Authoritarian Politics: Power-Sharing, Institutions, and Party Role - Prof. Aznar, Apuntes de Ciencia Política

¡Descarga Authoritarian Politics: Power-Sharing, Institutions, and Party Role - Prof. Aznar y más Apuntes en PDF de Ciencia Política solo en Docsity! The Politics of Au APT Milan W. Svolik more information - www.cambridge.org/9781107024793 Cambridge Studies in Comparative Politics General Editor Margaret Levi University of Washington, Seattle Assistant General Editors Kathleen Thelen Massachusetts Institute of Technology Erik Wibbels Duke University Associate Editors Robert H. Bates Harvard University Stephen Hanson The College of William and Mary Torben Iversen Harvard University Stathis Kalyvas Yale University Peter Lange Duke University Helen Milner Princeton University Frances Rosenbluth Yale University Susan Stokes Yale University Other Books in the Series Ben W. Ansell, From the Ballot to the Blackboard: The Redistributive Political Economy of Education David Austen-Smith, Jeffry A. Frieden, Miriam A. Golden, Karl Ove Moene, and Adam Przeworski, eds., Selected Works of Michael Wallerstein: The Political Economy of Inequality, Unions, and Social Democracy Andy Baker, The Market and the Masses in Latin America: Policy Reform and Consumption in Liberalizing Economies Lisa Baldez, Why Women Protest: Women’s Movements in Chile Stefano Bartolini, The Political Mobilization of the European Left, 1860–1980: The Class Cleavage Robert Bates, When Things Fell Apart: State Failure in Late-Century Africa Mark Beissinger, Nationalist Mobilization and the Collapse of the Soviet State Pablo Beramendi, The Political Geography of Inequality: Regions and Redistribution Nancy Bermeo, ed., Unemployment in the New Europe Series list continues following the Index. The Politics of Authoritarian Rule MILAN W. SVOLIK University of Illinois, Urbana-Champaign Contents Figures page xiii Tables xv Acknowledgments xvii 1 Introduction: The Anatomy of Dictatorship 1 1.1 The Two Problems of Authoritarian Rule 3 1.1.1 The Problem of Authoritarian Power-Sharing 5 1.1.2 The Problem of Authoritarian Control 9 1.2 The Authoritarian Setting 13 1.3 Plan of the Book 17 2 The World of Authoritarian Politics 19 2.1 What Counts as a Dictatorship? 22 2.2 Making Sense of Institutional Heterogeneity under Dictatorship 26 2.3 This Book’s Approach to Political Organization of Dictatorships 32 2.4 Leadership Change in Dictatorships 39 2.5 The Survival of Authoritarian Ruling Coalitions 41 2.6 Conclusion: The Rich World of Authoritarian Politics 43 2.7 Appendix: Authoritarian Spells, 1946–2008 45 part i the problem of authoritarian power-sharing 3 And Then There Was One! Authoritarian Power-Sharing and the Path to Personal Dictatorship 53 3.1 Authoritarian Power-Sharing and the Emergence of Personal Autocracy 57 3.2 A Formal Model 63 3.2.1 Authoritarian Power-Sharing 65 3.2.2 Balance of Power and Authoritarian Power-Sharing 68 3.2.3 A Model with Endogenously Evolving Balance of Power 71 ix x Contents 3.3 Implications for the Empirical Study of Authoritarian Tenures 72 3.4 The Path to Personal Autocracy and Authoritarian Exit from Office 75 3.5 Conclusion: The Management of Established Autocracy 78 3.6 Appendix: Proofs 81 4 When and Why Institutions Contribute to Authoritarian Stability: Commitment, Monitoring, and Collective Action Problems in Authoritarian Power-Sharing 85 4.1 Political Institutions and Authoritarian Power-Sharing 88 4.2 The Allies’ Collective-Action Problem and Credible Power-Sharing 94 4.3 A Formal Model 100 4.3.1 Allies’ Rebellion as a Collective-Action Problem 102 4.3.2 Authoritarian Power-Sharing without Institutions 106 4.3.3 Authoritarian Power-Sharing with Institutions 108 4.4 Power-Sharing Institutions and Authoritarian Stability 110 4.5 Conclusion: The Distinctly Authoritarian Purpose of Nominally Democratic Institutions in Dictatorships 116 4.6 Appendix: Proofs 117 part ii the problem of authoritarian control 5 Moral Hazard in Authoritarian Repression and the Origins of Military Dictatorships 123 5.1 The Moral Hazard Problem in Authoritarian Repression 127 5.2 Bargaining in the Shadow of Military Intervention 134 5.3 A Formal Model 138 5.4 Empirical Analysis 148 5.5 Conclusion: The Political Price of Authoritarian Repression 159 5.6 Appendix I: Proofs 159 5.7 Appendix II: Multiple Imputation 160 6 Why Authoritarian Parties? The Regime Party as an Instrument of Co-optation and Control 162 6.1 The Logic of Party-Based Authoritarian Co-optation 167 6.1.1 Hierarchical Assignment of Service and Benefits 168 6.1.2 Political Control over Appointments 178 6.1.3 Selective Recruitment, Promotion, and Repression 182 6.2 Regime Parties and Authoritarian Resilience 184 6.3 Conclusion: Why Authoritarian Regime Parties? 192 7 Conclusion: Incentives and Institutions in Authoritarian Politics 196 7.1 Why Dictators Preside over Policy Disasters 197 Figures 1.1 Nonconstitutional exits from office of authoritarian leaders, 1946–2008. page 5 2.1 The number and the change in the number of dictatorships, 1946–2008. 25 2.2 Dictatorship around the world, 1946–2008. 27 2.3 Military involvement in authoritarian politics, 1946–2008. 34 2.4 Restrictions on political parties in dictatorships, 1946–2008. 35 2.5 Joint distribution of restrictions on political parties and military involvement in authoritarian politics, 1946–2008. 36 2.6 Legislative selection in dictatorships, 1946–2008. 36 2.7 Executive selection in dictatorships, 1946–2008. 37 2.8 Constitutional exits from office of authoritarian leaders, 1946–2008. 40 3.1 Authoritarian power-sharing game in extensive form. 65 3.2 The probability that the dictator reneges, α∗ (dashed line), and the probability that the ruling coalition rebels if it observes a high signal, β∗H (solid line), in an equilibrium of the multiperiod authoritarian power-sharing game. 72 3.3 Probability density of time-to-rebellion based on the numerical example. 75 3.4 Probability density of time-to–established autocracy based on the numerical example. 76 3.5 The empirical density of coups d’état, 1946–2008. 76 3.6 The improving odds of dying in bed, 1946–2008. 77 4.1 A model of an authoritarian polity. 101 4.2 Payoffs to ally i , given the proportion of allies that rebel φ. 103 4.3 Effect of the dictator’s power vis-à-vis the allies λ on threshold discount factors δ I (solid line) and δ∼I (dashed line) for λ < κ0. The dotted line plots a hypothetical discount factor δ = 0.85. 110 xiii xiv Figures 4.4 Nonconstitutional leader exits from office in dictatorships with and without legislatures, 1946–2008. 112 5.1 The effect of the expected magnitude of the mass threat R̂ on the equilibrium choice of the military’s resources r∗. 146 5.2 The effect of the magnitude of the mass threat R̂ on the equilibrium probability of a successful military intervention. 147 5.3 Average annual frequencies of military interventions in leader entry or exit, 1946–2002. 151 5.4 The estimated effect of economic inequality on the probability of military intervention. 156 6.1 The share of legislative seats controlled by authoritarian regime parties in dictatorships with multiple parties, 1946–2008. 188 6.2 The association between legislative seat share and the age of authoritarian ruling coalitions, 1946–2008. 191 Tables 1.1 An Outline of the Outcomes Explained in This Book page 18 2.1 The Origin and End of Authoritarian Spells, 1946–2008 26 2.2 Restrictions on Political Parties and Military Involvement in Authoritarian Politics, 1946–2008 32 2.3 Legislative and Executive Selection in Dictatorships, 1946–2008 33 2.4 Joint Distribution of Legislative and Executive Selection, 1946–2008 38 2.5 Political Affiliation of Authoritarian Leaders, 1946–2008 42 4.1 Institutions, Balance of Power, and the Success of Authoritarian Power-Sharing 99 4.2 Legislatures and the Survival of Authoritarian Ruling Coalitions, 1946–2008 111 4.3 Legislatures, Parties, and the Survival of Authoritarian Leaders, 1946–2008 114 5.1 Moral Hazard in Authoritarian Repression and Military Intervention in Politics 138 5.2 Military Intervention in Dictatorships by Level of Economic Inequality 152 5.3 The Impact of Economic Inequality on Military Intervention in Dictatorships 155 6.1 A Comparison of Co-optation Via Transfers and Co-optation Via a Regime Party 171 6.2 Restrictions on Political Parties and the Survival of Authoritarian Ruling Coalitions, 1946–2008 186 6.3 Do the Survival Functions of Authoritarian Ruling Coalitions Differ Depending on Their Restrictions on Political Parties? 187 6.4 Three Subgroups of Authoritarian Regime Parties by Legislative Seat Share 189 6.5 The Effect of a Regime Party’s Legislative Seat Share on the Survival of Authoritarian Ruling Coalitions 190 xv xviii Acknowledgments Center for the Study of Democratic Politics at Princeton University and sta- tioned at the Cline Center for Democracy at the University of Illinois. I would like to thank Larry Bartels and Pete Nardulli for their support during these periods. I have benefited greatly from the generosity and insights of my colleagues at the University of Illinois. Jake Bowers, Xinyuan Dai, Paul Diehl, Tiberiu Dragu, Zach Elkins, Sam Frost, Brian Gaines, Jude Hays, Jim Kuklinski, Bob Pahre, Gisela Sin, Bonnie Weir, and Matt Winters have offered comments and criticisms at various stages of this book. My thanks go especially to José Cheibub, whose encouragement and guidance make this project seem so easy in retrospect. I was fortunate to receive excellent and detailed comments on the entire manuscript from Giacomo Chiozza, Lucan Way, and several anonymous refer- ees. Carles Boix, José Cheibub, Jennifer Gandhi, Monika Nalepa, Pete Nardulli, and Duncan Snidal offered invaluable guidance on the book’s publication. I am also grateful to Hein Goemans, whose data were incredibly helpful at the early stages of this project. Seden Akcinaroglu, Svitlana Chernykh, Aya Kachi, Donksuk Kim, Dan Koev, Alex Sapone, Tatiana Švolı́ková, and Nini Zhang all provided valuable research assistance at various stages of this project. I am especially indebted to Michael Martin for his outstanding help with data collection, editing, and indexing. Students in my 2010–2012 undergraduate and graduate classes on the politics of dictatorships suffered through the early drafts of several chapters – their feedback was instrumental in helping me frame the book’s overarching argument. I would also like to thank my editor at Cambridge University Press, Lew Bateman, for his interest in the project and his consideration of the professional pressures faced by a junior political scientist, as well as Margaret Levi for including the manuscript in the Cambridge Studies in Comparative Politics series. Finally, I thank my family and friends for their support along the way. My wife Bonnie has been a source of endless love and energy at every stage of the writing process. She patiently read and reread, edited and re-edited the entire manuscript. Exactly when it counted, she has been my toughest critic and my greatest supporter. This book is dedicated to my parents, to whom I owe the most. Portions of this book rely on research that has been published or draws on collaborative work. Parts of Chapter 3 have appeared as “Power-Sharing and Leadership Dynamics in Authoritarian Regimes” in the American Journal of Political Science (2009). More importantly, I have had the good fortune to col- laborate with and learn from Carles Boix; parts of Chapter 4 draw on our joint paper “The Foundations of Limited Authoritarian Government: Institutions and Power-Sharing in Dictatorships.” 1 Introduction The Anatomy of Dictatorship Still democracy appears to be safer and less liable to revolution than oligarchy. For in oligarchies there is the double danger of the oligarchs falling out among themselves and also with the people . . . Aristotle, The Politics, Book 5 [W]herein men live without other security, than what their own strength, and their own invention shall furnish them . . . , the life of man [is] solitary, poor, nasty, brutish, and short. Thomas Hobbes, Leviathan Bashar al-Asad was not meant to be a dictator. Although he was the son of Syria’s long-serving president, Hafez al-Asad, Bashar’s education and career were nonpolitical. In 1988, at the age of twenty-three, he received a degree in ophthalmology from the University of Damascus and moved to London four years later to continue his medical residency. Hafez al-Asad had instead groomed Bashar’s older brother, Basil, as his successor. Yet Bashar’s seclusion from politics ended in 1994 when Basil died in an automobile accident. Bashar was recalled from London, entered a military academy, and quickly advanced through the ranks, while his father spent the last years of his life eliminating potential challengers to Bashar’s succession.1 Consider Bashar al-Asad’s delicate position on July 17, 2000, when he became the Syrian president. Given his unexpected path to power, how does he best ensure his survival in office? What threats should he expect and how will he deal with them? Alas, the contemporary political scientist is not well equipped to become the new Machiavelli. If Bashar al-Asad were concerned about politically succeeding in a democracy, students of politics might offer him suggestions ranging from how to best target voters in campaigns to the implications of electoral systems 1 See Hinnebusch (2002), Leverett (2005), and Perthes (2006). 1 2 The Politics of Authoritarian Rule for partisan competition.2 But of course, if Bashar al-Asad lived in a democracy, he would not have been in a position to inherit a presidency. Although growing at a fast pace, contemporary scholarship on dictatorships has so far generated only a fragmented understanding of authoritarian politics. Extant research increasingly studies authoritarian parties, legislatures, bureau- cracies, and elections, as well as repression, leadership change, and regime stability across dictatorships.3 Yet in most cases, these facets of authoritarian- ism are examined individually, in isolation. In turn, we lack a unified theoretical framework that would help us to identify key actors in dictatorships; locate the sources of political conflict among them; and thereby explain the enor- mous variation in institutions, leaders, and policies across dictatorships.4 At both the empirical and theoretical level, we are without a general conceptual heuristic that would facilitate comparisons across polities as diverse as Mexico under the Institutional Revolutionary Party (PRI), Saddam Hussein’s Iraq, and contemporary China. This book attempts to fill that void. I argue that two conflicts fundamentally shape authoritarian politics. The first is between those who rule and those who are ruled. All dictators face threats from the masses, and I call the political problem of balancing against the majority excluded from power the problem of authoritarian control. Yet dictators rarely control enough resources to preclude such challenges on their own – they therefore typically rule with a number of allies, whether they be traditional elites, prominent party members, or generals in charge of repression. A second, separate political conflict arises when dictators counter challenges from those with whom they share power. This is the problem of authoritarian power-sharing. To paraphrase Aristotle’s warning in this chapter’s epigraph, authoritarian elites may fall out both with the people and among themselves. Crucially, whether and how dictators resolve the problems of power-sharing and control is shaped by two distinctively dismal features of authoritarian pol- itics. First, dictatorships inherently lack an independent authority with the power to enforce agreements among key political actors, especially the dicta- tor, his allies, and their repressive agents. Second, violence is an ever-present and ultimate arbiter of conflicts in authoritarian politics. These two intrinsic features uniquely shape the conduct of politics in dictatorships. They limit the role that political institutions can plausibly play in resolving the problems of power-sharing and control, and they explain the gruesome manner in which so many dictators and dictatorships fall. Authoritarian politics takes place in the shadow of betrayal and violence. In brief, the central claim of this book is this: Key features of authoritari- anism – including institutions, policies, as well as the survival of leaders and regimes – are shaped by the twin problems of power-sharing and control against 2 See, e.g., Green and Gerber (2004) and Cox (1997), respectively. 3 See subsequent chapters for a detailed discussion of this literature. 4 Bueno de Mesquita et al. (2003) and Wintrobe (1998) are two notable exceptions to the tendency for fragmentary explanations of authoritarian politics. The Politics of Au APT Milan W. Svolik more information - www.cambridge.org/9781107024793 Cambridge Studies in Comparative Politics General Editor Margaret Levi University of Washington, Seattle Assistant General Editors Kathleen Thelen Massachusetts Institute of Technology Erik Wibbels Duke University Associate Editors Robert H. Bates Harvard University Stephen Hanson The College of William and Mary Torben Iversen Harvard University Stathis Kalyvas Yale University Peter Lange Duke University Helen Milner Princeton University Frances Rosenbluth Yale University Susan Stokes Yale University Other Books in the Series Ben W. Ansell, From the Ballot to the Blackboard: The Redistributive Political Economy of Education David Austen-Smith, Jeffry A. Frieden, Miriam A. Golden, Karl Ove Moene, and Adam Przeworski, eds., Selected Works of Michael Wallerstein: The Political Economy of Inequality, Unions, and Social Democracy Andy Baker, The Market and the Masses in Latin America: Policy Reform and Consumption in Liberalizing Economies Lisa Baldez, Why Women Protest: Women’s Movements in Chile Stefano Bartolini, The Political Mobilization of the European Left, 1860–1980: The Class Cleavage Robert Bates, When Things Fell Apart: State Failure in Late-Century Africa Mark Beissinger, Nationalist Mobilization and the Collapse of the Soviet State Pablo Beramendi, The Political Geography of Inequality: Regions and Redistribution Nancy Bermeo, ed., Unemployment in the New Europe Series list continues following the Index. The Politics of Authoritarian Rule MILAN W. SVOLIK University of Illinois, Urbana-Champaign Contents Figures page xiii Tables xv Acknowledgments xvii 1 Introduction: The Anatomy of Dictatorship 1 1.1 The Two Problems of Authoritarian Rule 3 1.1.1 The Problem of Authoritarian Power-Sharing 5 1.1.2 The Problem of Authoritarian Control 9 1.2 The Authoritarian Setting 13 1.3 Plan of the Book 17 2 The World of Authoritarian Politics 19 2.1 What Counts as a Dictatorship? 