Intra-State Conflict, Governments and Security: Dilemmas of Deterrence and Assurance (Contemporary Security Studies)

Intra-State Conflict, Governments and Security

This volume seeks to understand the central role of governments in intrastate conflicts. The book explores how the government in any society plays two pivotal roles: as a deterrent against those who would use violence; and as a potential danger to the society. These roles come into conflict with each other, as those governments that can best deter potential rebels are also those that can do the most harm to their citizens. Therefore, a balance must be achieved, raising difficult tradeoffs for policy-makers. This volume marks a departure from studies of ethnic conflict and civil war in recent years, which have focused on failed states, in considering the idea that governments themselves may be the source of violence. The contributors not only explore the balancing act that governments must perform, but also the positive and negative roles that the international community can play in these conflicts. In doing so, the book covers a range of cases from both advanced and newer democracies to the most conflict-prone parts of the world. This book will be of interest to students of ethnic conflict, civil wars, peacekeeping, security studies, and international relations in general. Stephen M. Saideman is Canada Research Chair of International Security and Ethnic Conflict at McGill University. Marie-Joëlle Zahar is Associate Professor of Political Science at the Université de Montréal.

Contemporary security studies

NATO’s Secret Armies Operation Gladio and terrorism in Western Europe Daniele Ganser

The Political Economy of Peacebuilding in Post-Dayton Bosnia Tim Donais

The US, NATO and Military Burden-Sharing Peter Kent Forster and Stephen J. Cimbala

The Distracted Eagle The rift between America and old Europe Peter H. Merkl

Russian Governance in the TwentyFirst Century Geo-strategy, geopolitics and new governance Irina Isakova

The Iraq War European perspectives on politics, strategy, and operations Edited by Jan Hallenberg and Håkan Karlsson

The Foreign Office and Finland 1938–1940 Diplomatic sideshow Craig Gerrard

Strategic Contest Weapons proliferation and war in the greater Middle East Richard L. Russell

Rethinking the Nature of War Edited by Isabelle Duyvesteyn and Jan Angstrom

Propaganda, the Press and Conflict The Gulf War and Kosovo David R. Willcox

Perception and Reality in the Modern Yugoslav Conflict Myth, falsehood and deceit 1991–1995 Brendan O’Shea

Missile Defence International, regional and national implications Edited by Bertel Heurlin and Sten Rynning

Globalising Justice for Mass Atrocities A revolution in accountability Chandra Lekha Sriram Ethnic Conflict and Terrorism The origins and dynamics of civil wars Joseph L. Soeters Globalisation and the Future of Terrorism Patterns and predictions Brynjar Lia Nuclear Weapons and Strategy The evolution of American nuclear policy Stephen J. Cimbala Nasser and the Missile Age in the Middle East Owen L. Sirrs War as Risk Management Strategy and conflict in an age of globalised risks Yee-Kuang Heng Military Nanotechnology Potential applications and preventive arms control Jurgen Altmann NATO and Weapons of Mass Destruction Regional alliance, global threats Eric R. Terzuolo Europeanisation of National Security Identity The EU and the changing security identities of the Nordic states Pernille Rieker

International Conflict Prevention and Peace-building Sustaining the peace in post conflict societies Edited by T. David Mason and James D. Meernik Controlling the Weapons of War Politics, persuasion, and the prohibition of inhumanity Brian Rappert Changing Transatlantic Security Relations Do the US, the EU and Russia form a new strategic triangle? Edited by Jan Hallenberg and Håkan Karlsson Theoretical Roots of US Foreign Policy Machiavelli and American unilateralism Thomas M. Kane Corporate Soldiers and International Security The rise of private military companies Christopher Kinsey Transforming European Militaries Coalition operations and the technology gap Gordon Adams and Guy Ben-Ari Globalization and Conflict National security in a ‘new’ strategic era Edited by Robert G. Patman Military Forces in 21st Century Peace Operations No job for a soldier? James V. Arbuckle

The Political Road to War with Iraq Bush, 9/11 and the drive to overthrow Saddam Nick Ritchie and Paul Rogers Bosnian Security after Dayton New perspectives Edited by Michael A. Innes Kennedy, Johnson and NATO Britain, America and the Dynamics of Alliance, 1962–68 Andrew Priest Small Arms and Security New emerging international norms Denise Garcia The United States and Europe Beyond the neo-conservative divide? Edited by John Baylis and Jon Roper Russia, NATO and Cooperative Security Bridging the gap Lionel Ponsard International Law and International Relations Bridging theory and practice Edited by Tom Bierstecker, Peter Spiro, Chandra Lekha Sriram and Veronica Raffo Deterring International Terrorism and Rogue States US national security policy after 9/11 James H. Lebovic

Vietnam in Iraq Tactics, lessons, legacies and ghosts Edited by John Dumbrell and David Ryan Understanding Victory and Defeat in Contemporary War Edited by Jan Angstrom and Isabelle Duyvesteyn Propaganda and Information Warfare in the Twenty-first Century Altered images and deception operations Scot Macdonald Governance in Post-Conflict Societies Rebuilding fragile states Edited by Derick W. Brinkerhoff European Security in the TwentyFirst Century The challenge of multipolarity Adrian Hyde-Price Ethics, Technology and the American Way of War Cruise missiles and US security policy Reuben E. Brigety II International Law and the Use of Armed Force The UN charter and the major powers Joel H. Westra Disease and Security Natural plagues and biological weapons in East Asia Christian Enermark

Explaining War and Peace Case studies and necessary condition counterfactuals Jack Levy and Gary Goertz

Russian Energy Policy and Military Power Putin’s quest for greatness Pavel Baev

War, Image and Legitimacy Viewing contemporary conflict James Gow and Milena Michalski

The Baltic Question During the Cold War Edited by John Hiden, Vahur Made, and David J. Smith

Information Strategy and Warfare A guide to theory and practice John Arquilla and Douglas A. Borer Countering the Proliferation of Weapons of Mass Destruction NATO and EU options in the Mediterranean and the Middle East Thanos P. Dokos

America, the EU and Strategic Culture Renegotiating the transatlantic bargain Asle Toje Afghanistan, Arms and Conflict Post-9/11 security and insurgency Michael Bhatia and Mark Sedra

Security and the War on Terror Edited by Alex J. Bellamy, Roland Bleiker, Sara E. Davies and Richard Devetak

Punishment, Justice and International Relations Ethics and order after the Cold War Anthony F. Lang, Jr

The European Union and Strategy An emerging actor Edited by Jan Hallenberg and Kjell Engelbrekt

Intra-State Conflict, Governments and Security Dilemmas of deterrence and assurance Edited by Stephen M. Saideman and Marie-Joëlle Zahar

Causes and Consequences of International Conflict Data, methods and theory Edited by Glenn Palmer

Intra-State Conflict, Governments and Security Dilemmas of deterrence and assurance Edited by Stephen M. Saideman and Marie-Joëlle Zahar

First published 2008 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN Simultaneously published in the USA and Canada by Routledge 270 Madison Ave, New York, NY 10016 Routledge is an imprint of the Taylor & Francis Group, an informa business

