Kazakhs, and Turkmen – despite their cultural differences from Russians – hung back. In Czechoslovakia, fiscal considerations can explain Klaus’s bargaining position and therefore Meciar’s willingness to accept separation, despite the fact that most Slovaks and Czechs opposed separation in opinion polls. Fiscal considerations of course cannot explain why Slovenes, Croats, and Slovaks had such well-defined group identities and broadly nationalist movements in the first place. They can only explain why nationalism turned to secessionism. Even then, fiscal considerations are not the whole story. In Yugoslavia the Kosovo Albanians and Bosnian Muslims eventually set themselves firmly in favour of independence despite their fiscal benefits from the existing system. In the Bosnian case, the secession of Slovenia and Croatia probably meant that there were no fiscal benefits to remaining within Yugoslavia, as Bosnia and Serbia had similar per capita incomes.44 The Croats and Slovenes had long been hurt by central government fiscal policies, but only attempted to secede in the late 1980s and early 1990s. In democracies, comparatively wealthy regions generally pay more in taxes, because under majority rule, there is almost always some positive level of redistribution from the wealthy to the poor, possibly because the decisive median voter is generally poorer than the mean income earner.45 In autocracies, however, the decisive decisionmaker may be a member of a very small ruling coalition at the centre of the government. If that ruling coalition is a wealthy group, they may redistribute from the poor, especially if the poor are less able to resist.46 At the same time, the Spanish example demonstrates that an autocracy could be even harsher on rich regions than a ­democracy could ever be. In democracies, therefore, economically better off regions are more secessionist than economically worse off regions; in autocracies, this relationship could be weaker or nonexistent.

34 Secessionism

Horowitz has observed that poor groups in poor regions are early, frequent secessionists, while advanced groups in advanced regions are late, rare secessionists.47 Ayres and Saideman have not found any relationship between economic differentials and secessionism.48 Horowitz’s hypothesis appears to hold up better in postcolonial auto­cracies, which constitute the bulk of the cases from which he draws his conclusions. In these countries, relatively well-off ethnic groups may have been better positioned than worse-off groups to compete for and obtain political power at the centre after the departure of the colonial power. Once obtaining political power, the autocratic rulers could redistribute to themselves. There are other ways to interpret the wealth effect. Gourevitch suggests that economically growing but politically weak regions develop secessionism because of a decline in the legitimacy of the centre.49 Another possibility is that more affluent regions are seen as more viable after independence. However, neither interpretation is as plausible as the fiscal cost-benefit hypothesis. Catalonia and the Basque regions have been the industrial centres of Spain for more than a century; there has been no sudden crisis of legitimacy. In Italy, Lombardy and Veneto steamed ahead of the rest of the country in the 1960s and 1970s, and a legitimacy explanation of secessionism is more plausible there.50 Nevertheless, all four cases are ones of strong secessionism; thus, recent declines in legitimacy of the c­ entre do not seem to be the main explanation. As for viability, relative affluence should matter little to viability. Total affluence and economic size matter more for viability. Even a region that is fairly low-income by Western European standards – Wallonia, for example – would be much more affluent after independence than almost all already independent African and South Asian countries. Affluence is certainly not a viability concern for any regions in advanced democracies, analyzed in chapter 3. The reason relative poverty suppresses secessionism in Wallonia is that independence would mean a loss of significant subsidies for the region. Neither the viability nor legitimacy arguments could explain why the relative affluence-secessionism link might be found in democracies but not in autocracies. Concerns about other aspects of viability do have a high ­profile in debates over independence, with opponents of secession t­ ypically stressing that the region concerned is not viable. Larger territories are thought to be more viable as independent states than small ones:



The Positive Theory of Secession

35

in the language of economics, provision of most public goods enjoys declining average costs over population. Separation can restrict access to external markets and require greater s­elf-sufficiency, although globalization appears to be changing this outcome, as this book will demonstrate. Historically, smaller countries have had lower growth rates than larger ones when they did not open to international trade.51 Presumably, smaller countries are also less able to defend themselves in war. Thus, smaller groups may derive fewer benefits from independence. The success of many small countries seems to render suspect these arguments and the economic logic underlying them; however, there is no denying that arguments to viability in terms of size have a prominent place in anti-secessionist rhetoric. Thus, variables perceived to affect viability (whether they actually do or not) should make a difference to some voters and would-be participants in secessionist movements. Distance is a factor that can affect both net benefits of independence and military capability for a region (and thereby the observed secessionist mobilization of the regional group in countries that ban secession). If a region is cut off from the rest of the country by open sea or by the territory of another country, then the region should be better able to resist invasion than a region attached to the rest of the country. The case of Bangladesh is instructive; the Indian army was able to intervene decisively in favour of the secessionists because of the fact that then-East Pakistan was cut off from West Pakistan by a vast stretch of Indian territory. The military rationale makes less sense in advanced democracies. Distance may also affect ­culture; human communities may develop distinctive features in isolation, in a manner roughly analogous to the evolution of different ­species in far-flung territories. Another important effect of distance is that it usually renders the region less dependent on trade with the rest of the country. If independence means a disruption of trade ­patterns with the former country, then this disruption will be less costly for distant regions than adjacent ones. The introduction also mentioned the salt-water standard in international law for the right to self-determination, but this aspect does not matter here, because this book ignores all non-self-governing territories as defined by the United Nations. For economic reasons, then, geographically separate regions should be more secessionist. Even without geographical separation, sea access could be important to a region hoping to secede, as it would permit more opportun-

36 Secessionism

ities to trade and make the new country less politically dependent on its neighbours. Regions with sea access should be more secessionist than regions without. Another economic factor to consider is globalization. What effects might economic globalization, understood as the worldwide increase in foreign trade and capital mobility, have on secessionist movements? I do not address this issue in depth in this book, as it has been amply explored elsewhere. Economists and political scientists have developed theory suggesting that as the world becomes more economically integrated, smaller political units will become more efficient, and those political units will support a globalized, free-trading system.52 Globalization does not cause regions without existing secessionist parties to develop them, but it does appear to cause already-existing secessionist parties to increase their vote share over time.53 In summary, relatively affluent (in democracies only), populous regions enjoying geographic separation from the rest of the country or at least sea access should show greater demand for extensive self-government. Political Factors HYPOTHESIS 2 (H2) For reasons of international politics, regions with irredentist potential are less secessionist than regions without such potential. HYPOTHESIS 3 (H3) Groups that have little power in the current state and expect to obtain little power in the future are more likely to consider secession than groups with access to power. Therefore, ideological differences from the rest of the country (in democracies only) and small relative size should promote secessionism. HYPOTHESIS 4 (H4) In countries where at least limited self-government is available, the current level of self-government might not affect secessionist ­support, but increases in self-government should reduce secessionist support, and failed decentralization attempts may increase secessionist ­support. Where governments have indicated, through d ­ iscriminatory policies, that even limited self-government is unavailable, secessionist support will be higher. Offers of autonomy to other ethnic groups in a country should cause a group to make its own demands.