22 2.2 Making Sense of Institutional Heterogeneity under Dictatorship 26 2.3 This Book’s Approach to Political Organization of Dictatorships 32 2.4 Leadership Change in Dictatorships 39 2.5 The Survival of Authoritarian Ruling Coalitions 41 2.6 Conclusion: The Rich World of Authoritarian Politics 43 2.7 Appendix: Authoritarian Spells, 1946–2008 45 part i the problem of authoritarian power-sharing 3 And Then There Was One! Authoritarian Power-Sharing and the Path to Personal Dictatorship 53 3.1 Authoritarian Power-Sharing and the Emergence of Personal Autocracy 57 3.2 A Formal Model 63 3.2.1 Authoritarian Power-Sharing 65 3.2.2 Balance of Power and Authoritarian Power-Sharing 68 3.2.3 A Model with Endogenously Evolving Balance of Power 71 ix x Contents 3.3 Implications for the Empirical Study of Authoritarian Tenures 72 3.4 The Path to Personal Autocracy and Authoritarian Exit from Office 75 3.5 Conclusion: The Management of Established Autocracy 78 3.6 Appendix: Proofs 81 4 When and Why Institutions Contribute to Authoritarian Stability: Commitment, Monitoring, and Collective Action Problems in Authoritarian Power-Sharing 85 4.1 Political Institutions and Authoritarian Power-Sharing 88 4.2 The Allies’ Collective-Action Problem and Credible Power-Sharing 94 4.3 A Formal Model 100 4.3.1 Allies’ Rebellion as a Collective-Action Problem 102 4.3.2 Authoritarian Power-Sharing without Institutions 106 4.3.3 Authoritarian Power-Sharing with Institutions 108 4.4 Power-Sharing Institutions and Authoritarian Stability 110 4.5 Conclusion: The Distinctly Authoritarian Purpose of Nominally Democratic Institutions in Dictatorships 116 4.6 Appendix: Proofs 117 part ii the problem of authoritarian control 5 Moral Hazard in Authoritarian Repression and the Origins of Military Dictatorships 123 5.1 The Moral Hazard Problem in Authoritarian Repression 127 5.2 Bargaining in the Shadow of Military Intervention 134 5.3 A Formal Model 138 5.4 Empirical Analysis 148 5.5 Conclusion: The Political Price of Authoritarian Repression 159 5.6 Appendix I: Proofs 159 5.7 Appendix II: Multiple Imputation 160 6 Why Authoritarian Parties? The Regime Party as an Instrument of Co-optation and Control 162 6.1 The Logic of Party-Based Authoritarian Co-optation 167 6.1.1 Hierarchical Assignment of Service and Benefits 168 6.1.2 Political Control over Appointments 178 6.1.3 Selective Recruitment, Promotion, and Repression 182 6.2 Regime Parties and Authoritarian Resilience 184 6.3 Conclusion: Why Authoritarian Regime Parties? 192 7 Conclusion: Incentives and Institutions in Authoritarian Politics 196 7.1 Why Dictators Preside over Policy Disasters 197 Figures 1.1 Nonconstitutional exits from office of authoritarian leaders, 1946–2008. page 5 2.1 The number and the change in the number of dictatorships, 1946–2008. 25 2.2 Dictatorship around the world, 1946–2008. 27 2.3 Military involvement in authoritarian politics, 1946–2008. 34 2.4 Restrictions on political parties in dictatorships, 1946–2008. 35 2.5 Joint distribution of restrictions on political parties and military involvement in authoritarian politics, 1946–2008. 36 2.6 Legislative selection in dictatorships, 1946–2008. 36 2.7 Executive selection in dictatorships, 1946–2008. 37 2.8 Constitutional exits from office of authoritarian leaders, 1946–2008. 40 3.1 Authoritarian power-sharing game in extensive form. 65 3.2 The probability that the dictator reneges, α∗ (dashed line), and the probability that the ruling coalition rebels if it observes a high signal, β∗H (solid line), in an equilibrium of the multiperiod authoritarian power-sharing game. 72 3.3 Probability density of time-to-rebellion based on the numerical example. 75 3.4 Probability density of time-to–established autocracy based on the numerical example. 76 3.5 The empirical density of coups d’état, 1946–2008. 76 3.6 The improving odds of dying in bed, 1946–2008. 77 4.1 A model of an authoritarian polity. 101 4.2 Payoffs to ally i , given the proportion of allies that rebel φ. 103 4.3 Effect of the dictator’s power vis-à-vis the allies λ on threshold discount factors δ I (solid line) and δ∼I (dashed line) for λ < κ0. The dotted line plots a hypothetical discount factor δ = 0.85. 110 xiii xiv Figures 4.4 Nonconstitutional leader exits from office in dictatorships with and without legislatures, 1946–2008. 112 5.1 The effect of the expected magnitude of the mass threat R̂ on the equilibrium choice of the military’s resources r∗. 146 5.2 The effect of the magnitude of the mass threat R̂ on the equilibrium probability of a successful military intervention. 147 5.3 Average annual frequencies of military interventions in leader entry or exit, 1946–2002. 151 5.4 The estimated effect of economic inequality on the probability of military intervention. 156 6.1 The share of legislative seats controlled by authoritarian regime parties in dictatorships with multiple parties, 1946–2008. 188 6.2 The association between legislative seat share and the age of authoritarian ruling coalitions, 1946–2008. 191 Tables 1.1 An Outline of the Outcomes Explained in This Book page 18 2.1 The Origin and End of Authoritarian Spells, 1946–2008 26 2.2 Restrictions on Political Parties and Military Involvement in Authoritarian Politics, 1946–2008 32 2.3 Legislative and Executive Selection in Dictatorships, 1946–2008 33 2.4 Joint Distribution of Legislative and Executive Selection, 1946–2008 38 2.5 Political Affiliation of Authoritarian Leaders, 1946–2008 42 4.1 Institutions, Balance of Power, and the Success of Authoritarian Power-Sharing 99 4.2 Legislatures and the Survival of Authoritarian Ruling Coalitions, 1946–2008 111 4.3 Legislatures, Parties, and the Survival of Authoritarian Leaders, 1946–2008 114 5.1 Moral Hazard in Authoritarian Repression and Military Intervention in Politics 138 5.2 Military Intervention in Dictatorships by Level of Economic Inequality 152 5.3 The Impact of Economic Inequality on Military Intervention in Dictatorships 155 6.1 A Comparison of Co-optation Via Transfers and Co-optation Via a Regime Party 171 6.2 Restrictions on Political Parties and the Survival of Authoritarian Ruling Coalitions, 1946–2008 186 6.3 Do the Survival Functions of Authoritarian Ruling Coalitions Differ Depending on Their Restrictions on Political Parties? 187 6.4 Three Subgroups of Authoritarian Regime Parties by Legislative Seat Share 189 6.5 The Effect of a Regime Party’s Legislative Seat Share on the Survival of Authoritarian Ruling Coalitions 190 xv xviii Acknowledgments Center for the Study of Democratic Politics at Princeton University and sta- tioned at the Cline Center for Democracy at the University of Illinois. I would like to thank Larry Bartels and Pete Nardulli for their support during these periods. I have benefited greatly from the generosity and insights of my colleagues at the University of Illinois. Jake Bowers, Xinyuan Dai, Paul Diehl, Tiberiu Dragu, Zach Elkins, Sam Frost, Brian Gaines, Jude Hays, Jim Kuklinski, Bob Pahre, Gisela Sin, Bonnie Weir, and Matt Winters have offered comments and criticisms at various stages of this book. My thanks go especially to José Cheibub, whose encouragement and guidance make this project seem so easy in retrospect. I was fortunate to receive excellent and detailed comments on the entire manuscript from Giacomo Chiozza, Lucan Way, and several anonymous refer- ees. Carles Boix, José Cheibub, Jennifer Gandhi, Monika Nalepa, Pete Nardulli, and Duncan Snidal offered invaluable guidance on the book’s publication. I am also grateful to Hein Goemans, whose data were incredibly helpful at the early stages of this project. Seden Akcinaroglu, Svitlana Chernykh, Aya Kachi, Donksuk Kim, Dan Koev, Alex Sapone, Tatiana Švolı́ková, and Nini Zhang all provided valuable research assistance at various stages of this project. I am especially indebted to Michael Martin for his outstanding help with data collection, editing, and indexing. Students in my 2010–2012 undergraduate and graduate classes on the politics of dictatorships suffered through the early drafts of several chapters – their feedback was instrumental in helping me frame the book’s overarching argument. I would also like to thank my editor at Cambridge University Press, Lew Bateman, for his interest in the project and his consideration of the professional pressures faced by a junior political scientist, as well as Margaret Levi for including the manuscript in the Cambridge Studies in Comparative Politics series. Finally, I thank my family and friends for their support along the way. My wife Bonnie has been a source of endless love and energy at every stage of the writing process. She patiently read and reread, edited and re-edited the entire manuscript. Exactly when it counted, she has been my toughest critic and my greatest supporter. This book is dedicated to my parents, to whom I owe the most. Portions of this book rely on research that has been published or draws on collaborative work. Parts of Chapter 3 have appeared as “Power-Sharing and Leadership Dynamics in Authoritarian Regimes” in the American Journal of Political Science (2009). More importantly, I have had the good fortune to col- laborate with and learn from Carles Boix; parts of Chapter 4 draw on our joint paper “The Foundations of Limited Authoritarian Government: Institutions and Power-Sharing in Dictatorships.” 1 Introduction The Anatomy of Dictatorship Still democracy appears to be safer and less liable to revolution than oligarchy. For in oligarchies there is the double danger of the oligarchs falling out among themselves and also with the people . . . Aristotle, The Politics, Book 5 [W]herein men live without other security, than what their own strength, and their own invention shall furnish them . . . , the life of man [is] solitary, poor, nasty, brutish, and short. Thomas Hobbes, Leviathan Bashar al-Asad was not meant to be a dictator. Although he was the son of Syria’s long-serving president, Hafez al-Asad, Bashar’s education and career were nonpolitical. In 1988, at the age of twenty-three, he received a degree in ophthalmology from the University of Damascus and moved to London four years later to continue his medical residency. Hafez al-Asad had instead groomed Bashar’s older brother, Basil, as his successor. Yet Bashar’s seclusion from politics ended in 1994 when Basil died in an automobile accident. Bashar was recalled from London, entered a military academy, and quickly advanced through the ranks, while his father spent the last years of his life eliminating potential challengers to Bashar’s succession.1 Consider Bashar al-Asad’s delicate position on July 17, 2000, when he became the Syrian president. Given his unexpected path to power, how does he best ensure his survival in office? What threats should he expect and how will he deal with them? Alas, the contemporary political scientist is not well equipped to become the new Machiavelli. If Bashar al-Asad were concerned about politically succeeding in a democracy, students of politics might offer him suggestions ranging from how to best target voters in campaigns to the implications of electoral systems 1 See Hinnebusch (2002), Leverett (2005), and Perthes (2006). 1 2 The Politics of Authoritarian Rule for partisan competition.2 But of course, if Bashar al-Asad lived in a democracy, he would not have been in a position to inherit a presidency. Although growing at a fast pace, contemporary scholarship on dictatorships has so far generated only a fragmented understanding of authoritarian politics. Extant research increasingly studies authoritarian parties, legislatures, bureau- cracies, and elections, as well as repression, leadership change, and regime stability across dictatorships.3 Yet in most cases, these facets of authoritarian- ism are examined individually, in isolation. In turn, we lack a unified theoretical framework that would help us to identify key actors in dictatorships; locate the sources of political conflict among them; and thereby explain the enor- mous variation in institutions, leaders, and policies across dictatorships.4 At both the empirical and theoretical level, we are without a general conceptual heuristic that would facilitate comparisons across polities as diverse as Mexico under the Institutional Revolutionary Party (PRI), Saddam Hussein’s Iraq, and contemporary China. This book attempts to fill that void. I argue that two conflicts fundamentally shape authoritarian politics. The first is between those who rule and those who are ruled. All dictators face threats from the masses, and I call the political problem of balancing against the majority excluded from power the problem of authoritarian control. Yet dictators rarely control enough resources to preclude such challenges on their own – they therefore typically rule with a number of allies, whether they be traditional elites, prominent party members, or generals in charge of repression. A second, separate political conflict arises when dictators counter challenges from those with whom they share power. This is the problem of authoritarian power-sharing. To paraphrase Aristotle’s warning in this chapter’s epigraph, authoritarian elites may fall out both with the people and among themselves. Crucially, whether and how dictators resolve the problems of power-sharing and control is shaped by two distinctively dismal features of authoritarian pol- itics. First, dictatorships inherently lack an independent authority with the power to enforce agreements among key political actors, especially the dicta- tor, his allies, and their repressive agents. Second, violence is an ever-present and ultimate arbiter of conflicts in authoritarian politics. These two intrinsic features uniquely shape the conduct of politics in dictatorships. They limit the role that political institutions can plausibly play in resolving the problems of power-sharing and control, and they explain the gruesome manner in which so many dictators and dictatorships fall. Authoritarian politics takes place in the shadow of betrayal and violence. In brief, the central claim of this book is this: Key features of authoritari- anism – including institutions, policies, as well as the survival of leaders and regimes – are shaped by the twin problems of power-sharing and control against 2 See, e.g., Green and Gerber (2004) and Cox (1997), respectively. 3 See subsequent chapters for a detailed discussion of this literature. 4 Bueno de Mesquita et al. (2003) and Wintrobe (1998) are two notable exceptions to the tendency for fragmentary explanations of authoritarian politics. Introduction 5 68% 11% 10% 7% 5% 0 50 100 150 200 Foreign intervention Assassination Transition to democracy Popular uprising Coup d’état At least one day in office At least one year in office figure 1.1. Nonconstitutional exits from office of authoritarian leaders, 1946–2008. Note: Percentages refer to a category’s share of all nonconstitutional exists. Exits of interim leaders are not included. Unambiguous determination of exit was not possible for thirteen leaders. short-lived leaders may have been more vulnerable because of their inexperience in office or a weaker hold on power.9 Thus as far as authoritarian leadership dynamics are concerned, an over- whelming majority of dictators lose power to those inside the gates of the presidential palace rather than to the masses outside. The predominant polit- ical conflict in dictatorships appears to be not between the ruling elite and the masses but rather one among regime insiders. This is the second of the two problems of authoritarian rule that I identify: the problem of authoritar- ian power-sharing. The evidence I just reviewed suggests that to understand the politics of dictatorships, we must examine why and how a conflict among authoritarian elites undermines their ability to govern.10 I undertake this task in Part I of this book. 1.1.1 The Problem of Authoritarian Power-Sharing When he assumed office, Bashar al-Asad – like most dictators – did not per- sonally control enough resources to govern alone. Toward the end of his life, Bashar’s father Hafez al-Asad assembled a coalition of old comrades-in-arms, business elites, and Baath Party officials who would support his son’s succession to the Syrian presidency.11 This is what I call a ruling coalition – a set of 9 I elaborate on the latter rationale in Chapter 3. 10 Various aspects of such conflicts among authoritarian elites have been studied by Ramsayer and Rosenbluth (1995), Geddes (1999a), Bueno de Mesquita et al. (2003), Brownlee (2007a), Gehlbach and Keefer (2008), Magaloni (2008), Myerson (2008), and Guriev and Sonin (2009). 11 See Leverett (2005) and Perthes (2006). 