This edition published in the Taylor & Francis e-Library, 2008. “To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.” © 2008 Selection and editorial matter, Stephen M. Saideman and Marie-Joëlle Zahar; individual chapters, the contributors All rights reserved. No part of this book may be reprinted or reproduced or utilized in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging in Publication Data A catalog record for this book has been requested ISBN 0-203-92711-7 Master e-book ISBN

ISBN10: 0-415-46050-6 (hbk) ISBN10: 0-203-92711-7 (ebk) ISBN13: 978-0-415-46050-7 (hbk) ISBN13: 978-0-203-92711-3 (ebk)

Contents

List of illustrations List of contributors Acknowledgments

1 Causing security, reducing fear: deterring intra-state violence and assuring government restraint

xi xiii xviii

1

STEPHEN M. SAIDEMAN AND MARIE-JOËLLE ZAHAR

2 Fear, preemption, retaliation: an empirical test of the security dilemma

20

STATHIS N. KALYVAS

3 Six feet over: internal war, battle deaths and the influence of the living on the dead

33

DAVID A. ARMSTRONG II AND CHRISTIAN DAVENPORT

4 Inequality, indivisibility and insecurity

54

S. MANSOOB MURSHED

5 Rulers as mass murderers: political institutions and human insecurity

72

AYSEGUL AYDIN AND SCOTT GATES

6 Resentment, fear, and the structure of the military in multiethnic states

96

ROGER PETERSEN AND PAUL STANILAND

7 Violence as politics: ETA and Basque nationalism ANDRÉ LECOURS

120

x Contents

8 Africa’s power sharing institutions as a response to insecurity: assurance without deterrence

138

DONALD ROTHCHILD

9 The “chicken or the egg”? External support and rebellion in ethnopolitics

161

YASEMIN AKBABA, PATRICK JAMES, AND ZEYNEP TAYDAS

10 Tackling the anarchy within: the role of deterrence and great power intervention in peace operations

182

SARAH-MYRIAM MARTIN-BRÛLÉ

11 Conclusion: dilemmas of insecurity – implications for research and policy

205

MARIE-JOËLLE ZAHAR AND STEPHEN M. SAIDEMAN

References Index

222 245

Illustrations

Figures 1.1 2.1 2.2 2.3 2.4 2.5 2.6 3.1 3.2 5.1 6.1 6.2 7.1

Deterrence versus assurance Control zone and indiscriminate violence at t1 Control zone and indiscriminate violence at t4 Control zone and selective violence at t1 Control zone and selective violence at t4 Distribution of selective violence in the Argolid Viet Cong selective violence, 1969 Trends over time in battle deaths and deaths from genocide/democide Contention and casualties Genocides/politicides and limits to political power Fear, resentment and military structure Ethnicity and militaries: composition and recruitment Attitudes of the Basques toward ETA

13 25 25 26 26 27 29 37 41 88 100 103 128

Tables 3.1 Descriptive statistics for battle deaths and genocide/ democide deaths in different contexts 3.2 Negative binomial regression results – conventional battle deaths for all observations and only country-years in conflict 3.3 Negative binomial regression results – conventional battle deaths for civil wars with and without genocide/politicide 3.4 Negative binomial regression results – conventional battle deaths plus genocide/politicide deaths 4.1 GDP per capita (1995 constant US$) growth rates 4.2 Number of countries in different ‘social’ groups 5.1 Random effects logit analysis: dependent variable, genocides/politicides

42

43 46 48 57 60 86

xii Illustrations 5.2 Random effects logit analysis: ethnicity and genocides/ politicides 5.3 Autocracies and genocides/politicides 6.1 Overall SANDF racial composition over time 6.2 SANDF officer corps racial composition (October 2003) 8.1 Recent post-civil war power sharing peace agreements in Africa 9.1 SUR analysis of external support for minorities at risk 9.2 SUR analysis of rebellion for minorities at risk 10.1 Typology of peace operations in Liberia and Sierra Leone (1990–2005) 10.2 Typology of peace operation’s success

87 91 115 115 146–149 175 177 188 189

Contributors

Yasemin Akbaba is an Assistant Professor at Gettysburg College. She received her PhD from University of Missouri-Columbia. Her research focuses on mobilization of ethnic and religious groups as well as international conflict. She is the co-author of “One Sided Crises in World Politics: A Study of Oxymoron, Violence and Outcomes” in International Interactions, and a book chapter (with Patrick James) “The Evolution of Iranian Interventionism: Support for Radical Islam in Turkey, 1982–2003” that will appear in Efraim Inbar and Hillel Frisch edited volume Radical Islam and International Security: Challenges and Responses (Routledge Press). David A. Armstrong II is a Post-Doctoral Fellow in Quantitative Methods in the Department of Politics and International Relations as well as Nuffield College at the University of Oxford. His research focuses mainly on the relationship between democratic institutions and state repression. His research has appeared in Political Geography and the American Journal of Political Science. Aysegul Aydin is Assistant Professor of Political Science at the University of Colorado, Boulder and is a research affiliate at the Centre for the Study of Civil War, PRIO, Norway. Dr Aydin works on civil wars, mass killings, international conflicts, diplomacy, and conflict resolution. Her most recent articles include “Diplomacy and Other Forms of Intervention in Civil Wars,” Journal of Conflict Resolution (2006), coauthored with Patrick M. Regan; and, “Diplomatic Interventions and Civil War: A New Dataset,” Journal of Peace Research (forthcoming), co-authored with Patrick M. Regan and Richard Frank. Christian Davenport is a Professor of Political Science at the University of Maryland – College Park, as well as Director of the Radical Information Project (RIP) and Stop Our States (SOS). His primary research interests include human rights violations/state repression, social movements, measurement, and racism. Professor Davenport is the author of numerous articles appearing in American Political Science Review,

xiv Contributors American Journal of Political Science, Journal of Politics, Journal of Conflict Resolution, Political Research Quarterly, Comparative Political Studies, and Monthly Review (among others). He is the recipient of numerous grants (including five from the National Science Foundation and one from the Carnegie Foundation), author of one book, State Repression and the Domestic Democratic Peace? (Cambridge University Press, 2007), and editor of two books: Repression and Mobilization: What Do We Know and Where Do We Go From Here? (University of Minnesota Press, 2004), and Paths to State Repression: Human Rights Violations and Contentious Politics (Rowman & Littlefield, 2000). Currently, one book is under review: The Rashomon Effect: Data Generation and Perspective in Black Panther Party Repression. For more information see www.christiandavenport.com. Scott Gates is Research Professor and Director of the Centre for the Study of Civil War. He has a PhD in Political Science from the University of Michigan. His current research is on civil conflict, post-conflict peacebuilding, and organization theory. He has been an Associate Editor of Journal of Peace Research since 1998. He has received several grants, among others, from the World Bank, EU, Research Council of Norway, and the Russell Sage Foundation. He has published in American Political Science Review, American Journal of Political Science, Journal of Conflict Resolution, Journal of Environmental Economics and Management, and Journal of Peace Research, inter alia. In addition to his three published books, Functions of the Public Executive: Teaching, Managing Tasks, Brokering Trust, co-authored with John Brehm is forthcoming with Russell Sage Press. A volume co-edited with Simon Reich, Child Soldiers. Children in Armed Conflict in the Age of Fractured States is under review. Patrick James is Professor of International Relations and Director of the Center for International Studies at the University of Southern California (PhD, University of Maryland, College Park). James specializes in comparative and international politics. His interests at the international level include the causes, processes, and consequences of conflict, crisis, and war. With regard to domestic politics, his interests focus on Canada, most notably with respect to the constitutional dilemma. James is the author of 11 books and over 100 articles and book chapters. Among his honors and awards are the Louise Dyer Peace Fellowship from the Hoover Institution at Stanford University, the Milton R. Merrill Chair from Political Science at Utah State University, the Lady Davis Professorship of the Hebrew University of Jerusalem, the Thomas Enders Professorship in Canadian Studies at the University of Calgary, the Senior Scholar award from the Canadian Embassy, Washington, DC, the Eaton Lectureship at Queen’s University in Belfast, and the Quincy Wright Scholar Award from the Midwest International Studies