The Positive Theory of Secession

37

Political factors also affect whether full – as opposed to limited – selfgovernment is a relevant solution to a regional group’s grievances. By “political factors” I mean factors having to do with ideological differ­ences and the availability of strategies for obtaining power. Key political factors affecting demand for full ­ self-government include irredentist potential, relative size of the group in autocracies, partisan-­ideological differences in democracies, territorial autonomy, discriminatory and restrictive policies, and decentralization to other groups. If a regional group has kin that dominate an adjoining state, then the group may seek attachment to that state (irredentism) rather than independence. Thus, German speakers in South Tyrol ­(officially, the Province of Bolzano) are more likely to seek reattachment to Austria than an independent South Tyrol. Independence could be used as a stepping-stone to reunion, but the states involved are aware of this and usually take precautions. For instance, K ­ osovo obtained supervised independence (against the wishes of Serbia) on condition that Kosovo could never join Albania. Indeed, the potential irredentism in Kosovo was part of Serbia’s excuse for implacably opposing secession. Armenians in Nagorno-Karabakh eventually dropped their goal of reunion with Armenia in favour of independence. In Kashmir, the most popular and oldest s­ eparatist group, the Jammu and Kashmir Liberation Front (JKLF), has always favoured independence rather than reunion with Pakistan; the ­Pakistani-controlled irredentist groups have little grassroots support within Kashmir. Border revisions should generally have more direct and destabilizing consequences for the regional balance of power than secessions, because they involve handing over territory to a potential adversary. States should therefore be less likely to concede defeat to a rebellious secessionist group with irredentist potential than to a rebellious secessionist group without irredentist potential, and the capabilities for secessionist rebellion might therefore be lower for groups with irredentist potential. On the other hand, irredentist potential implies the existence of a potentially sympathetic foreign country that could support a rebellion. As Horowitz notes, however, states rarely seek to retrieve minorities abroad, partly because retrieval could alter the domestic balance of political power.54 Thus, groups with irredentist potential are unlikely to become either irredentist or secessionist: they will usually (but not always) accept their fate in the existing state, unless the kin-dominated government decides to support the “stranded” minority.

38 Secessionism

In practice, cases of potential irredentist conflict are sometimes alle­viated by negotiations between the countries involved, which wish to avoid persistent international tension over the issue. Thus, these potentially irredentist enclaves typically receive high levels of autonomy and “special status” within their state. South Tyrol (ItalyAustria), Aaland (Finland-Sweden), and Northern Ireland are cases where inter-state negotiation or arbitration was used to settle an irredentist dispute. This pattern especially holds for democracies, which are reluctant to go to war with each other. The only persistent irredentist conflict in advanced democracies is that in Northern Ireland, where the irredentist group is actually the minority in the region itself, and therefore autonomy would not answer their demands (in fact, the situation of Catholics was worse under the Stormont parliament than under direct rule). The next hypothesis relates directly to central governments’ commitment problems with respect to minority groups. Groups that have less prospect for future influence in the existing state are more likely to seek independence. In ethnically based regimes, where political cleavages follow ethnic boundaries, one’s ethnic membership determines one’s political prospects, as elections take on a censuslike character.55 This state of affairs is common in countries that are less than fully democratic. Large ethnic groups will have more influence in coalition politics at the centre and will also enjoy a greater prospect of controlling the state by coup d’état if necessary. Generally, these groups will not fear union in a state with many smaller ethnic groups. Rather than developing autonomist nationalism, these groups will fight for power over the state.56 Thus, Shi’as in Iraq are unlikely to be secessionist, even when out of power, because they constitute the majority of the population there and enjoy a reasonable prospect of affecting or controlling policy in the future. In Nigeria, Hausa-Fulani leaders have favoured a strong central government since 1967, because although the group is not a majority in Nigeria, it is the largest group, and it would be very difficult for the many other groups in the country to form a united coalition against them in the central government. The Yoruba are the second largest group in the country and have also spurned secessionism, unlike the Ibo, the third largest group. In ethnically diverse non-democracies, ethnic party systems will predominate, and therefore in these countries the ethnic groups with a chance of attaining power – generally the most populous groups in the country – will be less secessionist.



The Positive Theory of Secession

39

In well-established democracies, ideological differences between regions can play the role that relative ethnic size does elsewhere, indi­cating systematic divergence between the central government’s policies and what the minority group wants. If a region’s voters are well to the left or right of the country as a whole, they can expect to have little influence on important public policies at the central level. Independence would offer voters the opportunity to shape important policies more according to their own preferences. Thus, four consecutive election victories for the Conservative Party in Britain (1979–1992) provoked a nationalist reaction in Scotland, where Labour won a majority of MPs in each election. Many Scots considered the Conservative government to have no mandate to govern Scotland. Observers generally agree that this experience solidified Labour support for devolution to Scotland.57 Even when Labour governs in Westminster, it must be attentive to the preferences of voters in Britain as a whole rather than just those of Scottish voters, if it hopes to be re-elected. In addition to absolute size of ideological difference, the direction of regional ideological difference may matter. Traditionally seces­sionist parties have been left wing, but this tendency may be ­reversing now that much of the left sees decentralization of the welfare state as a threat because of the possibility of inducing taxcutting competition for capital. Chapter 3 examines the effects of both absolute ideological difference and left-right ideological orientation using election data. The effects of existing levels of territorial self-government on demands for further self-government are a matter of intense and inconclusive debate. There seem to be inconsistent effects on latent demand for self-government and collective action for ­self-government, with autonomy reducing the demand for secession but increasing the ability of the region to pursue secession.58 Horowitz argues that smaller regions impede collective action against the state, but, while this structure may reduce communal conflict at the central-state level, it should not have any effect on demands for ­self-government,59 and chapter 3 finds no relationship between relative size of region and secessionist electoral support. One way to parse the effects of autonomy is to compare longstanding autonomy arrangements to recently proposed or implemented institutions. In democracies, if a central government offers a regional legislature to a region that did not previously have one, or concedes