6 The Politics of Authoritarian Rule individuals who support a dictator and, jointly with him, hold enough power to guarantee a regime’s survival. This terminology is inspired by its semantic counterpart in Soviet politics: Stalin’s inner circle came to be known as the “select group,” the “close circle,” or – most commonly – the “ruling group.”12 Chapters 3 and 4 explain why power-sharing between a dictator and his ruling coalition so frequently fails. A key obstacle to successful authoritarian power-sharing is the dictator’s desire and opportunity to acquire more power at the expense of his allies. In dictatorships, the only effective deterrent against such opportunism is the allies’ threat to replace the dictator. Throughout this book, I refer to such elite-driven attempts to remove an authoritarian leader as allies’ rebellions, mirroring the language of the right to a “baronial rebellion” recognized by the Magna Carta of 1215. Of course, the closest empirical coun- terpart of such rebellions are the coups d’état that I just discussed. Quite often though, leaders of successful rebellions characterize them in a language that is more suggestive of their righteous motives – as in the case of the Corrective Revolution of 1970 that brought Hafez al-Asad’s faction of the Baath Party to power in Syria. Chapter 3 examines the most blatant failure of authoritarian power-sharing: the emergence of personal autocracy. I explain why a power trajectory along which an authoritarian leader, like Joseph Stalin, assumes office as the “first among equals” but succeeds over time in accumulating enough power to become an invincible autocrat is both possible and unlikely. The possibility of such “upward mobility” is intimately tied to the distinctively toxic condi- tions under which authoritarian elites must operate. When they cannot rely on an independent authority to compel the dictator to share power as agreed and when violence looms in the background, a small dose of uncertainty about a rebellion’s success will limit the allies’ ability to credibly deter the dictator from attempting to usurp power at their expense. If he succeeds in several such attempts, the dictator may accumulate enough power to entirely undermine the allies’ capacity to stop him. Hence the emergence of personal autocracy should be a rare but nevertheless systematic phenomenon across dictatorships. This logic implies that the interaction between a dictator and his allies gen- erally takes only two politically distinct forms. Under the first, which I call contested autocracy, politics is one of balancing between the dictator and the allies – the allies are capable of using the threat of a rebellion to check the dic- tator’s opportunism, albeit imperfectly. By contrast, established autocrats have acquired so much power that they can no longer be credibly threatened by their allies – they have effectively monopolized power. In fact, many accounts by classical philosophers and historians identify precisely this analytical distinc- tion: Machiavelli distinguishes between the King of France, who cannot take away the privileges of his barons “without endangering himself,” and the Turk, whose ministers are his “slaves.” Meanwhile, historians of the Soviet Union distinguish between the pre–Purges and the post–Purges Stalin that achieved 12 The corresponding Russian terms are uzkii sostav, blizhnii krug, and rukovodiashchaia grupa, respectively. See Gorlizki and Khlevniuk (2004, 47). Introduction 7 “limitless power over the fate of every Soviet official”; and historians of China distinguish between the pre–1958 Mao, who “listened to interests within the system,” and the “later Mao,” who simply overrode them.13 Hence the tran- sition from contested to established autocracy represents the degeneration of authoritarian power-sharing into personal autocracy. Chapter 3 thus explains the emergence of a prominent class of dictator- ships that have been alternatively referred to as personalist, neopatrimonial, or sultanistic.14 In these regimes, leaders have managed to wrestle power away from the individuals and institutions that originally brought them to power – whether they be parties, militaries, or dynastic families. My arguments clarify why such dictators – like Fidel Castro, who ruled Cuba for a half-century until his retirement in 2008 – emerge across all kinds of dictatorships, develop personality cults, and enjoy long tenures: They have effectively eliminated any threats from their ruling coalition. This last point helps us understand not only the variation in the length of dictators’ tenures but also the manner by which they lose office. When established autocrats ultimately leave office, it is most likely by a process that is unrelated to the interaction with their allies. Accord- ingly, Saddam Hussein was brought down by a foreign occupier, Muammar Qaddafi by a popular uprising, and Joseph Stalin by a stroke – none of them at the hands of their inner circle. My emphasis shifts from the failure of authoritarian power-sharing to its potential success in Chapter 4. One factor that exacerbates the gruesome char- acter of dictatorships is the secrecy that typically pervades interactions among authoritarian elites. Yet unlike the potential for violence or the lack of an inde- pendent authority that would enforce agreements among the dictator and his allies, the lack of transparency among authoritarian elites might be curtailed, if not eliminated, by adopting appropriate political institutions. These most often take the form of high-level, deliberative, and decision-making bodies – committees, politburos, or ruling councils – and are usually embedded within authoritarian parties and legislatures.15 Formal political institutions alleviate monitoring problems in authoritar- ian power-sharing in two distinct ways. Institutions like the Politburo Standing Committee of the Communist Party of China (1949–present), the Chilean Junta Militar de Gobierno under Pinochet (1973–1990), and the Consultative Coun- cil of Saudi Arabia (1993–present) typically establish formal rules concerning membership, jurisdiction, protocol, and decision making that both facilitate the exchange of information among the ruling elites and provide for an easy assessment of compliance with those rules.16 Thus regular, institutionalized 13 See Machiavelli (2005[1513], 16–17), Khlevniuk (2009, 247), and Teiwes (2001, 79). 14 On these concepts, see Zolberg (1966), Roth (1968), Jackson and Rosberg (1982), Snyder (1992), Bratton and Van de Walle (1997), Linz and Chehabi (1998), Geddes (1999a), and Brownlee (2002). 15 On authoritarian parties, see Brownlee (2007a), Geddes (2008), Gehlbach and Keefer (2008), Greene (2007), Magaloni (2006), and Smith (2005); on legislatures, see Gandhi and Przeworski (2007), Gandhi (2008), Malesky (2009), Ramseyer and Rosenbluth (1995), and Wright (2008a). 16 See Barros (2002), MacFarquhar (1997a), and Herb (1999) on these institutions in Chile, China, and Saudi Arabia, respectively. 10 The Politics of Authoritarian Rule At first glance, the difference between repression and co-optation may seem to be simply one between negative and positive incentives for compliance with the regime – “sticks and carrots” in popular parlance. Repression, however, is much more than co-optation’s evil twin. When we examine the two in isolation or treat them as substitutes, we may overlook that differences in their use have far-reaching consequences for the political organization and vulnerabilities of dictatorships. Heavy reliance on repression – typically by the military – entails a fundamen- tal moral hazard: The very resources that enable a regime’s repressive agents to suppress its opposition also empower it to act against the regime itself. Hence once soldiers become indispensable for a regime’s survival, they acquire polit- ical leverage that they can exploit. Militaries frequently do so by demanding privileges, perks, and policy concessions that go beyond what is necessary for suppressing the regime’s opposition – they claim a seat at the table when the spoils of their complicity are divided. As Machiavelli warns in The Prince, those emperors who come to power by “corrupting the soldiers” become hostages of “him who granted them the state.”22 This is why the former Tunisian President Zine El Abidine Ben Ali kept his military small and underequipped; why the Iraqi Baath regime disposed of its uniformed accomplices immediately after it came to power in 1968; and why Mao Zedong insisted that the Party must always command the gun. Nevertheless, no dictatorship can do away with repression. The lack of popular consent – inherent in any political system where a few govern over the many – is the “original sin” of dictatorships. In fact, many dictators do not have much leeway when deciding how much to rely on soldiers for repression. In regimes that face mass, organized, and potentially violent opposition, the military is the only force capable of defeating such threats. For dictators in these circumstances, political dependence on soldiers may be insurmountable. Meanwhile, other dictators simply inherit politically entrenched militaries when they come to power. These regimes, in turn, must concede to soldiers greater resources, institutional autonomy, and influence over policy. This is why the Egyptian military presides over a complex of commercial enterprises (Cook 2007, 19); why the Honduran military won complete autonomy over its budget and leadership positions after it brought President Ramón Villeda Morales to power in 1954 (Bowman 2002, Chap. 5); and why, in 1973, the Uruguayan military had its political influence institutionalized in a National Security Council that assisted several docile presidents in “carrying out national objectives” (Rouquié 1987, 251). Chapter 5 explains why bargaining over such concessions between a govern- ment and politically entrenched militaries takes a peculiar form: Each side con- sciously manipulates the risk of actual military intervention, even though both would prefer to avoid it. Military dictatorships emerge when, in the process 22 Chap. VII, “Concerning New Principalities Which Are Acquired Either by the Arms of Others or by Good Fortune,” in Machiavelli (2005[1513]). Introduction 11 of such brinkmanship, either the military or the government “rocks the boat” too much.23 Authoritarian reliance on repression is thus a double-edged sword: It sows the seeds of future military interventions. The analysis in Chapter 5 in turn clarifies why so many dictators wear a military uniform. Political control over militaries – in both dictatorship and democracies – is a political problem before it is a cultural or institutional one. When deciding how much to rely on repression, dictators make a trade-off between their exposure to external threats from the masses and their vulnera- bility to internal threats from their repressive agents. In dictatorships where a few in power control a disproportionate share of wealth, repression is simply more attractive than co-optation. In these regimes, it is cheaper for the regime to pay its repressive agents to suppress any opposition than to assuage it by co-optation – even after accounting for the Faustian bargain that such reliance on repression entails. In turn, we should see more sticks than carrots in coun- tries where a few wealthy landowners control the economy, where command of the government amounts to ownership of the country’s natural resources, and where a minority excludes a majority from power on ethnic or sectar- ian grounds. Such polity-wide, structural factors explain why some dictators maintain perfect political control over their militaries, why others are under effective military tutelage, and why military interventions threaten many new democracies. My focus shifts attention shifts from sticks to carrots in Chapter 6, which examines why some dictatorships establish and maintain a regime-sanctioned political party. Many authoritarian regimes favor one or several political par- ties, but only some – like PRI–era Mexico, Saddam Hussein’s Iraq, and con- temporary China – establish a party structure that effectively maintains a loyal, popular base for the regime. Chapter 6 identifies three core institutional fea- tures that turn authoritarian parties into effective instruments of authoritarian control: (1) hierarchical assignment of service and benefits, (2) political control over appointments, and (3) selective recruitment and repression. Briefly, the first feature entails assigning costly, politically valuable party service – often in the form of ideological proselytizing, intelligence gathering, and mobilization for regime-sanctioned events – early in a party member’s career while delaying the benefits of party membership – which typically entail better employment and promotion prospects or privileged access to education and social services – to 23 Existing research shows that military dictatorships are systematically associated with a range of outcomes. Geddes (1999b) and Hadenius and Teorell (2007) show that when compared to single-party and personalist dictatorships, military dictatorships are the most common form of authoritarian government prior to the 1990s, yet they also have the shortest lifespan (Geddes 1999b; Brownlee 2009); leaders of military dictatorships are less likely to survive in office than leaders of nonmilitary dictatorships (Geddes 1999b; Gandhi 2008) and they tend to be deposed by further coups (Nordlinger 1977; Debs 2009); and military regimes are also more resilient than personalist regimes or monarchies to international sanctions (Escribà-Folch and Wright 2008) and they also are more likely than single-party regimes to initiate military disputes (Lai and Slater 2006). 12 The Politics of Authoritarian Rule a later point. As a result, by the time party members reap the benefits of senior- ity, their costly service becomes “sunk investment”: Once expended, it cannot be recovered or transferred across political coalitions. These organizational features of authoritarian parties therefore accomplish more than simply distribute rewards in exchange for party members’ loyalty to the regime, as the extant literature frequently concludes. That could be accomplished without the institution of a party. After all, dictators frequently assuage popular discontent by redistributing land, subsidizing basic goods, or even distributing cash – as the Bahraini king did in the wake of the Arab Spring when he promised each family the equivalent of more than two thousand U.S. dollars.24 Rather, these features of internal party organization effectively exploit natural career aspirations among the population in order to foster an enduring stake in the perpetuation of the regime among its most productive and ideologically agreeable segments. As Bratton and Van de Walle (1997, 86) put it in their study of African transitions to democracy, members of such parties have little option but “to sink or swim” with the regime. Chapter 6 thus clarifies why authoritarian parties are best thought of as incentive structures that encourage sunk political investment by their members; why they serve to marginalize opposition rather than to co-opt it; and why party dictatorships with these organizational features survive under less favorable circumstances than dictatorships without them, even if the latter expand the same resources on co-optation. I also explain why dictatorships need the actual institution of the party; why some dictatorships find co-optation via parties less attractive than the alternatives of repression or co-optation by social spending alone; and why former authoritarian party elites so frequently continue to hold a firm grip over the politics of nascent democracies. This discussion outlines the first step in the overarching theoretical argu- ment that I develop in this book: In dictatorships, political battle lines emerge as often among those in power as they do between the elite and the masses. I identify these two distinct conflicts as the problems of authoritarian power- sharing and control. When I previously presented the two conflicts separately, it was primarily for analytical clarity and the heuristic value of such expo- sitional separation. As my discussion of repression and co-optation implies, the two problems are often interconnected: When indispensable in repression, soldiers transform from obedient agents into political rivals who demand a cut from the spoils of their complicity. Meanwhile, in order to co-opt effectively, authoritarian parties promise upward mobility that over time begets a new political elite. Repression and co-optation thus each empower different actors and institutions. Dictators’ response to the problem of authoritarian control therefore shapes the likely contours of the conflict over power-sharing. Jointly, the two problems clarify why many nominally democratic institu- tions – especially legislatures, parties, and even some elections – serve dis- tinctively authoritarian ends: They help dictators resolve the problems of power-sharing and control. Whereas legislatures serve to represent the diversity 24 See “Bahrain’s King Gives out Cash Ahead of Protests,” Reuters, 12 February 2011. Introduction 15 A related concern emerges in the context of authoritarian control. Dictators are wary about relying on their militaries for repression with good reason. When indispensable for a regime’s survival, repressive forces metamorphosize from an obedient servant into a potential political rival – regardless of any formal constraints on their prerogatives. This is what General Idi Amin Dada did in Uganda after he became indispensable in Milton Obote’s suppression of opposition to his eventual consolidation of dictatorial powers. Beginning in 1965, Obote used Amin’s loyal following within the armed forces to elim- inate opposition, first in the parliament, then from the country’s ceremonial president, and ultimately from within his own party. By the time Obote estab- lished a full-fledged dictatorship, he needed Amin and his army more than Amin needed Obote (Mutibwa 1992, 64). In 1971, Idi Amin deposed Obote in a military coup d’état and established what would become one of the most brutal dictatorships of the twentieth century. In authoritarian politics, therefore, no independent third party can be real- istically expected to enforce commitments among key actors – whether it be the dictator’s promise to share power with his allies, the repressive agents’ pledge to obediently serve their masters, or the dictator’s allies’ agreement to collectively replace him in a rebellion if he attempts to usurp power. This concern is compounded by the looming possibility of resolving political conflicts with violence. In authoritarian politics, the option of violence is never off the table: Political conflicts may be, and indeed frequently are, resolved by brute force. For every peaceful, negotiated, or institutional resolution of a politi- cal conflict, there is a crude alternative in which brute force plays a decisive role. The expulsion of the air force representative and Pinochet’s chief critic Gustavo Leigh from the Chilean junta in 1978 proceeded by a show of force: the occupa- tion of air force headquarters and installations by the army, in violation of the decree laws that were supposed to regulate decision making within the junta.29 Under dictatorship, therefore, institutionalized “rules of the game” cannot be taken at face value. But this does not amount to saying that institutions are epiphenomenal – that they merely mirror the power relations among the dictator, his allies, the regime’s repressive agents, and the masses excluded from power. Institutions do have the capacity to prevent unnecessary, regime- destabilizing conflicts in authoritarian politics, but only when institutionalized “rules of the game” rest on mutual advantage and respect the power of key participants. Put in the jargon of modern political science, authoritarian insti- tutions must be self-enforcing.30 Although settings in which actors cannot take their agreements to be binding and may resolve conflicts violently can, in principle, be analyzed in natural 29 These were Decree Laws 527 and 991; see Barros (2002, Chap. 2). 