Contributors xv Association. James was Distinguished Scholar in Foreign Policy Analysis for the International Studies Association (ISA; 2006–07). He was Vice President (2005–07) and is President (2007–09) of the Association for Canadian Studies in the United States, and Vice-President (2008–09) of the ISA. James also served a five-year term as Editor of International Studies Quarterly. Stathis N. Kalyvas is Arnold Wolfers Professor of Political Science and Director of the Program on Order, Conflict, and Violence at Yale. He is the author of The Logic of Violence in Civil War (Cambridge University Press, 2006) which received the Woodrow Wilson prize for the best book on government, politics or international affairs published in 2006, and The Rise of Christian Democracy in Europe (Cornell University Press, 1996) which received the J. David Greenstone Award for the best book in the field of politics and history. His work has appeared in many journals, and in five languages. He has received the Gregory Luebbert Award for the best article in comparative politics, a Jean Monnet Fellowship from the European University Institute, and fellowships and grants from the John Simon Guggenheim Memorial Foundation, the Harry Frank Guggenheim Foundation, the United States Institute of Peace, the World Bank, and the Folke Bernadotte Academy. He is currently researching the microdynamics of civil war, with a focus on warfare, recruitment, and violence, using disaggregated data from Colombia and Vietnam, among others. André Lecours is Associate Professor of Political Science at Concordia University. He holds a PhD from Carleton University. His research focuses on nationalism, with a particular interest for the Basque Country/Spain, Flanders/Belgium, Scotland/United Kingdom, and Québec/Canada. He is the Editor of New Insitutionalism. Theory and Analysis (University of Toronto Press, 2005) and the author of Basque Nationalism and the Spanish State (University of Nevada Press, 2007). He also has a forthcoming book (co-authored with Daniel Béland) Nationalism and Social Policy: The Politics of Territorial Solidarity (Oxford University Press). Sarah-Myriam Martin-Brûlé is a PhD candidate in Political Science at McGill University. She has completed her Masters and undergraduate degrees at the Université de Montréal. Her research interests are peacekeeping operations and security issues related to civil war settings within failed or collapsed states. She is also interested in the integration of energy markets in the Americas. She recently co-authored an article with Professor Philippe Faucher entitled “Las implicaciones para Canada de la integracion regional de los mercados energeticos” published in Foro Internacional, 2004, vol. XLIV (2), April–June. S. Mansoob Murshed is at the Institute of Social Studies in the Netherlands and is also Professor of International Economics at the Birmingham

xvi Contributors Business School, University of Birmingham in the UK. He was the first holder of the rotating Prince Claus Chair in Development and Equity in 2003. He was a Research Fellow at UNU/WIDER in Helsinki where he ran Projects on Globalization and Vulnerable Economies and Why Some Countries Avoid Conflict, While Others Fail. He also ran a project on “The Two Economies of Ireland”, financed by the International Fund for Ireland at the Northern Ireland Economic Research Centre (NIERC), Belfast. He is the author of five books and over 75 book chapters and journal articles. His research interests are in the economics of conflict, aid conditionality, political economy, macroeconomics, and international economics. Roger Petersen is Associate Professor of Political Science at the Massachusetts Institute of Technology. Professor Petersen studies comparative politics with a special focus on conflict and violence. He has written two books: Resistance and Rebellion: Lessons from Eastern Europe (Cambridge University Press, 2001) and Understanding Ethnic Violence: Fear, Hatred, Resentment in Twentieth-Century Eastern Europe (Cambridge University Press, 2002). He also has an interest in comparative method and has co-edited, with John Bowen, Critical Comparisons in Politics and Culture (Cambridge University Press, 1999). Donald Rothchild was Professor of Political Science at the University of California, Davis. He was the author of Racial Bargaining in Independent Kenya: A Study of Minorities and Decolonization and Managing Ethnic Conflict in Africa: Pressures and Incentives for Cooperation; coauthor of Sovereignty as Responsibility: Conflict Management in Africa; and co-editor of The International Spread of Ethnic Conflict: Fear, Diffusion, and Escalation, and Ending Civil Wars: The Implementation of Peace Agreements. His most recent book, edited with Philip G. Roeder, is entitled Sustainable Peace: Democracy and Power-Dividing Institutions after Civil Wars. Stephen Saideman is Canada Research Chair in International Security and Ethnic Conflict and Associate Professor of Political Science at McGill University. He has written extensively on the international and domestic politics of separatism, as well as on other dimensions of ethnic conflict. In addition to his book, The Ties that Divide: Ethnic Politics, Foreign Policy and International Conflict, he has published articles in International Organization, International Studies Quarterly, Comparative Political Studies, Journal of Peace Research, as well as several other outlets. His next book, with Bill Ayres, For Kin or Country: Xenophobia, Nationalism, and War will be published by Columbia University Press in 2008. He spent a year on the US Joint Staff working in the Strategic Planning and Policy Directorate on Balkans issues as part of a Council on Foreign Relations International Affairs Fellowship, and he is

Contributors xvii currently the Director of the Montreal Research Group on Ethnic Conflict. Paul Staniland is a PhD candidate in the Department of Political Science and Security Studies Program at the Massachusetts Institute of Technology. He studies civil war and international security, with a focus on the organization of insurgent groups. He has published in Civil Wars, Security Studies and Washington Quarterly, and conducted field research in India and Northern Ireland. Zeynep Taydas is an Assistant Professor at Clemson University. She received her PhD (2006) from the University of Missouri-Columbia. Her research focuses on international and internal conflicts and third party interventions in ethnic conflicts. She is the co-author of a book (with David Carment and Patrick James) Who Intervenes? Ethnic Conflict and Interstate Crisis (Ohio State University Press, 2006), and an article (with Yasemin Akbaba and Patrick James) “One Sided Crises in World Politics: A Study of Oxymoron, Violence and Outcomes” in International Interactions. Marie-Joëlle Zahar is Associate Professor of Political Science at the Université de Montréal. She received her PhD (2000) from McGill University. Her research focuses on non-state actors in civil wars, civil conflict resolution, and post-conflict reconstruction. A recipient of grants from the Social Sciences and Humanities Research Council of Canada (SSHRC), the John D. and Catherine T. MacArthur Foundation, and the Carnegie Corporation, she was a research fellow at the Center for International Security and Cooperation, Stanford University (1998–2000). Her forthcoming book, Liban: les défis de la paix (forthcoming at Éditions Autrement/Centre d’études et de recherches internationales, Sciences-po (Paris), 2008) analyzes the impact of foreign intervention on the reconstruction of state institutions in post-war Lebanon.