40 Secessionism

further powers to an already existing regional legislature, then many voters in the region may feel satisfied that the new arrangements are sufficient. These voters, best thought of as “conditional secessionists,” will in the near term no longer vote for the secessionist party. By contrast, if a promised offer of autonomy falls through, dashed hopes may lead to greater resentment than would have existed if the offer had never been considered, and the secessionist alternative may benefit. Longstanding autonomous institutions protect a sense of regional separateness and provide the regional government with the resources it needs to cultivate a secessionist constituency if it so desires, effects that could counteract the mollification of conditional secessionist voters.60 In advanced democracies, recent increases in autonomy should reduce secessionist vote share, failed decentralization attempts should increase secessionist vote share, and longstanding autonomy arrangements could well have no consistent effect on secessionism (H4). I test these hypotheses in chapter 3, partly through qualitative anal­ysis. Outside the advanced democracies, I cannot test changes in support for secessionist movements. However, autonomy should play a role in ending armed secessionist conflicts, a role addressed in the section below on the consequences of secessionism. An important point to make is that the actual lack of self­government for a minority nation does not mean that limited selfgovernment is totally unavailable. The government of France has shown itself willing to offer legislative autonomy to regions, such as Corsica, that show sustained, significant demand for it. In countries where the government truly is intransigent on the issue of limited self-government, then we might expect that otherwise “regionalist” and anti-secessionist regions could become secessionist. Although no institution can guarantee the right policies, the lack of institutions of self-government creates a situation of insecurity for minorities. There is evidence that institutions such as democracy and federalism reduce discrimination against ethnic minorities.61 When minorities’ fears are realized and governments do enact harsh discriminatory measures, these policies presumably reflect the fact that the central government is unwilling to contemplate even limited self-government for the regional minority. When limited s­elf- government is unavailable, groups that might otherwise oppose independence might come to support it as a better option than the status quo (H4). By contrast, if the central government has



The Positive Theory of Secession

41

a­ ccommodated other autonomist movements, then there is a positive sign that self- ­government is available. More groups should make demands for self-government, but fewer should adopt intransigent independentist positions.

STRATEGY: COLLECTIVE ACTION FOR SELF-GOVERNMENT HYPOTHESIS 5A (H5a) The expected net benefits of forming secessionist movements are lower, all else equal, in countries that prohibit such movements than in countries that allow them, and therefore countries that prohibit secessionist movements will have fewer secessionist movements than countries that allow secessionist movements. HYPOTHESIS 5B (H5b) Secessionist organization will rise significantly during periods of state collapse. HYPOTHESIS 5C (H5c) Actual secession is more likely after a repressive state’s power collapses than after a non-repressive state’s power collapses, or than when a state’s power, of any type, is constant. That some people support independence does not necessarily mean that these people will undertake activities to achieve it. The first obsta­cle is a collective action problem. Independence is something that may benefit everyone in a group, regardless of whether everyone has contributed to the struggle for independence. Rational individuals will prefer to free-ride on the contributions of others to the public good of independence.62 This obstacle will be less important the lower the costs of the collective action in question and the higher the private benefits to participating in collective action. Voting has very low costs, and we see that millions of people in fact turn out for elections. In democracies that do not discriminate against secessionist parties and their supporters, the costs of voting for a secessionist party for any voter are likely to be no higher than the costs of voting for a non-secessionist party. Therefore we should expect that in those democracies that do not discriminate against secessionist parties and their supporters, the higher the expected net benefits

42 Secessionism

of independence for a national group, the greater the proportion of the group that votes for a political party supporting independence. For this reason, I am able to test most of the hypotheses of this chapter on advanced democracies using electoral data. I could not use electoral data to measure accurately popular support for secession in countries that prohibit secessionist parties or systematically discriminate against certain ethnic groups or organizations in the political process. Scholars have analyzed solutions to dissenters’ collective action problems at a “micro” level, considering in detail the strategies and tactics of dissenter groups and the state.63 The approach in this book is instead to examine the basic structural conditions, such as secessionist policy, that enhance or detract from secessionists’ ability to organize collectively. Most multinational countries in the world attempt to reduce the individual benefits of participating in a secessionist movement by imposing individual punishments. These countries legally ban secessionist political parties and pressure groups and sometimes even prohibit any kind of ethnic or cultural organizations for groups that the government suspects of disloyalty. These countries can be either democracies (Turkey, India) or dictatorships (Burma, Sudan). In such countries, the penalty for secessionist activism can be imprisonment or even death.64 The threshold for starting a secessionist movement will be higher in such situations. Thus, members of a national group are less likely to form secessionist movements where these movements are prohibited than where they are allowed (H5a). Note that banning secessionism does not reduce a group’s support for secession, just their propensity to form organizations to achieve secession. In fact, we should expect that when secession or secessionism is prohibited, the net expected benefits of independence are higher, and therefore support for independence among group members is higher. Would you join a group if you knew you could never leave? Once in the group, are you likely to feel secure about remaining in the group if you are told you may never leave? Groups should feel less secure about the future in countries where secession is prohibited, and where those groups do not have much influence on government policy. (Obviously, if a group controls the central government, the net expected benefits of independence are negligible, and so is support for independence among group ­members.) Where secessionism is prohibited, therefore, secessionist



The Positive Theory of Secession

43

­ ovements will be fewer but latent support for secession will be m higher than where secessionism is permitted. In fact, it is conceivable that prohibiting secession increases support for secession so much that the number of secessionist movements actually increases. As it turns out, banning secessionism actually does not appear to reduce it very much (chapter 2). The organization of secessionist movements will rise to catch up with the high level of latent secessionism when there is a sudden swing toward tolerance, especially where that tolerance is likely to disappear at some point in the future. Secessionist groups will try to take advantage of the lull in state repression to organize quickly and secure independence. With respect to state collapse, this argument is also supported by applications of the familiar Prisoner’s Dilemma game to the “ethnic security dilemma.”65 When exit (secession) is prohibited or impossible, groups will see to their own security for fear that other groups will try to exploit them first and, because one group’s security is another group’s threat, a dangerous spiral will ensue. If the state prevents ethnic groups from providing their own security, then it can solve the security dilemma. Another reason why secessionism can increase when the state collapses, then, is that the security dilemma has emerged. When a central government’s power collapses, it is also unable to combat the new secessionist strength that emerges. Thus, state collapse should facilitate actual secessions, especially when the previous regime was repressive (H5b). Given the paucity of actual secessions, it is easy to examine this hypothesis qualitatively. Where secession did not happen due to the interests of the central government (Czechoslovakia, MalaysiaSingapore), it generally came as a result of the central government’s economic (Soviet Union, Yugoslavia) or political (Ethiopia, Indonesia) collapse or defeat in war (Pakistan-Bangladesh, DenmarkIceland).66 Norway’s 1905 secession from Sweden provides a unique example of peaceful, lawful secession that also occurred during a period of stability.67 Leff argues that ethnic autonomy provides a platform for secession during periods of political instability and transition.68 She notes that centralized Spain did not break up during its democratization, while the federal Yugoslavia, USSR, and Czechoslovakia did. Federal structures presumably gave post-Soviet political entrepreneurs a useful base from which to organize nationalist movements.69 However, Spain did not experience an economic crisis at the level the USSR, Yugoslavia, and even Czechoslovakia did. It