30 Of course, the requirement that political outcomes be self-enforcing underlies most modern explanations of institutional choices in any political regime as well as transitions between regimes, as in the literature on “self-enforcing democracy” (Przeworski 1991, 2011; Boix 2003; Acemoglu and Robinson 2005; Fearon 2008). But unlike in the study of authoritarian politics, concerns about defection from key constitutional provisions can be safely assumed away in the study of democratic politics. 16 The Politics of Authoritarian Rule language, the conceptual issues involved in their analysis have been prominently articulated and rigorously examined in social-scientific applications of game theory.31 Accordingly, I draw on these techniques and develop new formal models of authoritarian power-sharing, institutional choice, repression, and co-optation in Chapters 3 through 6. While technical exposition tends to seem inviting to “authorized personnel only,” I hope this downside is outweighed by what I see as a commitment to the dictum “Trust but verify.” By formalizing my arguments, I can be more explicit about my assumptions, more transparent in my reasoning, and more specific about the empirical implications of my arguments than if I developed and presented them only verbally.32 The two distinctive aspects of authoritarian politics – the absence of an inde- pendent authority that would enforce mutual agreements and the ever-present potential for violence – also highlight why the nature of politics fundamen- tally differs between dictatorships and democracies. By definition, we consider a country to be democratic only if it resolves political conflicts nonviolently, typically by elections, legislative votes, and cabinet decisions. Furthermore, a country ceases to be a democracy the moment a few key mechanisms – espe- cially electoral rules and the respect of certain liberties – are circumvented, even if nonviolently. Thus when Cox (1997) examines how electoral rules shape vot- ers’ behavior or when Laver and Schofield (1990) study the politics of coalition governments, they can safely assume away any concerns about whether gov- ernments, parties, or voters will actually comply with constitutional provisions or the outcomes of elections. By definition, a failure to do so would turn a democracy into a dictatorship. Students of authoritarian politics cannot make such convenient assump- tions.33 While frequent, backstabbing is only metaphorical in democracies. In dictatorships, it is literal: According to the data described earlier, about one- third of leadership changes in dictatorships involve overt violence and about two-thirds of them are nonconstitutional – they depart from official proce- dures or established conventions. While not all dictatorships resolve political conflicts violently all of the time, and formal rules appear to constrain some dictators at least some of the time, this may be precisely because the option of violence looms in the background, thereby precluding the need to carry it out and enforce compliance with institutional rules. To paraphrase Thomas Hobbes’s famous line in this chapter’s epigraph: Because authoritarian elites live without other security than their own strength, their life is nasty, brutish, and often short. 31 For an introduction to game theory and formal political theory, see, e.g., McCarty and Meirowitz (2007), Morrow (1994), Myerson (1991), and Osborne (2004). 32 For a discussion of the value and limits of game-theoretic analysis in the social sciences, see Aumann (1985), Bates et al. (1998), Geddes (2003, Chap. 5), Kreps (1990), Morton (1999), Myerson (1992, 1999), Powell (1999, Chap. 1), Rubinstein (1991), and Tsebelis (1990, Chap. 2). 33 A similar point applies to the study of regime change; see Acemoglu and Robinson (2001, 2005), Boix (2003), and Przeworski (1991, 2005, 2011). Introduction 17 This sharp conceptual dichotomy between authoritarian and democratic politics guides how I collect and organize data on dictatorships. Chapter 2 defines a dictatorship to be a country that fails to elect its legislature and exec- utive in free and competitive elections. Empirically, then, I follow Alvarez et al. (1996) and think about the differences between dictatorships and democracies as first of all in kind and only then in degree.34 Meanwhile, the questionable relevance of political institutions under dictatorship leads me to complement data on formal institutions by other, more credible measures of their binding power. I therefore use original, detailed data on the timing and manner of entry into and exit from office for all authoritarian leaders throughout the period 1946–2008. Chapter 2 outlines how I organize the extraordinary diversity in institutions and leadership transitions observed across dictatorships. This diversity obtains partly because dictatorship is a residual category that contains all countries that do not meet established criteria for democracy and partly because of dic- tatorship’s richer and longer pedigree. I also argue that in our attempts to organize authoritarian politics, we should abandon the prevailing practice of classifying dictatorships into a few ideal types or according to their prominent descriptive features. That approach is flawed for several reasons: It collapses multiple and distinct conceptual dimensions of authoritarian politics into a single typology; it results in categories that are neither mutually exclusive nor collectively exhaustive; and it requires difficult classification judgments that weigh incommensurable aspects of authoritarian politics. These flaws compro- mise the validity and reliability of empirical inferences based on such data. I propose an alternative approach, one that explicitly identifies the concep- tual dimensions of authoritarian politics being measured and then develops appropriate scales for each dimension. 1.3 plan of the book The remainder of this book begins with Chapter 2, in which I define what I mean by dictatorship and organize the extraordinary heterogeneity in insti- tutions and leaders in authoritarian politics. I clarify why I essentially follow Alvarez et al.’s (1996) procedural and minimalist approach to the classification of regime types, and I illustrate the flaws of existing approaches to the clas- sification of dictatorships with a discussion of Geddes’s (1999b) typology of dictatorships. I also present the data used throughout this book, which cover the period 1946–2008 and are situated at three levels of observation: the country level, the ruling-coalition level, and the leader level. At the country level, I measure four dimensions of the political organization of dictatorships: military involvement in politics, restrictions on political parties, legislative selection, and executive selection. I also introduce a new measure of authoritarian stability, which I 34 I am paraphrasing Elkins’s (2000, 293) restatement of the position of Alvarez et al. (1996). 20 The Politics of Authoritarian Rule dictatorship is a residual category that contains all countries that do not meet established criteria for democracy. At a minimum, such criteria require that free, fair, and competitive elections determine the composition of the legis- lature and – often indirectly – the executive. More demanding criteria may require that governments respect certain civil liberties – such as the freedom of religion (Schmitter and Karl 1991; Zakaria 1997) – or that the incumbent government and the opposition alternate in power at least once after the first seemingly free election (Huntington 1993; Przeworski et al. 2000; Cheibub et al. 2010). Failure to satisfy any of these criteria may brand a country as a dictatorship. The understandable result is an exceptionally diverse set of polities, unified perhaps only by the failure to meet one or more criteria for democracy. To paraphrase Tolstoy, whereas democracies are all alike, each dictatorship may be undemocratic in its own way.3 The second challenge in organizing the world of authoritarian politics stems from the questionable relevance of formal institutions under dictatorship. As discussed in Chapter 1, authoritarian politics takes place under distinctively dismal conditions. Authoritarian elites cannot rely on an independent author- ity to enforce their agreements, and violence is the ultimate arbiter of political conflicts. Hence whether and which institutions actually matter for the conduct of authoritarian politics is far from apparent and neither is who matters. As the epigraph at the beginning of this chapter about Plutarco Calles’s continuing influence after he resigned from the Mexican presidency highlights, under dic- tatorship, the man who gives orders may not reside in the presidential palace but rather across the street from it.4 Before moving forward with the substantive arguments in this book, there- fore, it is essential to precisely define what I mean by dictatorship and to organize the extraordinary heterogeneity in institutions and leaders across dic- tatorships. As I clarify in this chapter, I essentially follow Alvarez et al.’s (1996) procedural and minimalist approach to the classification of regime types. After excluding any periods of foreign occupation, the collapse of state authority, or civil war, I say that a country is a dictatorship if it fails to elect its legislature and executive in free and competitive elections. This definition implies a sharp dichotomy between authoritarian and democratic politics. The arguments in this chapter clarify why the difference between dictatorship and democracy is best thought of first as one of kind and only then as one of degree. The data used throughout this book are situated at three levels of observa- tion: the country level, the ruling-coalition level, and the leader level. In devising 3 Collier and Levitsky (1997) discuss how the creation of diminished subtypes of democracy provides better differentiation among the various ways in which a polity may fail to satisfy the criteria for democracy. Przeworski et al. (2000) and Cheibub et al. (2010) defend the minimalist conception of democracy adopted herein. 4 Plutarco Elı́as Calles stepped down from the Mexican presidency in 1928 but nevertheless overshadowed his three successors; see Krauze (1997, 430). The World of Authoritarian Politics 21 the categories that organize these data, I attempt to overcome several impor- tant limitations of existing approaches to classifying dictatorships. Most exist- ing typologies fail to recognize that they implicitly collapse multiple, distinct conceptual dimensions of authoritarian politics into a single typology. This is a direct consequence of the prevailing practice of classifying dictatorships into a few ideal types, as in the case of a “totalitarian” dictatorship, or according to their prominent descriptive features, as in the case of a “bureaucratic authori- tarian” regime. The resulting categories are typically neither mutually exclusive nor collectively exhaustive, and they require difficult classification judgments that weigh incommensurable aspects of authoritarian politics, thereby compro- mising the validity and reliability of empirical inferences based on them. The prevailing usage of Geddes’s typology of personalist, military, and single-party dictatorships, which I discuss in this chapter, illustrates these shortcomings. Rather than classifying dictatorships into ideal types or according to their prominent descriptive features, we should instead explicitly identify the con- ceptual dimensions of authoritarian politics that we want to measure and then develop appropriate scales or typologies for each one. At the country level, I use four such dimensions: military involvement in politics, restrictions on political parties, legislative selection, and executive selection. Their choice is guided by the theoretical arguments developed throughout this book. As we shall see in this chapter, their joint distribution provides a more representative summary of the immense institutional heterogeneity across dictatorships than a typology based on a few ideal types or prominent descriptive features. Another major challenge in organizing authoritarian politics is the question- able relevance of formal institutions in dictatorships. Because dictatorships lack an independent authority that would enforce compliance with institutionalized “rules of the game,” the political significance of institutions that ostensibly govern authoritarian politics cannot be taken at face value. This is why I col- lected detailed data on leadership change across dictatorships. How dictators actually assume and lose power speaks louder than the rules that presumably regulate it. As in the case of authoritarian political institutions, the conceptual organi- zation of my data on authoritarian leadership change is guided by the empirical implications of theoretical arguments in subsequent chapters. For all dictators who held power for at least one day between 1946 and 2008, I record the timing and manner of their entry into and exit from office, as well as the use of violence and the participation of the military in these events. Finally, I introduce a new measure of authoritarian stability, which I refer to as a ruling-coalition spell. It consists of an uninterrupted succession in office of politically affiliated authoritarian leaders – typically from the same government, party, family, or military junta – and is based on my data on the institutional origin of authoritarian leaders and their political affiliation prior to assuming office. Some of my claims about the effect of institutions on the stability and longevity of authoritarian rule, as well as many claims in the existing literature, are evaluated most appropriately at this level of observation. Thus my claim 22 The Politics of Authoritarian Rule in Chapter 6 that regime-sanctioned political parties serve as effective tools of authoritarian co-optation does not pertain to the durability of dictatorship as a regime type or to the survival of individual dictators but rather to the continuity in power of leaders from the same political coalition – as in the case of Mexico under the PRI. By using data on ruling-coalition spells, I avoid confounding these distinct levels of analysis. All data codebooks and estimation code used throughout this book are available at a dedicated page on my Web site.5 2.1 what counts as a dictatorship? Although far from extinct, dictatorships have been declining both in number and as a proportion of all regimes since the early 1970s. As we will soon see, transitions to democracy have in spite of a few hiccups outnumbered democratic breakdowns during the last three decades.6 Especially since the end of the Cold War, surviving dictatorships have at least nominally come to resemble democracies in terms of their formal institutions (Diamond 2002; Levitsky and Way 2002; Schedler 2006). The few that defy this trend, especially the likes of North Korea or Saudi Arabia, appear to be stuck in an atavistic state, anachronistic and at odds with the rest of the world. In short, few contemporary dictatorships admit that they are just that. If we were to trust dictators’ declarations about their regimes, most of them would be democracies. According to President Aleksandr Lukashenko, present-day Belarus has had “so much so-called democracy that it has made [Belorus- sians] nauseated.”7 Even more often, contemporary dictatorships would be an improvement on democracy: Muammar Qaddafi’s Libyan Jamahiriya was a committee-governed “direct democracy,” solving the contradictions inherent in capitalism and communism (Mattes 2008, 59); Vladimir Putin’s Russia is a “sovereign democracy,” ensuring that the country is governed not by Western meddlers but rather by the Russian nation; and even China professes to be “the people’s democratic dictatorship.”8 It is therefore essential to explicitly state how we recognize a dictatorship when we see it. I follow Przeworski et al. (2000), Boix (2003), and Cheibub et al. (2010) in defining a dictatorship as an independent country that fails to satisfy at least one of the following two criteria for democracy: (1) free and competitive legislative elections and (2) an executive that is elected either directly in free and competitive presidential elections or indirectly by a legislature in parliamentary systems. Throughout the book, I use the terms dictatorship and authoritarian 5 See https://netfiles.uiuc.edu/msvolik/www/. 6 See also Merkel (2010); Diamond (2008) and Puddington (2008) have a more pessimistic view about the future of democracy. 7 On Lukashenko’s statement, see “Belarus Leader Blames Excess of Democracy for Bombing,” The New York Times, 21 April 2011. 