Acknowledgments

In the course of this project, we accrued a great many debts to those who helped us along the way. First, we would like to thank those that funded the conference and seminars in Montreal in 2005 that sparked this volume. The Research Group on International Security, a collection of scholars at McGill and Université de Montréal and funded by the Security and Defence Forum (SDF), provided the starting capital that made the rest of the project feasible. SDF, a unit of Canada’s Department of National Defence, provided additional funding as did the John Holmes Fund from the Department of Foreign Affairs. These contributors covered the lion’s share of the conference’s costs, and funds from McGill covered the rest via the Department of Political Science, the Faculty of Arts, the Montreal Research Group on Ethnic Conflict and the Vice President for Research, as well as the Canada Research Chairs program. Second, at the original workshop and subsequent seminars, we received very helpful comments. As a result, we are quite thankful to Jean-Paul Azam, Ravi Bhavnani, Tamra Pearson D’Estree, Major General (ret.) William Nash, Don Hubert, Colonel Denis Thompson, Robert GrossmanVermaas, Mark Brawley, Amy Cox, Michel Fortman, Michael Koplove, Spyridon Kotsovilis, Matthew Lange, Juliet Johnson, Michael Lipson, Oana Tranca, Brian Rathbun, Norrin Rippsman, Oskar Thoms, Laura Tuca, and Suranjan Weeraratne for their insightful suggestions. We also received much useful feedback when parts of the volume were presented at the 2006 meeting of the International Studies Association in San Diego, particularly from our discussant: Will Moore. Third, our research assistants made a tremendous contribution to this effort during the planning of the conference, during the conference itself and afterwards in the revising and editing of the volume. Sarah-Myriam MartinBrûlé was indispensable during the early stages of the project, helping to plan and run the main event. Amy Cox and Suranjan Weeraratne were very helpful during the middle stages of the project, along with Vania Draguieva and Claudia Martinez Ochoa. Ora Szekely and Lauren Van Den Berg did a tremendous job at the end of the project, doing much of the unpleasant detail-oriented stuff with good cheer and much dedication.

Acknowledgments xix Finally, we owe the contributors of this volume much thanks for their insights and the collegiality throughout the process. We learned a great deal throughout this project, with the contributors helping us to see the world from a variety of new and interesting angles. One, in particular, made a huge difference not only in our project but also in our careers. We both encountered Donald Rothchild early in our careers, finding ourselves lucky enough to be involved in volumes he edited. In each case, Don showed us how to be a good scholar and a great person. Unfortunately, he passed away while this volume made it through the publishing process. We lost not only a smart student of civil conflict, but also a good friend and amazing colleague. It is not only our loss but a blow to our profession. This will not be the last book dedicated to Donald Rothchild: friend, teacher, mentor, scholar, and, well, mensch.

1

Causing security, reducing fear Deterring intra-state violence and assuring government restraint Stephen M. Saideman and Marie-Joëlle Zahar

The 1990s saw a vast expansion of effort by a variety of actors to prevent, manage and/or resolve conflicts around the world, including Bosnia, Kosovo, Macedonia, Cambodia, Liberia, Rwanda, Sierra Leone, East Timor, and elsewhere with mixed results. The first years of the new millennium have seen more missions in even more apparently hopeless places such as Afghanistan and Iraq. The primary mission for the military inserted into these situations is to create a “safe and secure environment”1 so that the rest of the peace implementation (or nation-building) processes can take place. What seems to be relatively unproblematic in Bosnia in 2007 is extraordinarily difficult in Afghanistan and Iraq. While the situations in these three countries are most dramatic, all countries face the problem of providing security to their populations.2 This is the basic task of government, but it is not an easy one. In practice, a key complication is that governments not only provide security to their people but also serve as perhaps the most dangerous threat. Abuses by governments can threaten individuals and groups far more powerfully than their opponents, including dissidents and insurgents. The universal problem that is the focus of this book is that governments must provide security to their citizens but also remain restrained enough that they do not become the primary threat. We address both the domestic and international dynamics involved as the internal balancing act can be upset by outsiders, and external actors (including the United Nations, the North Atlantic Treaty Organization, the African Union, individual countries, and other organizations) have increasingly played a role in restoring the internal security of countries. The problem is that security, as both a concept and a reality, is poorly understood, and particularly in these contexts as the reigning approach, the ethnic security dilemma, largely assumes away the most relevant actor – the government. The starting assumption of this volume is that security is fundamental – that conflict within states has security as a key piece of the puzzle, and that governments are central. Scholars have argued that other factors motivate conflict. World Bank economists, among others, have argued that greed in some form drives conflicts within states (Collier 2000; Collier and Hoeffler 2001). Some political scientists have argued

2 S.M. Saideman and M.-J. Zahar along similar, but not identical, lines of thought that those seeking domination – predators – rather than security are the key actors (Snyder and Jervis 1999). Increasingly, scholars have focused on insurgency, which directs attention away from security of the group and focuses on the weaknesses of states and the tactics and strategies of small numbers of individuals. However, these and other lines of argument ignore a crucial dynamic – that for any of these processes to play out, group members beyond its elites must give support.3 Likewise, for stability to develop after a conflict, individuals and groups must support the peacekeepers and new government: they must deny opponents assistance and provide the new security providers intelligence.4 Below, we consider each argument directly and identify the role security plays, implicitly or explicitly. In the past dozen years, the primary way in which scholars have thought about security in the context of civil wars has been the security dilemma (Posen 1993; Snyder and Jervis 1999; Rose 2000). They have taken a concept from International Relations (Herz 1951; Waltz 1979; Jervis 1978) and applied it to the context of civil wars and ethnic conflict. There is much to be gained, as the key insight from the security dilemma is quite important – that unilateral efforts to make oneself more secure leads others to try to do the same, ultimately leaving all less secure. The challenge is that the translation from the international context to the domestic one has been problematic (David 1997). In particular, the security dilemma best applies in the absence of government, as it was originally intended – in the relations of sovereign states in a world without a central government. In relations between substate units (ethnic groups, for example), the context is one of hierarchy – where the government exists in some form, playing a crucial role in ameliorating or exacerbating the violence. Indeed, one of the essential roles of government is to deter violence – to possess a monopoly of the legitimate use of force and to wield it so that individuals and groups refrain from engaging in serious conflict. This can be quite difficult, as a delicate balance must exist between deterring violence and repressing dissent. Because the state possesses much of the coercive capacity in most states, it is the greatest threat to most groups. Indeed, in most cases of genocide, the perpetrator is the state (Krain 1997; Rummel 2000; Midlarsky 2005). This leads to two dilemmas that reinforce each other. The first is that the state, to prevent violence, must be sufficiently threatening to deter potential opponents while accommodating enough dissent so that frustration does not build into challenges. This dilemma is most clearly relevant pre-conflict, particularly in existing studies, but it also matters in the postconflict peace-building process, as the new institutions must be capable so that past rebels do not have an easy way out of the political process. The second is that the state must be assertive enough to protect any group, but not seen as too favorable toward anyone or else groups will compete to gain control of the state, creating a political security dilemma. This