44 Secessionism

is hard to evaluate the counterfactual: would these Soviet countries have broken up just as readily had their institutions been more centralized? Regardless, it is clear that extreme political instability was a critical factor in disintegration. Cetinyan and Jenne approach the issue of ethnic demands from the perspective of relative capacities of the state and ethnic minority.70 The stronger the state relative to the minority group, the more the group will moderate its demands, because it is less willing to risk a rebellion that it would likely lose. Thus, we might expect that a minority group will shrink from demanding full independence or perhaps even widespread autonomy when its capability is low, even if the political and economic benefits of sovereignty dealt with in the previous section are high. Jenne argues that the support of a kin-dominated state will therefore encourage organized separatism among minority groups. One problem with this argument is that most kin-dominated states will not offer strong support to their kindred across the border, because they will shrink from initiating the kind of interstate conflict necessary to achieve irredentist aims. Jenne would predict a positive association between irredentist potential and secessionism, while this book predicts a negative one.

STRATEGY: THE CONSEQUENCES OF SECESSIONISM FOR REBELLION AND DECENTRALIZATION HYPOTHESIS 6 (H6) Moderately weak, ethnoregionally diverse states will prohibit secession because they are willing to tolerate a certain amount of secessionist violence in order to prevent the actual breakup of the state. HYPOTHESIS 7A (H7a) Secessionist and irredentist movements should generally undertake more violent rebellion against the state than non-separatist national and ethnic movements, because their inability to guarantee not to use autonomy to press further demands causes states generally to be unwilling to compromise with them. HYPOTHESIS 7B (H7b) Exception: groups enjoying quasi-legal secession should rebel less than other groups, and there should be almost no positive association between secessionism and rebellion where secession is quasi-



The Positive Theory of Secession

45

legal, because in these countries secessionists can commit to using the political process and governments can commit to respecting autonomy. HYPOTHESIS 7C (H7c) Other factors that affect credibility of autonomy, such as checks and balances in the central government and valuable natural resources in the peripheral region, will affect likelihood of both rebellion and (de)centralization. HYPOTHESIS 7D (H7d) Actual decentralization implies that both sides have resolved their credibility problems, and rebellion should be less likely after an increase in regional autonomy. HYPOTHESIS 8 (H8) For reasons of partisan competition, democracies with secessionist parties will decentralize faster than democracies without ­secessionist parties. Governments that depend on the political support of a partially secessionist minority should be more willing to offer autonomy to that minority than governments that do not depend on such support. Thus far, the argument has been that a credible commitment problem with respect to minorities causes minority nations to demand self-government. When limited self-government is unavailable, or when the benefits of full self-government are clear, minority nations demand full self-government or secession. Sometimes these demands remain latent, however. The government’s policies toward secessionists affect whether latent demands translate into actual organization. Governments that prohibit secessionist political parties or punish pro-secession speech can prevent some secessionists from forming organizations or promoting their cause. One cost of this governmental strategy is that it exacerbates the problem of non-credible commitments to respecting minority rights. By taking away the option of even threatening secession, governments reduce the ability of minorities to threaten governments into respecting their rights. Minorities are more insecure when secession and secessionism are prohibited, and therefore, when they get the opportunity, they are more likely to pursue independence – illegally if necessary – than

46 Secessionism

groups not prohibited from seceding. Counterintuitively, banning secession can promote secessionism. The analysis of secessionist rebellion starts with the rationalist model of war.71 Because violent conflict always destroys resources, it seems that there should always be some compromise solution for any conflict of interest that avoids violence and that all sides prefer to violence. The only conditions under which violence should occur are: informational asymmetries, non-credible commitments, and indivisible goals. Indivisible goals are rare in politics, and, as this book will show, independence is not necessarily indivisible: autonomy compromises are always feasible in theory and frequently so in practice.72 Informational asymmetries occur when adversaries have incentives not to reveal their true capacities for war to each other. Informational asymmetries do a better job of explaining why conflict breaks out than why it persists, since capabilities should be proved on the battlefield fairly quickly. As chapter 4 shows, secessionist civil wars tend to last longer than other civil wars when combined with what Fearon calls “sons of the soil dynamics.”73 Credible commitment problems are a more fruitful place to look for reasons why conflicts cannot be resolved. I now argue that credible commitment problems are also responsible for secessionist rebellion.74 The first problem is a credibility problem for secessionists. Central governments should be reluctant to offer regional autonomy as a compromise to prevent or end a secessionist conflict, because they are worried that secessionists will use the resources of autonomous government to build a stronger nationalist constituency, accumulate economic and political resources, and reopen demands for further autonomy or independence. Secessionists are unable to offer credible guarantees of forsaking secessionist aims after they have attained some political power. Likewise, central governments enjoy a natural advantage over regional governments and cannot usually guarantee that they will not abrogate or erode autonomy in the future. Because governments and secessionists are often unwilling to trust each other to abide by the outcome of a negotiated settlement, secessionist revolts will tend to last longer than other kinds of ethnic conflicts. An apparent example of just this kind of mistrust is the situation in Southern Sudan as of this writing, where the autonomous Government of Southern Sudan is stockpiling weapons on the rationale that they do not trust the



The Positive Theory of Secession

47

central government to abide by the results of a 2011 referendum on independence.75 An alternative explanation of the separatism-rebellion link in the literature is that governments are more likely to fight separatists because they believe that compromising with separatists sends a stronger signal of weakness to foreign adversaries and other potential separatists at home.76 However, when governments and separatists fight, the results on the battlefield should inform all observers of the relative capabilities of the two sides. This reputational argument would not explain why separatist conflicts would last longer than others, as chapter 4 shows that they do. Providing a credibly legal path to secession may reduce or eliminate the secessionism-violence link because it sharply limits the ability of secessionists in an autonomous government to legitimize an insurgency against the government; instead, they must direct their efforts toward persuading their own people to support independence, typically in a referendum. Also, if secession is permitted, a regional minority can better trust the central government not to repeal or infringe on their autonomy in the future, since they enjoy a strong remedy in such an event. Of course, if the secession right is itself in question, as it might be in Southern Sudan, then these beneficial effects would not obtain. In general, then, the tie between secessionism and rebellion should hold in countries that prohibit secession, but not in countries that allow it. Furthermore, countries that permit secession should be more willing to concede autonomy than other countries (H7b). Since the vast majority of governments do not permit secession, secessionism will usually correlate with rebellion (H7a). Why would governments prohibit secession even if doing so leads to violence? The answer must be that these governments are willing to exchange a certain level of violence in exchange for something else of value. Prohibiting secession should discourage some regional minorities from launching secessionist drives. The policy in a sense raises the bar, and the stakes, of secession attempts. Winning a democratic referendum is not sufficient; instead, nationalists must organize a successful rebellion in order to achieve independence. Fewer will bother. Thus, governments that are particularly worried about many secessionist challenges in the future will be less likely to offer a legal path to independence. These might tend to be moderately weak states that might be worried that legalizing secession sends a signal of weakness. Clearly strong and clearly weak states should