8 On “sovereign democracy,” see Masha Lipman, “Putin’s ‘Sovereign Democracy’,” The Wash- ington Post, 15 July 2006. The World of Authoritarian Politics 25 40 50 60 70 80 P er ce nt ag e of a ll co un tr ie s − 15 0 15 40 60 80 10 0 12 0 C ha ng e N um be r 1946 1955 1964 1973 1982 1991 2000 Year Number of dictatorships (left axis) Change in the number of dictatorships Dictatorships as a percentage of all countries (right axis) figure 2.1. The number and change in the number of dictatorships, 1946–2008. In a second departure from Przeworski et al. (2000), Boix (2003), and Cheibub et al. (2010), I exclude from my data on dictatorships any periods of foreign occupation, the collapse of state authority, or a major civil war. Although such periods trivially fail to satisfy the criteria for democracy, poli- ties in such circumstances cannot be meaningfully considered to be authori- tarian. Bosnia between 1992 and 1995 was neither a democracy nor a dicta- torship – it was at civil war. Periods like this one are best characterized by the lack of any sovereign political authority.16 I identify them using the Polity IV data (Marshall and Jaggers 2008), the Correlates of War (Sarkees 2000), and the UCDP/PRIO Armed Conflict Dataset (Gleditsch et al. 2002; Themnér and Wallensteen 2011).17 The resulting data cover the period 1946–2008 and contain 4,696 authori- tarian country-years. As shown in Figure 2.1, between 41 and 114 dictatorships exist in any given year during this period.18 The number of dictatorships gener- ally grew from 1946 to the late 1970s and declined thereafter. In proportional terms, however, the share of dictatorships among all countries already began to decline at the beginning of the 1970s. As the lower part of Figure 2.1 illustrates, 16 Previous large-N research on dictatorships has ignored such periods. This risks conflating the instability that typically accompanies foreign occupations, the collapse of state authority, or civil wars with authoritarian politics. 17 This parallels the category of anarchy in Magaloni and Kricheli (2010). See the codebook for further details. 18 These numbers correspond to 1947 and 1977/1979, respectively. 26 The Politics of Authoritarian Rule table 2.1. The Origin and End of Authoritarian Spells, 1946–2008 Frequency (%) Origin End Breakdown of/transition to democracy 58 (29.15) 97 (48.74) Independence 72 (36.18) 11 (0.50) Lack of sovereign authority 34 (17.09) 38 (19.10) Continuance from before 1946 to after 2008 35 (17.58) 63 (31.66) Note: The unit of observation is an authoritarian spell. See the codebook for details about coding rules for each category. the largest increase in the number of dictatorships occurred during the decolo- nization era of the 1950s and 1960s, whereas the period following the end of the Cold War witnessed the sharpest decrease in the number of dictatorships throughout the period 1946–2008. Overall, dictatorships comprise between a minimum of 39 percent and a maximum of 75 percent of all countries between 1946 and 2008, corresponding to the years 2008 and 1972, respectively. As a graphical summary of the geographic distribution of dictatorships around the world, Figure 2.2 plots the number of authoritarian years that each country contributes to the data throughout the period 1946–2008. I refer to an uninterrupted period of dictatorship in a particular country as an authoritarian spell. The countries in the data have experienced between one and four authoritarian spells throughout the period 1946–2008. As summarized in Table 2.1, most authoritarian spells originate in newly independent countries (i.e., 36 percent) and after democratic breakdowns (i.e., 29 percent); this was the case of Cambodia in 1953 and Chile in 1973, respectively. The remaining 33 percent of authoritarian spells either began prior to 1946 (e.g., the Soviet Union) or emerged after a period of foreign occupation, collapse of state authority, or civil war (e.g., Mao’s China in 1949). On the other hand, the largest number of authoritarian spells – about one half – end in a transition to democracy. A further 19 percent end as a result of foreign occupation, the collapse of state authority, or civil war; a negligible number end after a country ceases to exist. As of 2008, there were 63 (i.e., 32 percent) surviving authoritarian spells. Brazil in 1985, Afghanistan in 2001, Yugoslavia in 1991, and Cuba in 2008 are respective examples of these cate- gories. The origin and end of all 199 authoritarian spells in the data are listed in the appendix at the end of this chapter. 2.2 making sense of institutional heterogeneity under dictatorship As is apparent in Figure 2.2, any attempt to explain authoritarian politics must confront its extraordinary scope and diversity. A major source of this diversity is that dictatorship is a residual or negative category defined in the 46 − 5 7 ye ar s (3 2 co un tr ie s) 38 − 4 5 ye ar s (3 3 co un tr ie s) 27 − 3 7 ye ar s (3 2 co un tr ie s) 2 − 2 6 ye ar s (3 4 co un tr ie s) D em oc ra cy o r N ot In de pe nd en t fi g u re 2. 2. D ic ta to rs hi p ar ou nd th e w or ld , 19 46 –2 00 8. N ot e: W or ld ge og ra ph y as of 20 02 ; no t di sp la yi ng co un tr ie s th at ce as ed to ex is t pr io r to 20 02 ; co un tr ie s th at w er e pa rt of an ot he r di ct at or sh ip pr io r to 20 02 do no t in he ri t th at di ct at or sh ip ’s ye ar co un t. 27 30 The Politics of Authoritarian Rule can range from a complete ban on political parties (e.g., Chile under Augusto Pinochet’s military dictatorship, 1973–1987) to a single government-sanc- tioned party (typical for communist countries) to only minor restrictions on party organization (e.g., Mexico under the PRI, especially 1946–2000). The category of personalist dictatorship reflects high values of at least two other distinct conceptual dimensions of the political organization of dictator- ships. The first is the concentration of power in the hands of the dictator. Although it is rare, power in some dictatorships may be dispersed across indi- viduals, institutions, and levels of government, as it was in Mexico toward the end of the PRI’s rule. On the other hand, Joseph Stalin’s and Mao Zedong’s despotic rule exemplifies the upper limits on the power that a single individ- ual can acquire. Thus they correspond to the other extreme of this conceptual dimension. Yet in other cases, the category of personalist dictatorship also appears to reflect a high level of another conceptual dimension: the personalization of political interactions.24 This facet of personalism is sometimes characterized as neopatrimonial rule (Jackson and Rosberg 1982; Snyder 1992; Bratton and Van de Walle 1997; Brownlee 2002) or sultanism (Linz and Chehabi 1998). When Jackson and Rosberg (1982), Bratton and Van de Walle (1997), and Linz and Chehabi (1998) write of personalism, they refer to polities in which key political interactions are based on personal ties and traditional authority rather than formal institutions and rules. Whereas a high concentration of power in the hands of a dictator allows him to supersede many formal rules and institutions, the understanding of personalism as a high degree of concentration of power is distinct from one that emphasizes the personal nature of political interactions. Although both Stalin and Mao acquired immense amounts of power, they relied heavily on impersonal, formal rules, parties, and the bureaucracy in order to govern.25 As a consequence of this conceptual inconsistency, the categories of per- sonalist, military, and single-party dictatorship are neither mutually exclusive nor collectively exhaustive. As an example of this difficulty, consider the case of Syria: During Hafez al-Asad’s thirty-year rule, the Syrian regime ruled via a single party, the Baath Party.26 Yet at the same time, key posts in the gov- ernment and security apparatus were held by military officers who supported Asad’s takeover of the Baath Party in 1970. Hence the Syrian regime between 1970 and 2000 was ruled by military officers and maintained a single ruling party as well. Meanwhile, by the time of his death in 2000, Hafez al-Asad was an undisputed leader of the Syrian government, military, and Baath Party 24 See Geddes (1999b, 121–2). 25 On the operation of Stalin’s ruling circle, see Khlevniuk (2009) and Gorlizki and Khlevniuk (2004); on Mao, see MacFarquhar (1997a) and MacFarquhar and Schoenhals (2006). 26 Nominally, Syria has allowed for parties other than the Baath Party, but they must participate in the National Progressive Front, which is dominated by the Baath Party; see, e.g., Hinnebusch (2002, Chap. 4). I code as a “single-party” any dictatorship that nominally allows for multiple parties but requires that these operate under the leadership of a single party or as a single front. The World of Authoritarian Politics 31 (Hinnebusch 2002, Chap. 4) and built up a personality cult inspired in its aesthetic by the Soviet Union under Stalin (Wedeen 1999). As this case illus- trates, the simultaneous establishment of a single-party, military involvement in politics and a high concentration of power in a dictator’s hands are neither logically nor empirically exclusive.27 Meanwhile, some dictatorships do not fit any of the three categories and are missing from Geddes’s original data. This is especially the case for monarchies (e.g., Saudi Arabia) or regimes that simply do not experience military inter- ventions, do not have a single party, and are not governed by a leader who managed to monopolize power (e.g., contemporary Iran). Geddes’s typology thus fails to be collectively exhaustive. Another difficulty that arises in Geddes’s typology is that classification judg- ments must weigh conceptually incommensurable aspects of authoritarian pol- itics. This is evident in the difficult judgments that must be made in classify- ing regimes that share the features of several of the three pure types (see, e.g., Geddes 1999a, 20–3). It is especially pronounced in judging whether a regime is personalist: Personalist leaders emerge in dictatorships governed by the military (e.g., Augusto Pinochet in Chile, 1973–1990), by a single party (e.g., Joseph Stalin in the Soviet Union, 1924–1953), and sometimes by both (e.g., Hafez al-Asad in Syria, 1970–2000). Because personalist dictators rarely banish their military or party from government after they consolidate power, it is particu- larly difficult to objectively ascertain the occurrence and timing of a transition from a military or single-party dictatorship to a personalist one – primarily because each of the three types measures a different aspect of authoritarian politics.28 These difficulties – the lack of exclusiveness or exhaustiveness across cat- egories and the use of categories that weigh conceptually incommensurable aspects of authoritarian politics – limit the validity and reliability of data based on Geddes’s typology as well as the type of inferences that can be drawn from it. If, for instance, we are interested in the role of single parties in dictatorships, the relevant comparison groups are not military or personalist dictatorships but rather those dictatorships that do not restrict partisan organization to a single party. That group is best ordered by differentiating between regimes that ban parties entirely and those that allow for multiple parties. The two subsets may function very differently and therefore compare to single-party regimes in different, possibly opposing, ways. We entirely miss such differences when we instead use military and personalist dictatorships as a comparison 27 Geddes (1999a) recognizes this difficulty when she creates hybrid regime types (see also Hade- nius and Teorell 2007), which comprise 25 percent of her data. Nevertheless, in empirical work that relies on Geddes’s typology, these hybrids have often been subsumed under one of the pure types, treated as a separate hybrid category, or ignored. 28 Slater (2003) discusses such transitions within parties and militaries. He uses the terms machine and junta to describe an oligarchic balance of power within the two institutions and the terms bossism and strongman to describe personalization of power within parties and militaries, respectively. 32 The Politics of Authoritarian Rule table 2.2. Restrictions on Political Parties and Military Involvement in Authoritarian Politics, 1946–2008 Military Involvement Restrictions on in Politics Political Parties None Parties banned Indirect Single party Personal Multiple parties Corporate Note: See the codebook for details about coding rules for each category. group. Therefore, the use of categories that weigh conceptually incommen- surable aspects of authoritarian politics results in poorly formed comparison groups and potentially flawed inferences. These limitations are not confined to Geddes’s typology – I used it only as an illustrative example. Most typologies that classify dictatorships into a few ideal types or according to prominent descriptive features collapse multiple, conceptually distinct dimensions of authoritarian politics onto a single typol- ogy. They are, in turn, neither mutually exclusive nor collectively exhaustive and often ask for difficult classification judgments that weigh incommensurable aspects of authoritarian politics. The resulting categories have limited validity, reliability, and use as conceptually appropriate comparison groups. 2.3 this book’s approach to political organization of dictatorships As a foundation for the empirical work in this book, I devised and collected data on the political organization of dictatorships that attempt to overcome the limitations of existing typologies. Rather than classifying dictatorships into ideal types or according to their prominent descriptive features, I first identified the conceptual dimensions of authoritarian politics that I want to measure and then developed an appropriate scale for each dimension. The choice of dimen- sions is guided by the key aspects of the political organization of dictatorships that I explain throughout this book. At the country level, I measure four conceptual dimensions: military involve- ment in politics, restrictions on political parties, legislative selection, and exec- utive selection. Tables 2.2 and 2.3 summarize the categories used in their measurement. When measuring military involvement in authoritarian politics, I distin- guish among none, indirect, personal, and corporate military involvement. As outlined in the discussion of Geddes’s typology, some dictatorships experience no military involvement in politics (or a purely civilian government), whereas, in others, the military indirectly affects politics, even if the formal head of the government is a civilian. I consider military involvement in authoritarian The World of Authoritarian Politics 35 C um ul at iv e P ro po rt io n 1946 1955 1964 1973 1982 1991 2000 2009 Year Parties banned (17%) Single party (36%) Multiple parties (47%) 1 0. 8 0. 6 0. 4 0. 2 0 0 0. 2 0. 4 0. 6 0. 8 1 figure 2.4. Restrictions on political parties in dictatorships, 1946–2008. Note: Overall distribution of individual categories in parentheses. 1975–2006, to cover my entire sample period 1946–2008 and to match my modified coding scheme. On the conceptual dimension of legislative selection, I distinguish among dictatorships with no legislature, an unelected or appointed legislature, a legislature with one party or candidate per seat, a legislature with multiple parties in which the largest party controls more than 75 percent of the seats, a legislature with multiple parties in which the largest party controls less than 75 percent of the seats, and dictatorships with elected but nonpartisan legislatures.33 The largest party in authoritarian legislatures is most often what I will call in Chapter 6 a regime party – that is, sanctioned by the regime’s lead- ership. Saudi Arabia (1946–2008), Indonesia (1960–1965), Romania (1947– 1989), Mexico (1946–1975, 1994–2000), Mexico (1976–1993), and Kuwait (1963–1975, 1982–1985, 1992–2008) are instances of these six respective cat- egories. Figure 2.6 summarizes their distribution throughout the period 1946– 2008. When it comes to executive selection, I distinguish among dictatorships with an unelected executive, an executive that is elected but with only one party or candidate, an executive that is selected by a small, unelected body, an executive that is elected in a competition among multiple candidates and received more than 75 percent of the vote, and an executive that is elected in a 33 I present the 75 percent threshold because I consider it a reasonable (but nonetheless arbitrary) metric of significant oppositional presence. In the complete data, I record the exact fraction of legislative seats that the largest party controls. In the case of bicameral legislatures, all percentages refer to the lower house. 36 The Politics of Authoritarian Rule M ili ta ry In vo lv em en t i n P ol iti cs 17% 39% 44% 10% 37% 54% 29% 27% 44% 0 500 1,000 1,500 Frequency corporate personal civilian No party Single party Multiple parties (71%) (18%) (11%) figure 2.5. Joint distribution of restrictions on political parties and military involve- ment in authoritarian politics, 1946–2008. Note: Categories of indirect military involve- ment and restriction to two parties not included. Percentages under the categories of military involvement refer to each category’s share in the data. Percentages next to bars refer to each category’s share within the corresponding form of military involvement. C um ul at iv e P ro po rt io n 1946 1955 1964 1973 1982 1991 2000 2008 Year No legislature (19%) Unelected or appointed legislature (11%) One party or candidate per seat (37%) Largest party controls more than 75% of seats (13%) Largest party controls less than 75% of seats (17%) Nonpartisan legislature (3%) 1 0. 8 0. 6 0. 4 0. 2 0 0 0. 2 0. 4 0. 6 0. 8 1 figure 2.6. Legislative selection in dictatorships, 1946–2008. Note: Overall distribu- tion of individual categories in parentheses. The World of Authoritarian Politics 37 C um ul at iv e P ro po rt io n 1946 1955 1964 1973 1982 1991 2000 2008 Year Unelected executive (36%) One party or candidate (32%) Selected by a small, unelected body (7%) Elected by more that 75% of the vote (12%) Elected by less that 75% of the vote (14%) 1 0. 8 0. 6 0. 4 0. 2 0 0 0. 2 0. 4 0. 6 0. 