Causing security, reducing fear 3 dynamic has largely been studied from the perspective of post-conflict design of institutions, but is also clearly significant for the prevention of conflict. The political security dilemma is that any effort to gain control over the state by any group will create doubts about its ability to protect and deter, leading to counter-efforts to control the government. This reinforces the initial instincts of the first group, leading to a spiral of competition, ultimately reducing the ability of the government to deter conflict (Saideman 1998). This type of interaction suggests that it might be more appropriate to think of security not so much as a good – the presence or absence of which defines rules of interaction – but as a continually renewed bargain between the state and constituent groups. In this volume, we seek a better understanding of security’s role in intra-state conflicts ranging from peaceful dissent to outbursts of violence to protracted civil wars. Perceptions of insecurity are central to the eruption of violence; securitization is necessary to restore sustainable peace. While other actors can sometimes step in to provide security on a temporary basis, the state remains the main provider of security, and outsiders can only depart if the state can provide security to all (or, at least, to most). State strategies are central to situations of war and peace alike. Although we focus on security, we approach it by applying a different set of concepts from International Relations – applying deterrence theory to the domestic context – while remaining sensitive to the differences between intra- and inter-state politics. The operating questions guiding this work are: • • • • • •

Under what conditions can the state deter violence? What strategies are best suited to this effect? Under what conditions is the state the most significant generator of violence? Under what conditions can the state assure the citizens that it is not a threat and cannot be captured by a particular group? How do external pressures affect the deterrence-assurance dynamics within states? Is the challenge of providing security different on the morrow of violent conflict? Can external actors step in to provide security should the state be unable or unwilling to do so?

Each chapter takes a different perspective on one or more of these questions, as the concepts of security and fear are inherently multi-disciplinary. The contributors come from a variety of fields, areas of expertise, methodological approaches, and theoretical orientations, and they consider a wide variety of cases, ranging from advanced democracies to failing autocracies. The contributions share a focus on security, doubts about existing understandings, and fresh approaches to the questions at hand.

4 S.M. Saideman and M.-J. Zahar The point of this book is not to throw out the baby with the bathwater – security is crucial to our understanding of intra-state violence, even if the application of security concepts from International Relations have, thus far, been flawed. Instead, we recast security as a central element of the puzzle of intra-state war and peace. By focusing on the strategies that yield more or less security for groups, we also bring the state back in. Indeed, we argue that government strategies and actions are fundamental to understanding outcomes. We also seek to identify opportunities for and constraints on the state’s ability to provide security to its constituent groups. In short, this contribution proposes an integrated approach to understanding the causes of civil war, civil peace, and situations in between these two extremes. Just as the study of interstate war needs to focus on the outbreak of war and the existence of peace, our effort here addresses both sides of the question, which is particularly important since civil and ethnic peace of various forms is far more prevalent than war. In the next section, we consider some of the most prominent accounts of civil conflict, indicating that they tend to overlook the security dynamics inherent in the processes that they consider. The following section addresses the security dilemma, the most common application of a securitybased logic to the topic at hand. We then delineate an alternative – that deterrence logic can help us understand the dynamics of intra-state violence. We then develop some of the implications of the logic of deterrence. The chapter concludes with a discussion of scope of the volume, the contributions and how they fit into the deterrence framework we sketch.

The role of security in theories of civil conflict A variety of interests may motivate conflict, ranging from prejudice to greed, from insecurity to hatred. In this section, we consider three approaches to understanding the causes of civil conflict, as each has implications for how such conflicts should be resolved.5 While there are many theories, we focus on three approaches: elite politics, greed, and insurgency. Some of these take security more seriously and more explicitly than the others, but a key question to be asked of all is: How does the rest of the population fit in? Efforts to mobilize conflict require support beyond a narrow cadre. One of the key problems in much of the work in this area is that the links between masses and elites are largely assumed or ignored. The thinking here is that the level of insecurity plays a key role in mediating the relationship between state and society and between elites and the population. Nationalist elites First, it has been commonly argued that elites promote conflict, particularly ethnic conflict, to divide opponents and divert domestic audiences

Causing security, reducing fear 5 away from issues that politicians want to avoid, like responsibility for a failing economy. Gagnon (1994/5, 2004), one of many, argues that Slobodan Milosevic used Serb nationalism to maintain his position, despite a collapsing economy. Snyder (1999) argues that this kind of dynamic is likely to play out during democratization processes as incumbents seek to hold onto their positions. While this can be opportunistic, as elites see an opening, it can also be compelled, as political competition may force politicians to take stronger and stronger stands on behalf of a group – outbidding (Rothschild 1981; Horowitz 1985). While elites can try to create divisions based on identity or other cleavages, they are limited by the support they receive. Politicians are not magicians and cannot dupe a large part of the electorate6 easily. That is, for leaders to be successful, they need followers. The question of under what conditions will extremist politicians be successful is poorly understood. The case of Yugoslavia is instructive as Milosevic was able to point toward the new leadership in Croatia, and vice versa (Gagnon 2004). Because individuals had some reason to fear what might happen in the future, they threw their support to a politician promising to defend their interests. Thus, we need to take seriously what causes individuals and groups to support politicians promising rather risky policies and potentially dangerous futures. Our hunch here is that extremist politicians are more likely to receive support if the individuals and groups perceive the context to be somewhat dangerous (Saideman 1998). That is, if the government is seen as unable to deter threatening actors, or, worse, is vulnerable to capture, in which case, the state becomes the threat. De Figueiredo and Weingast (1999) develop a game theoretic argument, asserting that voters will choose more extreme candidates if they possess only a modest degree of fear that failing to do so might risk domination or extinction. The calculation is that a modest chance of a horrible result multiplied by said outcome is greater than “risking” moderation. This rational logic coincides with findings from cognitive psychology. Prospect theory holds that individuals weigh more heavily possible losses more than potential gains, so they are willing to gamble more to avoid losing.7 Therefore, we should expect that constituents are likely to support politicians promising to defend them even if the threat is relatively unlikely as long as the consequences are severe – which they are if the state is potentially up for grabs. In addition, sociologists have long argued that people are more likely to support their group if they perceive an external threat (Coser 1956). Thus, the key to understanding the success of extremist politicians is the level of threat perceived by potential supporters. Therefore, we must understand what causes individuals and groups to be secure enough to ignore such appeals and what causes these folks to be insecure enough to follow extremists.