48 Secessionism

not worry about sending such a signal, since their strength, or lack thereof, is common knowledge. Also, states with many ethnoregional minorities that could organize secession drives in the future will be less likely to offer any of them a right to independence, out of a fear that all of them will want the same (H6). This book also investigates other factors that could adversely affect governments’ ability to tie their own hands and prevent themselves from meddling with autonomy arrangements. Governments with few checks and balances at the centre, especially autocracies with unlimited executive power, can repeal self-government more or less at will – which means that autonomy compromises are generally unavailable for ending secessionist rebellions. Thus, secessionist rebellions in autocracies could start more often and last longer than both non-secessionist rebellions in autocracies and secessionist rebellions in democracies, particularly highly constrained ones, and highly constrained regimes will also decentralize more often and centralize less often than autocratic ones (H7c). Of course, credibility is never perfect. Checks and balances are probably very nearly a necessary but not sufficient condition for relatively credible autonomy arrangements. It is always possible that all branches of the central government in a constitutional democracy will conspire to erode or eliminate autonomy for a minority. Another factor potentially affecting credibility adversely is natural resources. If a region possesses valuable natural resources, the regional government will wish to control the revenues from those resources, but the central government could worry that a wealthy regional government will be more likely to threaten secession and to prepare for that eventuality. Moreover, central governments will be tempted to tamper unilaterally with regional autonomy in order to take some of the revenues from mineral extraction. There are several arguments in the literature why mineral resources might provoke civil wars.77 Three are particularly relevant to secessionism. First, mineral resource deposits can increase an otherwise poor, peripheral region’s perceived viability as an independent state. Thus, natural resources can encourage secessionism directly, a possibility tested in the next chapter. Second, lootable mineral deposits encourage violent contests for control of the deposits. In this line of argument, mineral resources simply encourage insurgency and civil war in general. Third, natural resources might reduce regional and central government credibility as argued in the p ­ revious para-



The Positive Theory of Secession

49

graph. On this account, secessionist conflicts in regions with natural resources will last longer than secessionist conflicts in resource-poor regions, and resource-poor regions will be more likely to receive and retain autonomy (H7c). These last two arguments are tested in chapter 4. Now assume that in some countries regional autonomy is (relatively) credible. Governments in such countries should be able to reduce violence by decentralizing. Whenever we observe decentralization actually occurring, we can reasonably infer that the c­ entral government does not believe that autonomy will increase the probability of a future secession attempt, at least in the short term. Thus, decentralization should be associated with a short-term r­eduction in the likelihood of secessionist rebellion, if governments are rational (H7d). In one sense the logic of the autonomy bargain is obvious: if seces­sionist rebels get part of what they want, they may be w ­ illing to compromise and lay down arms rather than holding out for independence and getting nothing. Spatial logic goes further and implies that increases in autonomy generally reduce popular support for the secessionist option. In countries with secessionist parties, offers of autonomy should forestall secessionism if some voters are conditional secessionists, preferring independence to the status quo but not to substantial autonomy, and if there is full information about state capabilities, so that offering autonomy is not a signal of weakness. Regional autonomy could be more than a short-term solution for violent conflict and a long-term remedy for secessionist demands themselves. On the other hand, regional autonomy could promote popular confidence in regional self-government as well as secessionist capa­ bilities, in the manner earlier described. Does regional autonomy reduce or promote popular support for secession or the likelihood of secessionist rebellion in the long run? The foregoing analysis gives us no straightforward prediction. Chapter 3 tests the possibility that regional autonomy affects secessionist electoral vote share, which presumably reflects popular support for independence or nearindependence over the status quo.78 Chapter 4 tests the effects of autonomy on secessionist rebellion. If decentralization does not reduce secessionism, why would governments ever pursue it? In countries with competitive elections, the governing party may have direct or indirect interests in d ­ ecentralization

50 Secessionism

that would outweigh governments’ universal default position against alienating any of their authority. One hypothesis is that governing parties simply do what the voters want. If voters in a region would benefit from greater autonomy, that fact would be reflected in higher secessionist vote share, as some voters would prefer independence to the status quo, but also in greater willingness of the central government to decentralize power in order to win the votes of autonomist voters. I call this the “constituency interests” hypothesis. An alternative hypothesis is that governing parties decentralize when they expect to lose power at the central level but hope to win it at the regional level. I call this the “partisan decentralization” hypothesis (H8). Both of these hypotheses will be tested on case studies of Western democracies in chapter 5. In democracies, we should expect this relationship to hold: that secessionist regions would receive more autonomy than nonsecessionist regions. Thus, there is an interactive effect between ­ institutions and secessionist organization on decentralization. Governments that proscribe secession may worry that secessionist groups could use their autonomy to build up resources for rebellion. On the other hand, these countries might also see autonomy as the only way to prevent rebellion. However, since autonomy is credible only when executive constraints are high, autocracies should be relatively impervious to demands for autonomy. Highly constrained regimes should be more willing to grant autonomy to secessionists than autocracies, and in constrained regimes secessionists receive autonomy more often than non-secessionists do (H7c).