8 1 figure 2.7. Executive selection in dictatorships, 1946–2008. Note: Overall distribution of individual categories in parentheses. competition among multiple candidates and received less than 75 percent of the vote.34 Saudi Arabia (1946–2008), Syria (1971–2008), Brazil (1964–1979), Singapore (1979–2005), and Peru (1992–2000) respectively exemplify these five categories. Figure 2.7 portrays the changes in executive selection across dictatorships throughout the period 1946–2008. Because the two dimensions of legislative and executive selection are based on formal institutional criteria, they jointly provide a directly interpretable indicator of the degree of competitiveness across dictatorships.35 Table 2.4 dis- plays the joint distribution of these two dimensions. We see that dictatorships in which both the legislature and the executive are elected in a multiparty or multicandidate competition and the largest party or candidate received less than 75 percent of the vote account for about 450 country-years, or only about 10 percent of all country-year observations. These are dictatorships in which political competition is relatively open and are therefore most likely either to be misclassified as dictatorships or barely fail the criteria for democracy that I adopt. Table 2.4 shows that such “competitive” authoritarian regimes make up only about 10 percent of all dictatorships throughout 1946–2008, although they do account for between 20 percent and 30 percent of all dictatorships 34 In the case of multiple election rounds, these percentages refer to the last round. 35 This is in contrast to indexed measures or scales such as the Polity Score (Marshall and Jaggers 2008), which lack a direct institutional interpretation. Przeworski et al. (2000) and Cheibub et al. (2010) emphasize such direct interpretability as well as unambiguous operationalization of measurement in their discussion of regime classifications. 40 The Politics of Authoritarian Rule 40% 24% 22% 9% 5% 0 20 40 60 80 Consensus Elections Resignation Term limit Natural At least one day in office At least one year in office figure 2.8. Constitutional exits from office of authoritarian leaders, 1946–2008. Note: Exits of interim leaders are not included. Unambiguous determination of exit was not possible for thirteen leaders. number of constitutional and nonconstitutional forms of leader entry into and exit from office; coded for the use of violence and the participation of the military in these leadership transitions; and recorded the institutional and political affiliation prior to assuming office of all authoritarian leaders between 1946 and 2008. As a brief summary of these data, consider dictators’ constitutional and nonconstitutional exits from office. Recall from Chapter 1 that a leadership transition is constitutional when it follows an officially endorsed, typically constitutionally mandated process, such as an election, a vote by a ruling body, or a hereditary succession.38 Figure 2.8 summarizes the form and distribution of such exits. An exit due to natural causes – as in the case of a natural death or poor health – is the most frequent form of a constitutional exit from office. The second most frequent category is an exit due a binding term limit, followed by resignation, which describes those cases in which a dictator steps down without any apparent pressure by his inner circle to do so. Lost elections account for about 9 percent of constitutional exits. Cases in which an unelected, small body removes a leader from office account for an additional 5 percent – I refer to 38 The entries and exits of interim leaders – that is, leaders who hold office only temporarily, typically during a constitutional crisis – cannot be unequivocally characterized as constitutional or nonconstitutional; therefore, I treat them as a separate category (see the codebook for details.) With minor differences, my distinction between constitutional and nonconstitutional leadership changes mirrors those between regular and irregular leadership changes in Goemans et al. (2009); see also Goemans (2008). The World of Authoritarian Politics 41 these exits as consensus. Respective examples of these categories include Fidel Castro’s 2008 retirement from the Cuban presidency due to failing health; Carlos Salinas’s 1994 compliance with the “no reelection” provision that lim- its the Mexican presidency to a single term; Luis Somoza’s 1963 decision to leave the Nicaraguan presidency in favor of a loyal surrogate; apartheid-era departures of South African prime ministers from office after lost elections; and Field Marshal Thanom Kittikachorn’s resignation from the post of Thai Prime Minister, which he held briefly in Sarit Thanarat’s military regime of 1957–1963. I reviewed my data on dictators’ nonconstitutional exits from office in Chapter 1, where I distinguished among coups d’état, popular uprisings, transitions to democracy, assassinations, and foreign interventions. In fact, about two-thirds of all leadership transitions in dictatorships are nonconstitu- tional. Prominent examples of these subcategories include, respectively, Nikita Khrushchev’s 1964 removal by the Soviet leadership around Leonid Brezhnev; Fulgencio Batista’s 1959 overthrow by Fidel Castro’s Cuban Revolution; Rey- naldo Bignone’s 1983 resignation after the democratic election of Raúl Alfonsı́n to the Argentine presidency; the 1981 assassination of Egyptian president Anwar al-Sadat; and Saddam Hussein’s fall after the 2003 U.S. invasion of Iraq. We saw in Chapter 1 that by far the most frequent form of nonconsti- tutional leader exit is the coup d’état – which I defined as the removal of an authoritarian leader by his inner circle that is accompanied by the threat or actual use of force. These data on leadership change allow me to evaluate some of the key argu- ments in this book. Chapter 3 uses data on the duration of leaders’ tenures and the manner by which they leave office to test my claims about the differences in the allies’ ability to collectively constrain dictators in contested and established autocracies. I use information on the constitutionality of leadership transi- tions in Chapter 4, which assesses whether parties and legislatures alleviate commitment and monitoring problems in authoritarian power-sharing. That chapter also uses data on dictators’ exits from office due to natural causes as a benchmark for diagnosing potential endogeneity in the adoption of parties and legislatures. Finally, Chapter 5 uses data on military involvement in leaders’ entry and exit to test my predictions about the causes of military intervention in politics. 2.5 the survival of authoritarian ruling coalitions The data on institutions and leadership change across dictatorships reviewed so far cover many aspects of authoritarian politics. However, some of my claims in this book – as well as many claims in the extant literature on authoritarian politics – are best assessed not at the country or leader level of observation but rather at the level of what I call the authoritarian ruling coalition. For instance, if we are interested in whether a regime-sanctioned party contributes to authoritarian stability, we want evaluate whether such parties result in the 42 The Politics of Authoritarian Rule table 2.5. Political Affiliation of Authoritarian Leaders, 1946–2008 Political Affiliation Frequency Percentage Regime 408 60.45 Unaffiliated 135 20.00 Opposition 79 11.70 Independence 53 7.85 Note: The unit of observation is an authoritarian leader. See the codebook for details about coding rules for each category. continuity in power of the same political coalition – as in the case of Mexico under the PRI. Empirical studies that assess such claims by looking instead at the tenures of individual leaders or the duration of entire authoritarian regimes potentially confound the effect of institutions on the stability of authoritarian ruling coalitions with their effect on the survival of individual dictators or the stability of dictatorship as a regime type. I overcome this limitation by using information on the political and institu- tional affiliation of dictators in order to identify authoritarian ruling coalition spells. I define the latter as an uninterrupted succession in office of politically affiliated authoritarian leaders. More precisely, a leader was politically affil- iated with the previous leader and, hence, from the same ruling coalition if he was a member of the government, a government party, the royal or rul- ing family, or a military junta under the previous authoritarian leader. The survival of a particular ruling coalition thus corresponds to the continuation in power of what we colloquially refer to as a particular “political regime” or “dictatorship” – as in “the communist regime” in China (1949–), “the PRI regime” in Mexico (1929–2000), or “the military dictatorship” in Brazil (1964–1984). Table 2.5 summarizes the political affiliation of all authoritarian leaders during the period 1946–2008. It shows that about 60 percent of all dictators come from the same ruling coalition as their predecessor. Prominent exam- ples of ruling coalitions that span multiple leaders include the leadership in Communist and Baathist regimes, Mexican presidents under the PRI, hered- itary successions in many Middle Eastern monarchies, and the leadership of the Argentine and Brazilian military governments. As we saw in Chapter 1, however, leadership changes within the same ruling coalition do not always occur peacefully or constitutionally. To name one example, before becoming the president of Syria in 1970, Hafez al-Asad served as a Minister of Defence in the Baath government that came to power in 1966. Al-Asad came to power by staging a coup against the de facto head of government, Salah Jadid, after growing disagreements about the direction of the country between Jadid’s ide- ological and al-Asad’s pragmatic faction within the Baath Party and military (Seale 1990, chap. 11). The World of Authoritarian Politics 45 of authoritarian politics that these data describe: legislative and executive selec- tion, restrictions on political parties, and military involvement in authoritarian politics, as well as leadership change and ruling-coalition duration and compo- sition. I also explain why the difference between dictatorship and democracy is best thought of as first one of kind and only then one of degree; why we should abandon the flawed practice of classifying dictatorships into a few ideal types or according to their prominent descriptive features; and why our data on formal institutions need to be complemented by other, more credible measures of their binding power, such as the data on leadership change used throughout this book. The next two chapters present my substantive arguments about the first of the two conflicts that I argue drive the politics of dictatorships: the problem of authoritarian power-sharing. 2.7 appendix: authoritarian spells, 1946–2008 The following table lists all authoritarian spells during the period 1946–2008 that satisfy my definition of dictatorship. Each entry consists of a country name, the first and last year of the authoritarian spell, and its origin and end. The first and last year of each authoritarian spell corresponds to the first and last calendar year that a country entered as a dictatorship, respectively. For instance, 1956 is the first calendar year that Argentina entered as a dictatorship following the military coup that overthrew Juan Perón in September 1955, and 1958 is the last calendar year that Argentina entered as a dictatorship (with a democratic legislative and presidential election taking place that year). A missing last year indicates that a country was a dictatorship as of 2008. Country Years Origin End Afghanistan 1946–1977 Continuance No authority 1989–1991 No authority No authority 1996–2000 No authority No authority Albania 1946–1991 Continuance Democracy Algeria 1962–1993 Independence No authority 2002– No authority Continuance Angola 2002– No authority Continuance Argentina 1956–1958 Democracy Democracy 1963–1963 Democracy Democracy 1967–1973 Democracy Democracy 1977–1983 Democracy Democracy Azerbaijan 1991– Independence Continuance Bahrain 1971– Independence Continuance Bangladesh 1971–1991 Independence Democracy (continued) 46 The Politics of Authoritarian Rule (continued) Country Years Origin End Belarus 1997– Democracy Continuance Benin 1962–1991 Democracy Democracy Bhutan 1971– Independence Continuance Bolivia 1946–1982 Continuance Democracy Brazil 1965–1986 Democracy Democracy Brunei 1984– Independence Continuance Bulgaria 1946–1990 Continuance Democracy Burkina Faso 1960– Independence Continuance Burundi 1962–1992 Independence No authority 1996–1999 No authority No authority 2003–2005 No authority Democracy Cambodia 1953–1969 Independence No authority 1976–1978 No authority No authority 1988– No authority Continuance Cameroon 1960– Independence Continuance Cape Verde 1975–1991 Independence Democracy Central African Republic 1960–1993 Independence Democracy 2004– Democracy Continuance Chad 1960–1977 Independence No authority 1984– No authority Continuance Chile 1974–1990 Democracy Democracy China 1950– No authority Continuance Colombia 1951–1958 Democracy Democracy Comoros 1975–2004 Independence Democracy Congo (Brazzaville) 1964–1992 Democracy Democracy 1999– No authority Continuance Congo (Zaire) 1966–1995 No authority No authority 2001–2006 No authority Democracy Costa Rica 1946–1949 Continuance Democracy Cuba 1953– Democracy Continuance Cyprus 1960–1962 Independence No authority 1968–1983 No authority Democracy Czechoslovakia 1949–1990 Democracy Democracy Djibouti 1977– Independence Continuance Dominican Republic 1946–1964 Continuance No authority 1967–1978 No authority Democracy Ecuador 1946–1948 Continuance Democracy 1962–1979 Democracy Democracy 2001–2002 Democracy Democracy Egypt 1946– Continuance Continuance El Salvador 1946–1980 Continuance No authority Equatorial Guinea 1968– Independence Continuance Eritrea 1993– Independence Continuance Ethiopia 1946–1979 Continuance No authority 1992– No authority No authority The World of Authoritarian Politics 47 Country Years Origin End Fiji 1988–1991 Democracy Democracy 2007– Democracy Continuance Gabon 1960– Independence Continuance Gambia 1965– Independence Continuance Georgia 1991–2004 Independence Democracy German Democratic Republic 1954–1990 Independence Democracy Ghana 1959–1969 Democracy Democracy 1973–1979 Democracy Democracy 1982–1992 Democracy Democracy Greece 1968–1974 Democracy Democracy Grenada 1980–1984 Democracy Democracy Guatemala 1955–1986 Democracy Democracy Guinea 1958– Independence Continuance Guinea-Bissau 1974–1994 Independence Democracy 2000–2000 Democracy Democracy 2004–2005 Democracy Democracy Guyana 1966–1992 Independence Democracy Haiti 1946–1990 Continuance Democracy 1992–1994 Democracy Democracy 2000– Democracy Continuance Honduras 1946–1958 Continuance Democracy 1964–1971 Democracy Democracy 1973–1982 Democracy Democracy Hungary 1946–1990 Continuance Democracy Indonesia 1949–1999 Independence Democracy Iran 1946– Continuance Continuance Iraq 1946–2002 Continuance No authority Ivory Coast 1960–2001 Independence No authority 2007– No authority Continuance Jordan 1946– Independence Continuance Kazakhstan 1991– Independence Continuance Kenya 1963–2002 Independence Democracy Korea, North 1948–1949 Independence No authority 1954– No authority Continuance Korea, South 1948–1949 Independence No authority 1954–1988 No authority Democracy Kuwait 1961– Independence Continuance Kyrgyzstan 1991–2005 Independence Democracy Laos 1974– No authority Continuance Lebanon 2005– Democracy Continuance Lesotho 1971–1993 Democracy Democracy Liberia 1946–1989 Continuance No authority 1996–2006 No authority Democracy Libya 1951– Independence Continuance (continued) 50 The Politics of Authoritarian Rule (continued) Country Years Origin End Yemen 1990– Independence Continuance Yemen Arab Republic 1946–1961 Continuance No authority 1968–1989 No authority No authority Yemen People’s Republic 1967–1989 Independence No authority Yugoslavia 1946–1991 Continuance Independence Zambia 1964–1991 Independence Democracy Zimbabwe 1965–1975 Independence No authority 1980– No authority Continuance part i THE PROBLEM OF AUTHORITARIAN POWER-SHARING And Then There Was One! 55 book, I refer to such attempts as allies’ rebellions. Because rebellions may fail and the allies typically have only limited information about the dictator’s actions, they will be reluctant to rebel under most circumstances, thereby giv- ing the dictator an incentive to try his luck and attempt to acquire power at their expense. If he succeeds in several power grabs without being stopped, the dictator may accumulate enough power that the allies will no longer be able to stage a rebellion that could topple him. It is precisely this type of dynamic that allows for the emergence of a personal autocracy. This chapter thus answers one question that Stalin’s rise to the pinnacle of Soviet power frequently evokes: Why didn’t anyone stop him before it was too late? The short answer is as follows: The allies tried but failed. Indeed, shortly before his death in 1924, Vladimir Lenin was one of the first to warn about Stalin’s appetite for power (Suny 1998, 143–4); Martemian Riutin’s failed 1932 attempt to organize resistance against Stalin’s emergent dictatorship was the last before the Great Purges cemented Stalin’s personal autocracy (Suny 1998, 254–6). After the Purges, Stalin achieved “limitless power over the fate of every Soviet official, including the top leaders” (Khlevniuk 2009, 247). However, this answer – they tried but failed – is too short and this chapter clarifies why: The answer fails to appreciate that the reasons for the emergence of personal autocracy are structural. As emphasized in Chapter 1, authoritarian elites operate under distinctly hazardous conditions. They cannot rely on an independent authority to enforce mutual agreements, and violence is the ever- present, ultimate arbiter of their conflicts. These dismal circumstances ensure that any dictator’s aspiration to become the next Stalin is matched by the opportunity to do so. In fact, we will see that even if a dictator’s allies do their best to deter him from usurping power at their expense, their ability to reign him in will be limited and intimately tied to the distinctive conditions under which authoritarian elites operate. Rather than an accident of history, the emergence of personal autocracy is a systematic phenomenon. This chapter’s analysis of the emergence of personal autocracy also clarifies that authoritarian power-sharing across dictatorships generally takes two qual- itatively distinct forms. In the first, which I call contested autocracy, politics is one of balancing between the dictator and his allies, and the latter are capable of using the threat of a rebellion to deter the dictator’s opportunism, even if imperfectly. By contrast, established autocrats have acquired enough power so they can no longer be credibly threatened by an allies’ rebellion. Thus, even if the distribution of power between the dictator and his allies spans a contin- uum – as in the selectorate theory of Bueno de Mesquita et al. (2003) – there are, in fact, only two qualitatively distinct power-sharing regimes: Contested autocrats can be credibly threatened with a removal; established autocrats have effectively monopolized power.4 This theoretical difference between contested 4 In the next chapter, we will see that further distinctions can be drawn among contested auto- cracies depending on the role of institutions in power-sharing. In this chapter, I intentionally do not consider the potential role of institutions in power-sharing. 