6 S.M. Saideman and M.-J. Zahar Greed As the century turned, the discourse shifted to focus on greed – that leaders of insurgents, either as their primary goal or as necessity demanded, were driven by the existence of easily lootable resources. Spurred by a team of World Bank scientists led by Paul Collier, this debate focused on the necessities of rebellion.8 That is, it takes resources to finance violence against a government – to buy arms, to encourage followers, to bribe government officials, etc. This argument helped to make sense of the roles played by “conflict diamonds” in Africa, opium in Afghanistan, cocaine in Columbia, and other “easily lootable” resources elsewhere. However, recent studies (Ross 2004) suggest that while the existence of such resources may prolong civil wars, they are not necessary for their outbreak. Thus, this line of thinking provides some clues but also some questions for the topic at hand. While something else might trigger conflict, it may be that the proceeds from lootable resources may help to buy, or, at least, rent support from relevant populations. Even though small core groups can conduct much violence, they require support from the locals. Lootable resources may be used to gain the necessary support from the indigenous population. However, there is a problem – that resources are finite and needs are many. While leaders of a rebellion can use drugs, minerals and other “capital” to buy arms and pay off the inner circle, these assets are not without limit, so buying off entire populations seems unlikely to be a winning strategy. Indeed, Weinstein (2007) demonstrates that access to resources tends to lead to strategies that are less successful. Instead, coercion may be applied to the wider audience of this group. This raises a problem that insurgency theorists address more directly (see below) – the competition between government and opponents to provide security to the relevant populations. If a population’s security is largely guaranteed by the government, then the opponents will appear to be more like criminals as they run drugs, steal and sell minerals, and the like. If the government cannot provide security, and particularly if it is seen as a threat, then the criminal activities of the dissidents may be viewed as noble, worthy of support. Thus, it comes back to the state – can it deter conflict or is it a generator of violence? Insurgency Fearon and Laitin (2003) have argued that opportunity essentially drives civil war. That is, insurgency is the preferred strategy for fighting against a government, given the imbalance of forces. Governments may tolerate a little rebellion, and the ease of repression shapes how much violence they are willing to tolerate. Geography plays a role here as mountains and other terrain features that make counter-insurgency difficult increase the probab-

Causing security, reducing fear 7 ility of conflict. Level of economic development plays a large role, as individuals in richer states are more resistant to recruiting, and such states are more capable. This chapter, along with the events in Afghanistan and Iraq, suggests that insurgency presents a competitive dynamic between the rebels and the government – which side best protects and/or most severely threatens the relevant populations (Kalyvas 2006). Again, insurgency requires some level of popular support – so that the rebels can hide, rest, train, plan, and launch attacks. The government needs the public’s support to deny their opponents exactly this kind of refuge. As some economists have already noted (Grossman and Kim 1996), this creates a situation where there are, in essence, dueling protection rackets, as the combatants try to both persuade and compel the civilians in the middle to take their side (Kalyvas 2006). We need to know how such duels are resolved – what kinds of strategies can the government employ to assure and deter individuals so that they decrease support for the insurgents?

Security dilemmas Scholars of International Relations have sought to understand civil war, and particularly ethnic conflict, by applying the concept of the security dilemma to ethnic conflict. In International Relations, there is no higher authority than the state – that is, there is no world government. States, therefore, must depend upon themselves for their own security. The problem is that what one state does to enhance its own security may be seen as a threat by another. This second state, because of the uncertainty of the situation, is likely to respond by increasing its own efforts to be secure (more weapons, more allies, etc.). This, in turn, increases the insecurity of the first state. The dilemma is that any effort made by either to increase its security ultimately fails (Herz 1951; Waltz 1979; Jervis 1978). Posen (1993) was the first of several scholars to apply this idea to civil war.9 He argued that as empires and states collapse, groups must worry about what others might do. This fear leads to pre-emption as one or more groups will try to act first, before they themselves are attacked. Posen is most concerned with the translation of military variables from the International Relations dynamic to the domestic setting, so he is concerned about whether it is relatively easy to attack or defend (whether the offense or defense is dominant) and whether this is relatively straightforward to discern. He then turns to demographics to determine this, as he argues that highly intermixed groups are very vulnerable and hard to defend, whereas concentrated groups are easier to defend. Because it is hard to distinguish offensive capabilities from defensive ones in intermingled contexts, there is a strong temptation to attack first, to pre-empt, since one is so vulnerable. This view of the ethnic security dilemma focuses on a strictly physical notion of security – vulnerability to armed attack.

8 S.M. Saideman and M.-J. Zahar This approach has become one of the most prevalent ways in which scholars, particularly those whose primary field is International Relations, think about civil war and ethnic conflict. Unfortunately, this view suffers from at least two significant problems. First, its take on demographics is simply wrong – highly intermixed groups are much less likely to fight, as nearly all quantitative studies show.10 Thus, a key causal mechanism for the outbreak of violence is broken. Second, and, more importantly, the approach assumes a failing state and goes from there (David 1997). It essentially portrays ethnic conflict as primarily between groups rather than between a group and a government.11 This should restrict the scope of the ethnic security dilemma significantly. More importantly, it overlooks the causes of failed states, which might also be causing the subsequent violence. Thus, the dynamics that Posen and others see as critical may be spurious, as something else may be driving both the failures of states and the resulting violence. Trying to apply the security dilemma more broadly, Saideman (1998) considers it not merely as a matter of competition after state failure, but as a cause of imperial collapse, as groups compete with each other not just in the absence of the state but for control of an existing state. The idea is that the government is potentially the key threat to any group, as governments are usually the actors that commit genocide. Whoever controls the government is protected from extermination and those who are excluded from government are vulnerable to destruction. The case of Rwanda is instructive as the genocidaires held the commanding heights of governmental power – the Presidential Guard, the army, affiliated militias, etc. – and launched the genocide when their control was at risk. The problem in Yugoslavia was not the absence of government but precisely which government would exist and retain the state’s capacity to repress. While this revision helps us to focus more on the role of the state, it still borrows from the same strand of International Relations theory as Posen and others. Perhaps by considering an alternative approach, we can get a better grasp of the role of the state in causing or deterring conflict and the variation of strategies used by governments to maintain peace. This discussion should, as events in Iraq and elsewhere do, raise the dual issues of exit strategies and “Hearts and Minds.” To build support for peace and authority, we need consider how to provide enough security for individuals and groups so that they demobilize. Security here refers to both secure from attacks from the regime’s opponents and secure from abuses by the state. While outsiders may be seen as providers of credible guarantees (Walter 1997), real concerns exist about the willingness of international organizations and states to bear significant costs and their own desires to get out. That is, in any peace operation, there is not only the reality that the countries donating troops and organizations involved have to think about when to leave, but there is always a genuine desire to depart. The political necessity of having exit strategies and time limits

Causing security, reducing fear 9 tends to weaken the credibility of the interveners, undermining their ability to deter potential spoilers. Indeed, in most countries contributing troops to the missions in Afghanistan and Iraq, there is considerable controversy about the length of the mission, raising questions about how long they will remain. Further, as Somalia and Rwanda testify, killing a few soldiers may be a workable strategy for actors on the ground to immobilize peacekeeping operations.12 Again, the key dilemma for states and peaceenforcing efforts is how to be powerful enough to deter but accessible enough to assure.