SUMMARY OF HYPOTHESES Table 1.1 summarizes the hypotheses and provides a guide to the chapters in which they are tested. If my hypotheses about the effects of prohibiting secession and secessionism are correct, then one policy implication is that, if they are interested in reducing violent conflict, governments should ideally take a permissive line to secession (to reduce rebellion), but, if that is not possible, they should prohibit secessionist organizations altogether (to reduce secessionist agitation, which can lead to rebellion). Counterintuitively, toleration of secession appears more desirable because secessionist goals are otherwise so closely related to rebellion. Nevertheless, most governments will not take a tolerant approach to secession because they are

The Positive Theory of Secession



51

Table 1.1 Guide to Hypotheses Hypothesis

Description

CAUSES OF SUPPORT FOR SECESSION H1 Economic benefits H2 Irredentist potential H3 Potential access to power H4 Availability, levels, and change of regional autonomy; discrimination CAUSES OF COLLECTIVE ACTION FOR SECESSION H5a Toleration of secessionist political parties H5b-c State collapse, democratization, & secession CAUSES OF PROHIBITION OF SECESSION AND/OR SECESSIONIST PARTIES H6 Per capita income, oil exporters, ethnic diversity, new states

Chapters Tested 2; 3 2; 3 2; 3 2; 3

2 1; 4 4

CONSEQUENCES OF SECESSIONISM FOR REBELLION H7a Secessionism and irredentism H7b Quasi-legal secession H7c Checks and balances, natural resources H7d Decentralization

4 4 4 4

CONSEQUENCES OF SECESSIONISM FOR DECENTRALIZATION H7b Quasi-legal secession H7c Checks and balances, natural resources H8 Partisan decentralization

5 5 5

willing to endure some level of violence in order to send a signal to other governments that they will aggressively combat their enemies and in order to reduce, even by a tiny margin, the possibilities of actual secession from the state.

2 Explaining the Causes of Secessionism Worldwide

GOVERNMENT APPROACHES TO SECESSION When one examines patterns of secessionism worldwide, it immediately becomes clear that countries have followed different policies with respect to secessionists, with different results. In Western Europe and North America, secessionist movements organize as political parties and contest elections. Some of these countries, such as Great Britain and Canada, have admitted that secession is in principle negotiable, while other countries, such as France, Spain, and Italy, have asserted their eternal indivisibility.1 Interestingly, most democracies outside the developed West have consistently prohibited secessionist parties: India, Turkey, the Philippines, and Papua New Guinea are important examples. Ethiopia expressly permits secession in its constitution but has used methods of cooptation and control to suppress secessionist mobilization in some of its peripheral regions.2 Pakistan has inconsistently suppressed secessionist parties. Nevertheless, democracies still offer more room for collective action by national minorities than do autocracies. India, the Philippines, and Papua New Guinea have all allowed former secessionist rebels to contest elections and hold office after renouncing their former views. This chapter tests whether the economics of independence, the inter­national environment, and the prospects for power at the centre affect the extent of popular support for secessionist movements. I find fairly robust evidence that geography, irredentist potential, past autonomy, kin ties with another secessionist group, number of other secessionist groups in the country, and relative size importantly affect the degree of secessionism in an ethnic group. There is also weaker evidence that economic differentials and total population



Explaining the Causes of Secessionism Worldwide

53

matter. Finally, secessionist policy and regime type have some interesting effects on secessionism, suggesting that governments wishing to suppress secessionism altogether should be democratic but should also prohibit secessionist movements. The next section of this chapter presents the results of a global statistical analysis of the determinants of popular support for secessionist organizations. The third section takes a closer look at those cases that the model does not explain well. The final section of the chapter concludes with an overview of the findings.

THE CAUSES OF SECESSIONISM: GLOBAL ANALYSIS This section uses statistical analysis to look at the effects of regime type, secessionist policy, and the benefits of sovereignty on an ethno­ national group’s choice to pursue the goal of self-government. Chapter 1 argues that more members of a national group will support secession the higher the economic and political benefits of sovereignty compared to limited self-government, and the less credible or possible limited self-government is (H1–H4). Some democracies and autocracies punish secessionist mobilization, and therefore secessionist sentiment should remain more often latent and unobserved in these countries than in the advanced democracies (H5a). On the other hand, banning secessionism might cause collective insecurity to rise and increase the latent level of secessionism. Therefore, it is not completely certain that countries prohibiting secessionism will on balance have fewer active secessionist movements than advanced democracies. Groups that enjoy the prospect of controlling the central government through minority rule will see less need to secede (H3). Therefore, groups that enjoy less prospect of controlling the state in autocracies will be more secessionist. Measuring Secessionism If we were interested in the level of latent secessionist support in a population, the ideal way to measure it would be through p ­ ublic opinion polls. Public opinion polls asking voters’ opinions on independence are relatively common in Quebec, Scottish, or Basque contexts but are unknown in most of the world. In any case, what we ultimately want to know is what determines the strength of secessionist organizations, not what people might secretly believe. In advanced

54 Secessionism

democracies we could measure secessionist strength by vote share for secessionist parties in elections, controlling for electoral systems, and the next chapter does this. In the rest of the world we could measure secessionist strength by number of members of militant or clandestine secessionist groups. For obvious reasons, these figures are generally unavailable. The best way to do a worldwide quantitative study of secessionism, then, is to use a simple ordinal variable that measures quite roughly the support of organizations advocating extensive self­government among a territorially concentrated minority population. These organizations could be either rebel armies, extrapolitical pressure groups, or, in advanced democracies, political parties. This is the dependent variable, the outcome we want to explain. As chapter 1 noted, to explain the strength of secessionist organizations we have to employ variables capturing incentives for collective action for selfgovernment as well as variables capturing latent popular support for self-government. The observations we want to use are either “regions” or “national groups.” It does not matter a great deal which kind of observation is used. If we use regions, then one of the independent variables we need to use is a measure of the proportion of the regional population that belongs to a distinct national group. With this setup, we can learn the effects that national identity has on desire for self­government. If we use national groups as the observations, then we have tossed out regions without national groups, and we cannot learn the effects of national identity on desire for ­self-government, because the whole sample we are examining consists of groups that already have a national identity – there would be no control group. The next chapter uses regions in an analysis of secessionism in advanced ­democracies, because data on regions are available for those countries. Most countries in the world do not have good data available on their regions. There is, however, a dataset on “ethno­ political groups,” created by the Minorities at Risk (MAR) Project. This project codes over 270 communal groups on their economic, political, cultural, and demographic characteristics and environments. Studies have used these data to analyze over time trends and differences across countries in ethnic mobilization and ethnic conflict.3 Two studies have used this dataset to analyze the determinants of separatism and irredentism in the 1980s and 1990s.4