56 The Politics of Authoritarian Rule and established autocracy corresponds to the empirical difference between oli- garchy and personal autocracy. Dictators with seemingly unlimited powers have been labeled alternatively as personalist (Geddes 1999a), patrimonial (Zolberg 1966; Roth 1968), neopat- rimonial (Jackson and Rosberg 1982; Snyder 1992; Bratton and Van de Walle 1997; Brownlee 2002), and sultanistic (Linz and Chehabi 1998). Thus the theoretical equilibrium of established autocracy naturally corresponds to these notions and clarifies why such dictators typically enjoy long tenures and die in their bed: They have effectively eliminated any threats from their inner circle. Nevertheless, I intentionally refer to dictatorships in which a dictator’s inner circle cannot credibly threaten him with removal as established rather than personalist, neopatrimonial, or sultanistic: I emphasize that the former label is a theoretical equilibrium, whereas the latter three are empirical categories that carry additional connotations, such as charismatic leadership, reliance on traditional institutions rather than modern bureaucracy, or the lack of clear boundaries between the state and the leader’s personal domain. Such empirical connotations may be implied by but are neither necessary nor do they define the analytical distinction between contested and established autocracy. The category of a personalist dictatorship was introduced by Geddes (1999a), who distinguishes such regimes from military and single-party dic- tatorships. Unfortunately, she did not temporally distinguish the type of dicta- torship that existed prior to any personalist regime – which would correspond to the equilibrium of contested autocracy herein – from the period of the per- sonalist regime proper. Nevertheless, we do observe the consolidation of power that culminates in what Geddes calls a personalist dictatorship across all types authoritarian regimes.5 Prominent examples include both military and single- party regimes, as exemplified by Francisco Franco and Joseph Stalin. The the- ory presented here provides the theoretical microfoundations that explain why such degeneration into personal rule – to borrow Brooker’s expression (2000, Chap. 6) – occurs across various types of dictatorships: All dictatorships oper- ate in a setting that lacks an independent authority that can enforce mutual agreements and in which violence is the ultimate arbiter of political conflicts. The transition from contested to established autocracy analytically mirrors the rise to uncontested power by some of the most iconic personalist dictators: Mao Zedong, Saddam Hussein, and “Papa Doc” Duvalier, to name a few.6 On any of these trajectories, an authoritarian leader assumes office as the “first among equals” and succeeds over time in accumulating enough power to become an invincible autocrat. The logic outlined previously explains why such a trajectory is possible but at the same time highly unlikely. Thus, however fascinating Mao’s, Hussein’s, and Duvalier’s path to power may be, they should also be highly unrepresentative of the “average dictator.” The average dictator 5 See also Brooker (2000, 37), Hadenius and Teorell (2007), and Slater (2003). 6 See Khlevniuk (2009) and Gorlizki and Khlevniuk (2004) on Stalin; MacFarquhar (1997a) and MacFarquhar and Schoenhals (2006) on Mao; and Makiya (1998) and Karsh (2002) on Hussein. And Then There Was One! 57 does not survive in office long enough to have the privilege of becoming a household name. This empirical insight is developed in Section 3.3, which examines the sta- tistical implications of my theoretical arguments. The long-run statistical dis- tributions of several quantities of political interest – including the duration of tenure before a dictator is removed by a rebellion and the time in office until he becomes established, among others – can be derived directly from the theoreti- cal model in this chapter. In fact, when actual data on tenures of authoritarian leaders are discussed in Section 3.4, I show that the probability density of the time that a dictator stays in office implied by the present model closely mir- rors the actual distribution of dictators’ tenures. Consistent with the arguments in this chapter, I also find that the longer an authoritarian leader stays in office, the less likely he is to be removed by a rebellion as opposed to one manner certainly unrelated to interaction with his inner circle: a natural death. The next section outlines why and how power-sharing fails in dictator- ships. It also explains why the emergence of established autocracy is a rare but systematic feature of authoritarian politics and why its occurrence depends in part on plain luck and crucially on an endogenously evolving balance of power between a dictator and his allies. Section 3.2, develops a game-theoretic model that highlights limits to the allies’ ability to deter the dictator’s oppor- tunism under conditions that are characteristic of most dictatorships: the lack of an independent authority that would enforce agreements among the elites, the ever-present option of resorting to violence, and widespread secrecy. Sec- tion 3.3 derives the empirical implications of my arguments, which I evaluate by examining data on leadership change across dictatorships in Section 3.4. I conclude by discussing the rationale behind several distinctive practices used by established autocrats: personality cults as well as arbitrary and unexpected rotations, dismissals, and promotions of their key administrators or military commanders. I explain why personality cults as well as these other more pecu- liar practices are aimed not at the ideological conversion of the masses but rather as a public signal of the dictator’s paramount political status under established autocracy. 3.1 authoritarian power-sharing and the emergence of personal autocracy This chapter studies a political setting with two key players, the dictator and the ruling coalition. The latter is composed of the dictator’s allies who jointly, with him, hold enough power to be both necessary and sufficient for a regime’s survival.7 For instance, the Syrian government of Hafiz al-Asad (1971–2000) 7 I use the term power very broadly: Both the dictator and members of the ruling coalition may derive power from economic or military resources or by having a large number of loyal followers. Loyalty in turn may be the result of ethnic, sectarian, or tribal ties or patronage or it may have more elusive foundations, as in the case of personal charisma. 60 The Politics of Authoritarian Rule the bureaucracy rather than formal changes in the bureaucratic hierarchy. Such loyalties often develop institutionally, as in the case of many military dictators, but also may be tribal, ethnic, or sectarian (see, e.g., Bratton and Van de Walle 1997; Van de Walle 2001). An important step in Saddam Hussein’s rise in power, for instance, was the gradual elimination of the Baath Party’s indepen- dent institutional influence on the regime via the appointment of individuals from Tikrit – his place of origin – into key positions in the bureaucracy. In the late 1970s, the entrenchment of the Tikritis in the government reached such major proportions that Hussein felt the need to conceal it from public view by abolishing family names denoting place of origin (Karsh 2002, 182). The allies’ imperfect information about the dictator’s actions and a rebel- lion’s potential failure jointly undermine its ex-post credibility. This is how the two factors interact: The potential failure of any rebellion makes its staging costly – even if it is expected to succeed. The dictator’s allies would therefore like to avoid staging it unless they are certain that the dictator is indeed trying to usurp power at their expense. But, because they observe only an imperfect signal of the dictator’s actions, they never have such certainty. Stated simply, the allies would like to threaten a rebellion but will be reluctant to carry it out. Importantly, because the dictator anticipates this dilemma, the allies’ ability to deter his opportunism will be limited. The model in Section 3.2 clarifies that the more precise the allies’ information about the dictator’s actions or the closer the dictator is to consolidating power, the more willing the allies will be to act on their threat to rebel. Crucially, however, the allies’ reluctance to rebel will invariably tempt the dictator to usurp power in the hope that the allies will fail to either detect or act on it. More technically, the dictator will try to usurp power with a positive probability. This tenuous ex-post credibility of the allies’ threat of a rebellion, caused by an interplay of the allies’ imperfect information about the dictator’s actions and the potential failure of any rebellion, is a key obstacle to successful authoritarian power-sharing: Even if the ruling coalition acts optimally, the dictator may be sufficiently fortunate to accumulate enough power to eliminate them altogether. I investigate this possibility in a dynamic setting in which the balance of power between the dictator and the ruling coalition evolves endogenously. In some periods, the dictator will be fortunate enough that even when he behaves opportunistically, a rebellion will either not be staged or fail, thereby shift- ing the balance of power in the dictator’s favor. If the dictator succeeds in several power grabs, he may accumulate enough power that the ruling coali- tion will no longer be willing to rebel – the threat of a rebellion will lose ex-ante credibility. This, according to the Resolution on Party History (1949– 81), is what happened during Mao Zedong’s rule. The Resolution, which after the Chairman’s death summarized the Chinese Communist Party’s lessons from the Party’s total subjugation to Mao’s whims, explains that his growing arro- gance and arbitrariness “took place only gradually” and that “the Central Committee of the Party should be held partly responsible” for failing to pre- vent it (cited in MacFarquhar and Schoenhals 2006, 458). And Then There Was One! 61 In the formal model examined in the next section, I show that this dynamic results in the emergence of two qualitatively distinct power-sharing regimes. Under contested autocracy, a rebellion threatened by the ruling coalition has sufficient ex-ante credibility to deter his opportunism, even if only partially. Thus, contested autocracy is an equilibrium in which authoritarian politics is characterized by power-sharing – albeit imperfect – between the dictator and the ruling coalition. Although the dictator may be the most powerful member of the ruling coalition, he rules in the shadow of the threat of a rebellion. This type of power-balancing appears to have characterized the interaction between the General Secretary of the Communist Party and the Politburo after Stalin’s death. In Zemtsov’s (1991, 133) depiction, for instance, . . . the general secretary’s power or potential is inversely proportional to the influence of the Politburo members, who aim at maintaining a delicate balance between his power and theirs. They cannot let the general secretary accumulate too much power, for they would they find themselves devoid of influence in decision-making. . . . I call the second power-sharing regime established autocracy, which emerges after a dictator succeeds in consolidating enough power that he can no longer be credibly threatened by the ruling coalition. Under this “degenerated” power- sharing regime, rebellions do not occur and the dictator has effectively elimi- nated the ruling coalition, whose support is no longer necessary for his survival. In their study of personal rule in Africa, Jackson and Rosberg (1982, 143) call such dictators “African Autocrats” and emphasize – as I do herein – that what distinguishes the African Autocrat is . . . not ideology or ruling style but by his greater freedom to act as he sees fit. He is freer to break agreements (or not to make them in the first place) because those with whom he may have them are in no position to enforce them. There are no powerful rivals with whom he must contend. The transition from contested to established autocracy therefore can be seen as one from oligarchy to autocracy: Instead of allies who share power with the dictator and may constrain his choices, members of the ruling coalition become administrators who are fully subservient to the dictator and do not share power with him in any meaningful sense. In fact, historical accounts of authoritarian politics identify precisely such a dichotomy in the power trajectories of dictators.11 According to Teiwes (2001, 79), Mao Zedong’s tenure “can essentially be divided between the period before 1958 when the Chairman listened to interests within the system and sought results that took those interests into account . . . , and the subsequent ‘later Mao’ period when he simply overrode interests. . . .” In Jackson and Rosberg’s 11 Meanwhile, classical philosophers have drawn distinctions between political regimes that par- allel the difference between contested and established autocracy. In Politics, Aristotle distin- guished between the government of one or of a few; in The Spirit of the Laws, Montesquieu differentiated between monarchical and despotic regimes; and in The Prince, Machiavelli sepa- rated limited and absolute princes. 62 The Politics of Authoritarian Rule account (1982, 170–1), Mobutu Sese Seko’s tenure in office consists of the period before 1970, when he was consolidating power, and the period after 1970, when his “personal autocracy was firmly established” and “old political allies . . . who exhibited the slightest sign of independence were purged.” Finally, as Khlevniuk’s (2009, 246) summary in the epigraph at the beginning of this chapter highlights, the same can be said about Stalin’s trajectory in power – there were only two periods in terms of the structure of supreme authority: oligarchy and dictatorship. Among the possible power trajectories explained by the arguments in this chapter is one on which an authoritarian leader assumes office as the “first among equals” but over time, as a result of opportunism and luck, accumulates enough power to become an invincible autocrat. Observers are often puzzled by how – typically in several distinct stages along such a trajectory – the dictator’s old allies become his new enemies. Consider Karsh’s characterization of Saddam Hussein’s position as the apparent successor of Ahmed Hassan al- Bakr, after the latter resigned his presidency: [Saddam] was not content with the comfortable majority he enjoyed in the state’s ruling institutions. . . . He was at once far more powerful than all his comrades put together, and far more vulnerable to attack from them. (2002, 113) This chapter explains this dynamic: After every successful power grab by the dictator, members of the ruling coalition become more concerned about the possibility that he will become established and eliminate them. As dis- cussed in the next section, the dictator’s appetite for power also grows as he strengthens his position. Meanwhile, the ruling coalition counters this expand- ing appetite by rebelling with an increasing probability. The ladder to ultimate power becomes more slippery as the dictator advances to the top. As suggested at the beginning of this chapter, Stalin’s rise to power is perhaps the most prominent example of the transition from contested to established autocracy. Stalin rose from the position of an “obscure party functionary” (Suny 1998, 49) in the 1920s to an indomitable autocrat by the end of the 1930s. By the end of the 1920s, Stalin eliminated from the Communist Party the key opposition groups associated with Trotsky, Zinoviev, and Bukharin (Suny 1998, 165–6). In 1924, when the terminally ill Lenin warned that Stalin had accumulated too much power, the Party ignored him, and Stalin was retained as the general secretary (Suny 1998, 146–8). Ten years later, in another push to consolidate his power, Stalin’s Purges transformed the Communist Party from an ideological organization of elites and intellectuals, whose primary political interest was the promotion of Communism, into a party in which power rested in the hands of people of low-class origins whose primary personal loyalty was to Stalin. Stalin eliminated more than one-half of the 1,961 delegates and more than two-thirds of the 139 Central Committee members elected at the 17th Party Congress in 1934, the last such Congress before the Great Purges. He purged about one-half of the officer corps from the army and executed more Soviet generals than would be killed in World War II (Suny 1998, 261–8). In a And Then There Was One! 65 complyrenege Dictator low high Nature low high Nature Ruling coalition Ruling coalition no rebellion b + μb, 1 − rebellion no rebellion b, 1 rebellion no rebellion b + μb, 1 − rebellion no rebellion b, 1 rebellion fail b, 0 succeed 0,1 Nature fail b + μb, 0 succeed 0,1 Nature fail b, 0 succeed 0,1 Nature fail b + μb, 0 succeed 0,1 Nature figure 3.1. Authoritarian power-sharing game in extensive form. member of the ruling coalition prefers being at the mercy of a dictator who reneged to participating in a failed rebellion. Figure 3.1 portrays the timing of the actions and the payoffs in this author- itarian power-sharing game. First, the dictator chooses whether to renege or comply. Second, the ruling coalition observes an imperfect signal of the dicta- tor’s action and decides whether to rebel. Finally, if a rebellion is staged, then it either succeeds or fails. 