Deterring and assuring: a difficult balancing act The limitation of even the revised ethnic security dilemma is that it tends to focus on group competition, including struggling for control of the state, at the expense of the state, and its dual roles – in deterring violence and as the key source of threat. Instead, this section, and, indeed, this book, focuses largely on the role governments and their substitutes (peacekeeping operations, nation-building efforts) play in threatening those who might engage in violence while restraining themselves from harming everyone else.13 This focus on the state provides a unifying element that allows us to build bridges between analyses of the causes of inter-state war and studies of conflict resolution and peace implementation. Instead of focusing solely on war-torn countries, it allows us to cast the debate in wider terms. Indeed, while most states succeed in striking the difficult balance between deterrence and assurance, others constantly teeter on the edge of disaster, and others still fail. What determines success and failure? Some might suggest that it is regime type (Snyder 1999). Democracies may be good at assurance by providing access to the political system and transparency, but may be less good at deterrence as complex decision-making procedures may make it difficult to engage in policies that are costly and only work in the long term. Authoritarian states may excel at deterrence since it is easier for such regimes to use force, making their threats effective. However, they cannot assure as well since, in general, autocrats are not bound by institutions that might constrain their behavior. Anocracies face a particularly difficult challenge as they can neither deter nor assure particularly well. It may have to do with state capability (intelligence, economic well being etc. discussed above). On the other hand, the regional environment can provide constraints and opportunities. Instead of focusing only on states where violence has erupted, the proposed framework deals both with the dog that barks and the one that does not. From this perspective, it is not useful to single out failed and fragile states; rather we compare the constraints and opportunities faced by such states as they devise deterrence and assurance strategies with those of stronger and more capable states.

10 S.M. Saideman and M.-J. Zahar Before considering how deterrence works, we need to address the starting point: who and what is to be deterred? First, regime opponents or potential rebels can be broken down into several categories: those that merely need to be assured that the government will not abuse them; those that might consider violence if it is easy and unlikely to produce much of a response from the government; those that are likely to engage in violence unless the government is quite capable; and those that would be, essentially, undeterrable.14 The required balance of deterrence and assurance will vary, depending on the types of opponents present.15 What kinds of behavior is the government seeking to deter? To create a safe and secure environment, the state must deter groups from engaging in violence against each other. Wilkinson (2004) illustrates this quite nicely as ethnic riots in India are a significant threat, particularly to minorities, and governments at the regional level have the ability to prevent or limit such violence. The government must also prevent violence against itself as this not only spills over to the population, but also risks the capture of the state’s resources by one segment of society. Finally, the state needs to deter separatism, not just for the usual reasons of territorial integrity (Toft 2003), but also because there are usually minorities who do not want to depart along with the rest of the seceding region. Further, secession alters the balance of political power in the rump state, increasing the insecurity of some or all of the remaining groups (Saideman 1998). Indeed, Fearon (2004) has shown that separatist conflicts are the most enduring. The question then becomes under what conditions can the state deter violence, whether it is aimed at the government or at members of society? State as deterrent Deterrence theory tells us that several elements must be in place for an actor to be dissuaded from engaging in a preferred course of action:16 that the state can impose significant costs upon transgressors; the threat is credible; and that the status quo is possibly acceptable. First, the state must be capable enough to hurt violators, if need be. This requires a military and/or police force that has enough personnel, training, and equipment to pose a significant threat. If police or the army is under-manned, then potential rebels can find more opportunities to engage in violence at low cost.17 If the government’s forces are poorly trained, they will be easier to defeat. Weakness on the part of the government alters the deterrence equation, making successful defiance seem more likely and less costly. This is widely recognized in research on failed and fragile states which argues that fragility facilitates the eruption of violent conflict (Zartman 1995). Thus, governments with stronger economies (higher gross domestic product per capita) tend to face less conflict. One causal mechanism tying economic strength to civil peace is greater state capability.18 This is also the assumption behind state-building as a tool for conflict

Causing security, reducing fear 11 resolution. For conflict resolution to be long lasting,19 analysts and practitioners must address the weakness of the state. In other words, state building is fundamental to successful war-to-peace transitions. Second, potential rebels are unlikely to be deterred unless they believe that the government will follow through and use its capabilities. In the International Relations literature, there is much discussion about credibility and resolve.20 As difficult as it gets there, the problem of credible commitment to using force against one’s own population is much more complicated. Domestic politics matters, as some potential opponents may be allies of members of the ruling coalition. A government is unlikely to use force against a group if some of the group’s members are involved in making the decision. One of the problems with power sharing, as Donald Rothchild discusses in his chapter, is that it makes it exceptionally hard for the government to commit credibly to any forceful action. Including all of the major actors in a conflict in the government may assure each group, but it also limits the ability of the state to deter violence. Third, for deterrence to work, the status quo must be at least minimally acceptable. If the likelihood of successful rebellion is quite low and the threat imposed by the state is both credible and severe, then potential insurgents should be deterred – unless the present is so unappealing that the risks are worthwhile. This side of the deterrence equation is often ignored, but if the choice is between genocide and a rebellion that is unlikely to be successful, then the rational choice would be rebellion (De Figueiredo and Weingast 1999). In addition, the more appealing the status quo, the less threatening the state has to be to deter conflict. André Lecours’s chapter on the waning appeal of the Basque separatist organization ETA provides a good illustration of this type of situation. We, therefore, have three key components of deterrence within states that might usefully frame our analyses of the outbreak of violence and the requirements of sustaining a peace: state capacity, credibility to use force; and an acceptable status quo. Of course, as will become clear below, some of the requirements of deterrence tend to limit the ability of the state to assure the population that the government itself is not a threat. The state as threat While a threatening state may be quite good in posing significant costs to potential rebels, it also may give people something to rebel about. It bears repeating that governments are the most frequent perpetrators of genocide (Rummel 2000; Krain 1997). Repression may work in the short run, as it raises the cost of dissent. But dissent delayed can be much stronger and more violent.21 The more a government threatens its population, the more reason its citizens have to fear it and try to overcome it or leave it. So, governments need to assure their citizens that the capabilities of the state will not be abused and will not be used against those who are sufficiently