Explaining the Causes of Secessionism Worldwide

55

I created a secessionism variable from scratch, taking on a value of “4” for groups for which a majority of members appear to support (not necessarily participate in) secessionist organizations, “3” for groups for groups for which roughly 25 to 50 percent of members support secessionist organizations, “2” for groups for which roughly 5 to 25 percent of members support secessionist organizations, “1” for groups that have had any secessionist mobilization on the fringes, representing less than 5 percent of the group, and “0” for groups that have not had any evidence of secessionist mobilization at all.5 The variable does change over time for some groups, but in most of the world, it is difficult to tell when exactly transitions between different intensities of secessionism occur, since we cannot use electoral data to examine secessionist support directly. One oft-mentioned limitation of the MAR dataset is that it is limited to ethnic groups that are either politically active or somehow discriminated against or treated differently by government or social actors. Thus, any analysis limited to these groups will have to ignore those factors that cause an ethnic group to become politically active or targeted in the first place. To take those factors into account we would need a dataset with all ethnic groups. This is a potential problem for us because one of the interesting differences between countries that tolerate secessionist movements and countries that prohibit them is the fact that the latter are supposed to prevent latently secessionist groups from becoming actively secessionist. The problem is limited by the fact that a latently secessionist group is probably politically active in several other ways. I addressed this problem by researching and adding several dozen new cases, both from advanced democracies and from the rest of the world. Table A.1 in the appendix shows all the groups in this dataset, their country of residence, their region of residence, and secessionist status in 2003.6 I have also coded new variables, described below, for all the MAR cases. Finally, there were also some minor coding errors in the original MAR dataset, which I repaired.7 Because the original MAR dataset contains a wide variety of ethnic groups, some of which could never have any secessionist potential, the dataset was culled in the following way. All groups without a regional base were eliminated, where a regional base is defined as “a spatially contiguous region larger than an urban area that is

56 Secessionism

part of the country, in which 25 percent or more of the minority resides and in which the minority constitutes the predominant proportion of the population.”8 In other words, ethnic groups that are spread over a wide range of territory and are not a majority or significant minority anywhere were removed. The option of secession, or any kind of territorial autonomy, simply does not exist for these groups. In addition, groups that have a regional concentration but do not reside in their historical homeland9 were removed. A good example of a regionally concentrated group not residing in their historical homeland is ­African-Americans, who are regionally concentrated in the southeastern United States, but who do not constitute an absolute majority in any one state and do not consider the southeast to be their historic homeland. The data also showed that none of these removed groups are secessionist. Finally, groups that control their states (“dominant minorities”) were removed for those years that they had such control; as one might expect, none of these are secessionist.10 After removing these groups, 283 ethnonational minorities were left for the year 2003.11 Of these, 107 groups, 38 percent of the total, had secessionist organizations in that year. The prevalence of secessionism has increased slightly over time: in 1986, 31 percent of groups had secessionist organizations. The next step is to test the “risk factors” for secessionism in a regression format. When the dependent variable to be explained, secessionism, takes on ordinal values, an ordered maximum likelihood estimation procedure such as ordered probit or ordered logit is potentially appropriate. After ensuring that the model assumptions are met, I run an ordered logit analysis.12 With this format we can try to find a model, a set of explanatory variables, that successfully predicts which ethnopolitical groups achieve higher levels of secessionism. Each independent variable tests a separate hypothesis about the causes of secessionism. The next section recapitulates the hypotheses from chapter 1 about the causes of secessionism and describes the variables I use to test them. The Causes of Secessionism: Hypotheses and Variables The first set of hypotheses has to do with the political and economic costs and benefits of self-government. Because the dataset I am using in this chapter is already limited to ethnonational minority groups,



Explaining the Causes of Secessionism Worldwide

57

all groups in the dataset could derive cultural benefits from selfgovernment, including new linguistic or religious policies, symbolic place name changes, official support for traditional customs, and the like. The next chapter will test how much these benefits matter in advanced democracies, but we cannot test this here. Hypothesis H1 says that more affluent (in democracies only), resource-rich, populous, and geographically well-situated regions/ groups will be more secessionist. To test affluence, I use an indicator from MAR called “economic differentials,” which measures group economic disadvantage on a variety of dimensions (one of those dimensions is “income,” and I try using this variable by itself as well, with no difference in results). It takes positive values for worse-off groups, with “2” the maximum and “–2” the minimum, while “0” represents groups roughly equally affluent to other groups in society.13 Therefore, the variable should have a negative effect on secessionism in democracies and less or no effect in autocracies (because in democracies the median voter favours redistribution from the rich, while in autocracies redistribution does not necessarily flow from rich to poor). Unfortunately, I do not have access to economic data for the new cases I coded, so I run the analysis both with and without the economic differentials variable. The economic differentials variable is included on its own and is also multiplied by the commonly used “Polity” index of democracy, which runs from –10 to 10, to create an interaction variable, which tests whether economic differentials have different effects in democracies and autocracies. To test resource wealth, I collected data on mineral resource production value by subnational region, which is defined by the regional base of each ethnopolitical group in the dataset. These data are calculated only for MAR ethnopolitical groups, which explains why this variable is included only in the models that also include economic differentials. The figures represent the value of total mineral production in the region in a year, divided by the group’s population. The construction of the variable is described elsewhere.14 The large majority of ethnoregions in the dataset score “0” on the variable. The most resource-rich ethnoregion in the dataset in 2003 is the San Bushmen region of Botswana, followed by Yakutia, Indigenous Lowland Peoples in Peru, and the Shi’a region of Saudi Arabia. The variable (plus one) is logged before inclusion in the regressions. To test population, I use the natural logarithm of group population, taken from MAR and other sources,15 which should have a

58 Secessionism

positive effect. For geography, I use two dummy variables, one for geographic separation from the mainland (no road access with the rest of the country) and one for sea access for the group’s regional base. All geographically separate regions also have sea access. Both variables should positively influence secessionism. Hypothesis H2 says that groups with irredentist potential are less likely to be secessionist. I test this hypothesis with another dummy variable, scored “1” for groups living in regions near states where a close kin group holds political power, for those years for which that is true. Hypothesis H3 says that groups large enough to have the opportunity of seizing political power over the whole country are less likely to be secessionist. To some degree, the prediction of this hypothesis cuts against that of hypothesis H1, but a group large in absolute terms need not be large in relative terms. In order to test it separately, I’ve used a dichotomous (dummy) variable that has been scored “1” for those groups that are either: 1) the largest ethnic group in the country, or 2) the second largest ethnic group in a country in which no group constitutes 60 percent of the population or more. Hypothesis H4 says that groups suffering discrimination are more likely to support secession. To test it, I developed a three-point scale of systematic discrimination by public officials, scored “2” if a group suffers both economic and political discrimination from the government according to the MAR dataset, “1” if the group suffers one kind of discrimination but not the other, and “0” if the group does not suffer governmental discrimination. I do not test the effects of autonomy on secessionism in this chapter, because autonomy is endogenous: secessionist groups might receive more autonomy than non-secessionist groups, generating a spurious correlation. In the next chapter, I have enough over-time data to look at the effects of autonomy on secessionist vote share in developed democracies. The second set of hypotheses has to do with the incentives for groups to organize for collective action aimed at collective self-govern­ment. Since our measure of secessionism is whether a group does or does not have secessionist organizations, we need to use these hypotheses to explain the secessionism we observe. For example, secessionism should be rarer in countries that prohibit seces­sionist organizations. Hypothesis H5a says that groups in countries that permit secessionist organizations are more likely to develop them. I test this hypo­thesis with another dummy variable, scored “1” for groups