3.2.1 Authoritarian Power-Sharing Can the ruling coalition deter the dictator’s opportunism using only the threat of a rebellion? The threat of a rebellion would certainly deter the dictator’s opportunism if his actions were perfectly observable. The dictator would antic- ipate that if he reneged, the ruling coalition would plainly observe his actions and have no choice but to rebel. He would therefore always comply. This reasoning does not extend to the present setting, in which a dictator’s actions are not perfectly observable. Recall that the likelihood of a successful rebellion depends on the balance of power between the dictator and the ruling coalition. Thus staging a rebellion is costly to the ruling coalition because it may fail. As a result, the ruling coalition would prefer to threaten a rebellion if it observes a high signal, have the dictator believe this threat and therefore comply from the outset, but ultimately not carry out the rebellion despite having 66 The Politics of Authoritarian Rule observed a high signal. Of course, believing such a threat on the dictator’s part would not be consistent with the ruling coalition’s strategy. Instead, the dictator would anticipate the ruling coalition’s line of reasoning, not consider the threat of a rebellion credible, and thus renege. The threat of a rebellion is credible only if the ruling coalition has an incen- tive to carry out its threat after the dictator has acted and the ruling coalition has observed an imperfect signal of his action. This incentive exists only if the possibility that the dictator reneged is real. In other words, the threat of a rebellion is credible only if the dictator reneges with a positive probability. This logic can be verified by examining the perfect Bayesian equilibrium of this authoritarian power-sharing game. Allowing for mixed strategies, this game proceeds in four stages. First, the dictator reneges with probability α. Second, depending on the dictator’s action, nature determines the realization of the signal θ with probability πθa . Third, the ruling coalition rebels with probability βθ after it observes the signal θ . For example, βH is the probability that the ruling coalition rebels when it observes a high signal. Fourth, if a rebellion is staged, it succeeds with probability ρ. First, consider the ruling coalition’s equilibrium strategy βθ . Based on the previous discussion, we may verify that there is no equilibrium in which the dictator uses a pure strategy and the ruling coalition conditions its decision to rebel on the observed signal.17 In a mixed-strategy equilibrium, the ruling coalition rebels with probability βθ such that, given the correlation between his actions and the signal θ , the dictator is indifferent between reneging and complying. Thus we have∑ θ∈{H,L} πθc [βθ (1 − ρ)b + (1 − βθ )b] = ∑ θ∈{H,L} πθd [βθ (1 − ρ)(b + μb) + (1 − βθ )(b + μb)] , or, equivalently,∑ θ∈{H,L} πθc(1 − βθρ)b = ∑ θ∈{H,L} πθd(1 − βθρ)(b + μb). (3.1) Solving (3.1) for βH, we obtain βH = μ πHr (1 + μ) − πHc ( 1 ρ − βL ) + βL, which implies that βH > βL. Intuitively, the ruling coalition rebels with a greater probability after observing a high signal than a low signal. Among the possible pairs (βL, βH) that satisfy equality (3.1), only the pairs (βL = 0, βH > 0) and 17 As is the case with many extensive games with imperfect information, this game also has an implausible equilibrium in which the ruling coalition ignores the information conveyed by the signal θ , rebels with probability 1, and the dictator reneges with probability 1. And Then There Was One! 67 (βL > 0, βH = 1) constitute an equilibrium.18 Moreover, the equilibrium in which βL = 0 and βH > 0 is focal from the point of view of a dictator and a ruling coalition who would like to share power: Both actors prefer this equilibrium to that in which βL > 0 and βH = 1.19 Therefore the remainder of this chapter restricts attention to the relevant equilibrium, in which βL = 0 and βH > 0. In this equilibrium, the ruling coalition rebels with a positive probability only after observing a high signal, β∗L = 0 and β∗H = μ ρ [πHr (1 + μ) − πHc] > 0 . (3.2) Furthermore, (3.2) implies that the equilibrium probability that the ruling coali- tion rebels after observing a high signal β∗H is decreasing in both the probability that a rebellion succeeds (ρ) and the informativeness of the signal θ about the dictator’s actions (πHr/πHc), whereas it is increasing in the amount of power μ that the dictator acquires by reneging. This raises the possibility that μ could be so large that the dictator would renege even if βH = 1. That is, the dictator would renege even if the ruling coalition always rebelled after observing a high signal. To focus on settings in which a dictator is potentially deterrable, I make the following assumption. Assumption 3.1 (Limited Temptation to Consolidate Power). The additional power μ that the dictator acquires by reneging cannot be so large that the dictator reneges for all βH ≤ 1, μ < ρ(πHr − πHc) 1 − ρπHr . Assumption 3.1 admits a larger μ by the dictator when the signal θ is more informative about his actions and when the probability that a rebellion succeeds ρ is larger. Finally, what is the dictator’s equilibrium strategy α? In a mixed-strategy equilibrium, α must be such that the ruling coalition is indifferent between rebelling and not rebelling after observing a high signal. Although the ruling coalition does not directly observe the dictator’s actions, we can compute the conditional probability that the dictator reneged given the signal that the ruling coalition observes. I denote this probability as Pr(a|θ ). For example, Pr(d|H) is the probability that the dictator reneged, given that the ruling coalition observes a high signal. Using Bayes’ rule, we see that Pr(d|H) = πHrα πHrα + πHc(1 − α) . (3.3) 18 In any equilibrium in mixed strategies, the ruling coalition is indifferent between rebelling and not rebelling after observing a high signal or a low signal but not both. Therefore, in an equilibrium in mixed strategies, either βL = 0 or βH = 1. 19 A detailed proof of this claim, as well as of all the propositions that follow, is in the appendix to this chapter. 70 The Politics of Authoritarian Rule has accumulated more power? Under contested autocracy, the dictator reneges successfully when he reneges and (1) the ruling coalition observes a low signal; (2) the ruling coalition observes a high signal but does not rebel; or (3) the ruling coalition observes a high signal and stages a rebellion that fails. Thus the probability that the dictator gets away with reneging is Pr(Successful Reneging) = α∗[πLd + πHr (1 − β∗H) + πHrβ∗H(1 − ρ)]. The appendix to this chapter confirms that the probability that the dictator successfully reneges is indeed increasing in the dictator’s power. In other words, the moral hazard associated with authoritarian power-sharing intensifies as the dictator gains more power. Proposition 3.2 (Balance of Power). If b ∈ (0, ], then α∗, β∗H, and Pr(Success- ful Reneging are all increasing in b, whereas Pr(Successful Power-Sharing) is decreasing in b. We may also examine how the likelihood of successful power-sharing and the dictator’s successful reneging depend on the precision of the signal θ about his actions. We can check that the equilibrium probabilities α∗ and β∗H are decreasing in πHr and increasing in πHc. Therefore, when the signal θ about the dictator’s actions is more informative (i.e., πHr/πHc increases), the likeli- hood of successful power-sharing is greater. On the other hand, the relationship between the dictator’s probability of successfully reneging and the informative- ness of the signal θ is nonmonotonic. Proposition 3.3 (Transparency). If b ∈ (0, ], then Pr(Successful Power- Sharing) is increasing in the informativeness of the signal θ about the dictator’s actions, πHr/πHc. Finally, observe that the equilibrium probability with which the dictator reneges α∗ is decreasing in , the probability with which he eliminates a member of the ruling coalition if he successfully reneges. Although not an explicit part of this model, if larger ruling coalitions are associated with a lower probability of any ruling-coalition member being eliminated and, therefore, a lower , then larger ruling coalitions may be better able to deter the dictator’s opportunism and thus successfully share power. To illustrate the findings in this section, consider the following numeri- cal example. When πHr = 0.8, πHc = 0.2, μ = 0.2, b = 0.45, and = 0.5, a rebellion succeeds with the probability ρ = 0.55; the ruling coalition never rebels when the signal θ is low (β∗L = 0) but rebels when the signal θ is high with the probability β∗H = 0.48; and the dictator reneges with the probability α∗ = 0.69. The probability of successful power-sharing under these conditions is 0.28 and the probability that the dictator successfully reneges is 0.55. This is an example of contested autocracy because b < . However, if the dictator successfully reneges, his power grows to b = 0.54 > = 0.5. In that case, this regime becomes established with β∗L = β∗H = 0 and α∗ = 1. And Then There Was One! 71 3.2.3 A Model with Endogenously Evolving Balance of Power Although the results so far are based on a single-period extensive game, they suggest a dynamic interpretation of the dictator’s power trajectory. That is, we could conceive of a repeated game in which the balance of power between the dictator and the ruling coalition in each period depends on whether the dictator successfully reneged in the previous period. Proposition 3.1 implies that, under contested autocracy, the dictator will act opportunistically with a positive probability and the ruling coalition will rebel with a positive probabil- ity as well. Proposition 3.2 implies that the probability that the dictator indeed acquires more power is always positive and, in fact, increases with that power. Any contested autocrat therefore may become established if he succeeds in acquiring a sufficient amount of power, although such a trajectory is unlikely. I now examine such a multiperiod game. In an equilibrium of this game, the balance of power between the dictator and the ruling coalition evolves endogenously. The dynamic in this multiperiod game is qualitatively identical to that in the single-period game. In the next section, I use this multiperiod game to examine the implications of my theory for the statistical analysis of leaders tenures in authoritarian regimes. Periods are indexed by t = {T, T − 1, . . . , 1, 0} so in any period, T is the num- ber of times the dictator must successfully renege to become an established auto- crat. Thus, t = 1 denotes the period in which a single successful reneging turns contested into established autocracy. The game ends in period t = 0 in which b0 > and the ruling coalition’s threat to rebels is no longer ex-ante credible. In each period, the dictator and the ruling coalition receive one of the three possible payoffs portrayed in Figure 3.1. Recall that these payoffs depend on whether the dictator reneges, whether the ruling coalition rebels, and whether a rebellion succeeds. In any period, the existing balance of power bt summarizes the payoff-relevant history of play. Then Vt = (bt + δVt)(1 − αt) [πHcβt(1 − ρt) + πHc(1 − βt) + 1 − πHc] + (bt + μbt + δVt−1)αt [πHrβt(1 − ρt) + πHr (1 − βt) + 1 − πHr ] , and Ut = (1 + δUt)[αtπHrβtρt + (1 − αt)πHcβtρt + (1 − αt)πHc(1 − βt) + (1 − αt)(1 − πHc)] + (1 − + δUt−1) [αtπHr (1 − βt) + αt(1 − πHr )] are the discounted expected payoffs to the dictator and any member of the ruling coalition in period t, respectively, and δ ∈ (0, 1) is a discount factor. When the dictator becomes established, V0 = 1 and U0 = 1 − . Suppose that given an existing balance of power, the ruling coalition uses the threat of a rebellion in a way that is optimal from that period onward and ignores any previous history of play. That is, we examine a Markov perfect equilibrium of this multiperiod, authoritarian power-sharing game. Optimal strategies can be computed using backward induction by starting in period 72 The Politics of Authoritarian Rule 0.540.450.380.310.260.220.180.150.130.11 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0123456789 bt α , βH t figure 3.2. The probability that the dictator reneges, α∗ (dashed line), and the prob- ability that the ruling coalition rebels if it observes a high signal, β∗H (solid line), in an equilibrium of the multiperiod authoritarian power-sharing game. t = 1 and then proceeding recursively for the remaining periods. Explicit solu- tions obtained in this way are far too complicated algebraically to be useful. I therefore present a numerical example here instead. Suppose that πHr = 0.8, πHc = 0.2, μ = 0.2, = 0.5, and δ = 0.1 and the initial balance of power is 0.11. In this case, the dictator must renege suc- cessfully nine times to become established, T = 9 and bt = (0.11, 0.13, 0.15, 0.18, 0.22, 0.26, 0.31, 0.38, 0.45) for t = 9, . . . , 1. The equilibrium prob- abilities with which the dictator reneges and the ruling coalition rebels, α∗t and β∗Ht, are portrayed in Figure 3.2. The horizontal axis denotes both the periods t (i.e., upper axis) and the balance of power bt (i.e., lower axis) in these peri- ods. We see that the equilibrium probabilities that the dictator reneges and the ruling coalition rebels are increasing as he acquires more power until period 1, when b1 = 0.45. If the dictator successfully reneges in that period, he becomes established and the ruling coalition prefers to be at his mercy to staging a rebellion. This numerical example can be easily generalized to any number of periods T. 3.3 implications for the empirical study of authoritarian tenures Some of the key theoretical results so far have been stated with respect to the balance of power between the dictator and the ruling coalition – a factor And Then There Was One! 75 0 0. 02 0. 04 0. 06 0. 08 D en si ty 0 20 40 60 80 tau (periods) figure 3.3. Probability density of time-to-rebellion based on the numerical example. generalized negative-binomial distribution with T successes and a probability of success that decreases with t = T, . . . , 1. The equivalent continuous-time distribution is the generalized Gamma distribution (Gerber 1991). To illustrate these results, I continue with the numerical example from Section 3.2 and plot the probability density of time-to-rebellion and time- to-established autocracy in Figures 3.3 and 3.4, respectively. We may compare this numerical illustration with the distribution of successful coups d’état based on actual data in Figure 3.5. The model in Section 3.2 implies a distribution of successful coups that reasonably reflects real-world data.25 3.4 the path to personal autocracy and authoritarian exit from office A key implication of the theoretical analysis in this chapter concerns the empir- ical association between the length of dictators’ tenures and the manner by which they leave office.26 The longer a dictator is in office, the more likely it is 25 Understandably, existing large-N data do not record whether an autocracy is contested or established. We therefore should expect that observations of dictators’ tenures contain both contested and established autocrats. Whereas both the hazard of time-to-established autocracy and time-to-rebellions are increasing over time, the hazard of a successful rebellion declines relative to that of the dictator becoming established after a certain threshold time. In my numerical example, that time is thirty-five years. We therefore should expect the hazard of successful rebellions to be first increasing and then decreasing in actual real-world data. 26 Goemans (2008) comprehensively studies the manner in which leaders lose office across regime types. 76 The Politics of Authoritarian Rule 0 0. 01 0. 02 0. 03 D en si ty 0 20 40 60 80 100 tau (periods) figure 3.4. Probability density of time-to–established autocracy based on the numerical example. 0 0. 05 0. 1 0. 15 0. 2 D en si ty 0 10 20 30 40 Time (years) figure 3.5. The empirical density of coups d’état, 1946–2008. And Then There Was One! 77 18% 26% 48% 53% 44% 88% Ti m e in o ffi ce in y ea rs 0 20 40 60 80 100 Number of leader exits 26−30 21−25 16−20 11−15 6−10 1−5 Coups Natural Exits figure 3.6. The improving odds of dying in bed, 1946–2008. Note: Percentages refer to natural exits as a fraction of both types of exits in each tenure interval. that he is an established rather than a contested autocrat and the less likely it is that he will be removed from office by his inner circle. Hence long-lived author- itarian leaders should leave office more often in ways that are unrelated to their interaction with their inner circle, such as natural death, foreign intervention, or popular uprising.27 To evaluate this prediction, consider how the relative risk of coups d’état and natural deaths varies across dictators with different tenure durations.28 Figure 3.6 contrasts exits by coups to exits due to natural causes across six ordered, five-year intervals of tenure durations. Leaders who stayed in office for less than a year are not plotted because they may have been particularly vulnerable to coups. Leaders who stayed in office for more than thirty years also are excluded because only few of such long-serving leaders survive in any five- year tenure interval above thirty years and therefore may be unrepresentative.29 Consistent with the model in this chapter, the total number of leaders that manage to survive in office declines over time. More importantly, Figure 3.6 27 Nevertheless, there are examples of leaders (e.g., Haile Selassie of Ethiopia) who ruled for an unusually long time and managed to consolidate power in their hands but were later removed from office by their inner circle. The model in this chapter implies that such cases may occur but should be the exception rather than the rule. 28 Luttwak (1968), O’Kane (1981), Londregan and Poole (1990), Galetovic and Sanhueza (2000), and Belkin and Schofer (2003) examine coups d’état empirically. 29 There are 16, 44, and 4 leaders in the 31–35, 36–40, and 41–45 tenure intervals, respectively.