Explaining the Causes of Secessionism Worldwide

59

in countries that permit secessionist organizations.16 This variable should have a positive effect on the dependent variable, observed secessionism. Alternative arguments suggest that more repressive regimes would see more secessionism because of minority resentment. While policies of discrimination or repression should increase latent secessionist support (H4), legal suppression of secessionist organizations should probably reduce their observed frequency. Hypothesis H4 also suggested that democracies may see less secessionism than autocracies, because democracies can offer limited autonomy to groups that desire it, while such arrangements are not credible in autocracies with few executive constraints: thus, in auto­ cracies, groups that would ideally prefer limited autonomy instead see secession as their only option for achieving any autonomy. The Polity measure of democracy is included in the models on its own as well as in an interaction with economic differentials (see above).17 I expect it to have a negative effect on secessionism, except perhaps at low levels of economic differentials. We also need to consider some control variables. Territories with an international law right to independence or previous self-government that has since been lost should be more likely to develop secessionist organizations. To control for this possibility, I use a variable from MAR called “lost autonomy grievances” and convert it to a dummy variable: if the group formerly enjoyed independence or a traditional form of autonomy, or was granted a right to independence under international law that was not respected, it is scored “1.”18 Groups with autonomy grievances should at least view independence as an option and are more likely to be secessionist. Groups are more likely to develop secessionist movements when many other groups in the same country have also developed secessionist movements. One common claim about secessionism is that it is contagious, spreading from country to country or group to group. While I concur with Ayres and Saideman in finding little support for this thesis in general, there is one plausible construction of the idea: that secessionist groups influence other ethnic minorities within the same country. Thus, I have used a simple measure for this concept, the number of other secessionist national groups in the country. Some groups inhabit regions straddling interstate borders. The Kurds live in Turkey, Iraq, Iran, and Syria. If a group is secessionist with respect to one of the countries it inhabits, it is almost always secessionist with respect to the other countries it inhabits. I measure

60 Secessionism

secessionist kin with a dummy variable, scored “1” for groups with secessionist kin in another country, for those years for which that is true. In 2003 there were forty-four groups with secessionist kin, and only nineteen of them did not have secessionist movements of their own. In 1986 there were twenty-nine groups with secessionist kin, and only thirteen of them did not have secessionist movements of their own. Although all groups in the analysis constitute a “predominant proportion of the population” in their regional base, some groups are less than 50 percent, presumably a mere plurality of the local population. I created a dummy variable for these groups, which might be more reluctant to try to secede, fearing they would not enjoy true self-determination in any autonomous territory. In addition to the foregoing predictions, there are additional hypotheses about the causes of secessionism that we cannot test with the MAR data. Some have to do with the formation of cultural identity in the first place, some with over-time change in secessionism, and some with concepts for which we do not have reliable, quantifiable variables. I will discuss these factors later in this chapter and in the next. The Causes of Secessionism: Results Table 2.1 shows the results for four statistical models. Model I is for the year 2003 and includes the new cases I coded. Model II is for the year 2003 and includes economic differentials, its interaction with democracy, and logged mineral resources, thus requiring the exclusion of the new cases. Model III is for the year 1986 and includes my new cases, while model IV is for the year 1986 and includes economic differentials. Another option is to run pooled cross-sectional time-series regressions rather than just the cross-sectional regressions. However, it turns out that the lagged dependent variables soak up all the variance, and nothing else is significant when this is done. The problem is that secessionism changes gradually over the long term, and I do not have enough years in the dataset to go back and predict the emergence of secessionism in most cases.19 The hypotheses that receive the weakest support have to do with economic differentials and democracy, mineral resources, population, and irredentist potential. In 1986, the expected, negative effects of economic differentials and the interaction with democracy are

Table 2.1 Ordered Logit Tests of Explanations for World Secessionism

Variable

Model I (2003) Coef. (SE)

Economic differentials Econ diffs*Democracy (–) Logged population (+) Geographic separation (+) Sea access (+)

 0.08 (0.46) –0.80 (0.38)*  1.35 (0.39)*** –0.087 (0.031)** 1.00 (0.30)*** 1.76 (0.43)*** 0.17 (0.06)** 2.05 (0.39)*** –0.76 (0.40)

–0.067 (0.099) 0.016 (0.016) 0.26 (0.13)* 1.75 (0.67)** 0.50 (0.41) 0.012 (0.015) –0.16 (0.50) –1.03 (0.40)** 0.85 (0.48) –0.11 (0.04)** 0.88 (0.30)** 1.50 (0.63)* 0.21 (0.07)** 2.37 (0.46)*** –1.00 (0.46)*

268 42.5%

173 42.8%

 0.06 (0.10)  1.04 (0.47)*  0.90 (0.37)**

Mineral resources (+) Irredent. potential (–) Relative size (–) Secessionism permitted (+) Democracy (–) Discrimination (+) Lost autonomy Other secessionists Secessionist kin Regional minority

Model II (2003) Coef. (SE)

Soviet bloc N McKelvey-Zavoina Pseudo R2

Model III (1986) Coef. (SE)

0.09 (0.14) 0.14 (0.51) 1.14 (0.46)**

–0.64 (0.44) –1.17 (0.43)** 1.11 (0.58)* –0.01 (0.03) 0.71 (0.33)* 1.86 (0.67)** 0.23 (0.07)*** 1.61 (0.45)*** –0.62 (0.36) –0.64 (0.78) 248 40.9%

Model IV (1986) Coef. (SE) –0.20 (0.11) –0.023 (0.017) 0.19 (0.18) 1.08 (0.74) 1.31 (0.60)* 0.002 (0.017) –0.53 (0.48) –1.78 (0.58)** 0.50 (0.60) –0.02 (0.04) 0.32 (0.31) 1.05 (0.57) 0.35 (0.08)*** 1.53 (0.45)*** –0.59 (0.45) –2.46 (0.58)*** 171 45.2%

Supported?

No Yes, weakly Yes Yes No No Yes, strongly Yes Yes, weakly Yes

Notes: cut points not reported. Signs in coefficients represent hypothesized relationships. All hypotheses tested with one-tailed z-tests; all controls tested with two-tailed z-tests. Robust standard errors clustered on country reported. Data units: ethnic groups that are geographically concentrated in a homeland. The last column reports whether the results support the hypotheses or not. ***p