12 S.M. Saideman and M.-J. Zahar loyal.22 This requires good intelligence and discriminate targeting of rebels (Kalyvas 2006) as well as constraints on the state so that its power is not abused. As events in Iraq demonstrate on a daily basis, unless one knows who the insurgents are, the government (or occupier/peace-enforcer) risks sweeping up not only the rebels but innocent bystanders as well (Ricks 2006). This will antagonize those who did not otherwise have a significant grudge against the state. Of course, insurgents understand this, and do their best to mix in with the population to encourage the government’s application of force against non-combatants. The government then faces the choice of indiscriminate targeting, which may increase the threat it poses but at the expense of the status quo, or investing more effort into intelligence. Often governments choose the former, which is the road toward genocide or politicide (Valentino et al. 2004). The method of punishing entire populations (of villages, of regions) for the crimes of a few insurgents is no longer viable, at least not for democracies. Therefore, we need to think more seriously about this problem – how to identify opponents and how to use force discriminately.23 State capability might be a factor in this equation as intelligence may play a role in securing greater knowledge of the potential opposition. Otherwise, the government will have to be less discriminate in its use of force – which may still work to impose costs – but reduces the appeal of the status quo. The second challenge is to restrain the state itself. Rather than “merely” using violence indiscriminately, elements of the government may deliberately use violence against the citizens. While democracies may not always fully control or restrain their militaries, one of the defining characteristics of a democracy is civilian control over the military.24 We ought to expect democracies, on average, to do a better job of restraining their coercive branches, if only because the people have a voice and, if threatened, will choose new administrations that do a better job of protecting the people from the armed forces. Aydil and Gates, in their contribution to this volume, consider how democratic institutions restrain governments, preventing or limiting mass killing. Even if the government controls the military adequately, one still needs assurance that the government will not choose to use its coercive capacity against its political (non-rebellious) adversaries. Institutions can help, as they help to provide credible commitments of restraint. For instance, proportional representation is likely to increase the political power of smaller groups, compared to other electoral systems, making the government less likely to antagonize potential coalition partners.25 Power sharing is also seen as a way to limit the ability of a government to harm portions of the populace. Of course, as mentioned above, constraining the state may limit its ability to deter violence.

Causing security, reducing fear 13 Dilemma: deterrence versus assurance The problem is that the means to assure may weaken the tools to deter and vice versa. To prevent violence, governments may opt for a mix of strategies, institutions, and efforts to simultaneously assure and deter, or they may opt for more of one and less of the other, given the difficulties of balancing the two. Figure 1.1 illustrates the tradeoffs. At this point, we do not want to pre-judge and assert that a particular strategy or specific mix of assurance and deterrence produces less violence than another. It seems eminently possible that a strict focus on coercion might keep the peace for an extended period of time. Similarly, we can imagine political systems without violence even if there is very little deterrence going on. Most countries will be in between these extremes, and are likely to shift in their balance of strategies in response to changing circumstances, such as refugee flows or electoral outcomes. This discussion thus points to the real tradeoffs faced by most countries. The security dilemma versus deterrence This book is an attempt to re-cast how we think about security in the context of civil war and peace. As the security dilemma has dominated the discourse thus far, in this section, we contrast it with the deterrence approach presented here. First, as we discussed above, there is no state in the security dilemma, but there is in the deterrence approach. The government exists in nearly all societies, so the security dilemma is essentially restricted to failed states. It particularly is poorly suited for addressing developed democracies, whereas the deterrence approach can account for changing dynamics in such places as Spain, as Lecours demonstrates in his

Pure coercion: Ex. Nor th K orea P eace Deterrence Pure assur ance: Ex Costa Rica

Civil w ar, failed state

Assur ance

Figure 1.1 Deterrence versus assurance.

14 S.M. Saideman and M.-J. Zahar chapter. Consequently, the deterrence approach is useful for understanding the outbreak of severe conflict, the ending of conflict, and the absence of civil war. Second, there is little room for the successful deployment of threats, only promises, in the security dilemma. The security dilemma is built on the logic of the prisoners’ dilemma (Jervis 1976), where threats do not work, but instead produce counter-productive reactions. In the deterrence approach, threats do work to dissuade the other side. Given the reality that governments and peace-keeping operations frequently use threats to prevent or stop violence, the security dilemma’s omission is a serious one. Indeed, deterrence provides a logic for what peace-keeping operations are trying to do.26 Third, this approach is largely agnostic about the particular stakes involved. There are currently largely separate literatures on ethnic conflict and civil war. Our approach can serve as a bridge as the dilemmas we discuss are present in ethnic secessionist conflicts, ideological revolutions, and predatory coups d’etat. Indeed, one of the key insights of recent work (Kalyvas 2006) is that the fight on the ground in the provinces may not be the same as that at the elite level. Our take on that dynamic here is that once deterrence breaks down at the national level, there may be many kinds of disputes let loose at the local level. Fourth, the dilemma is inherently dynamic. Rather than focusing on the absence or presence of hierarchy or anarchy, it allows the government and potential opponents to respond to each other, influencing the capabilities and resolve of each. For any government facing possible violence (rebellion, communal rioting, coups, etc.), there is a constant effort to develop the right mix of coercion and assurance. Understanding this process is crucial for understanding when individuals and groups will be secure or fearful. Finally, this approach captures the dual role of the state as hero and villain, as savior and killer. Every government has responsibility for maintaining peace within its boundaries, but there are many ways to avoid unrest, including violently repressing the entire society. Much of the literature on civil violence focuses on either the government as threat or the state as the necessary peacekeeper. By considering both faces of the government, we both avoid the normative trap of considering one side or the other to be good or bad, and get a better grip of the difficulties facing most states. Implications of deterrence When should we expect violence? Returning to the basics, we need to focus on capability, resolve, and the status quo. We ought to expect violence to break out when the state’s capacity declines. Thus, external shocks, such as wars, refugee flows, and economic dislocations, may provide trig-

Causing security, reducing fear 15 gers to violence by weakening the ability of the state to deter violence. Similarly, such events may also increase the capabilities of rebels. The collapse of the Albanian government in the wake of failed pyramid schemes led to the flow of arms to Kosovo, helping to accelerate the development of the Kosovo Liberation Army. Domestic dynamics can also alter the relative capabilities of the government and dissidents. For instance, the earthquake in Armenia in 1988 weakened the regional government. Elections, repression, and other political dynamics can also alter what the various actors can and cannot do. The willingness to use force discriminately may change over time, as well as interest in assuring the population through compromises and restraint. Elections and electoral competition may cause some domestic actors to permit conflict in some cases and deter violence in others (Wilkinson 2004). Lecours, in his chapter, shows how the two main parties in Spain differ in their attitudes and strategies toward the Basques, so we ought to expect not only government policies varying over time, but also violence as the Basques react to these changes. Likewise, in authoritarian regimes, the threat and occurrence of coups may change not only who governs but also their interest in limiting violence. The third part of the equation – the status quo – sounds static, but is not. As time passes, the relative point of comparison changes, as actors consider what the possible future might look like, given their various choices, compared to that moment. Thus, the deterrence/assurance balance may tip one way or another as the situation improves or declines. In Romania, for instance, it may be far easier for the government and potential dissidents to restrain themselves now than ten years ago, as the economy has improved significantly. Politicians have learned that losing power does not mean either a death sentence or a ticket to obscurity, but perhaps merely a timeout from office-holding. This deterrence/assurance approach is, ultimately, a framework which highlights certain tradeoffs, but, by itself, does not provide determinate predictions. It helps to orient how we may approach the problems of civil war, reconstruction, and peace, but the dynamics described here are incomplete. In the rest of the volume, the contributors will focus on various aspects of the problem of maintaining security and peace, providing more depth to the framework in many cases and challenging it in others.

Fitting the pieces together Our volume has both a narrow and broad scope. We focus on only one aspect of human security – the threats posed to the lives of citizens by governments, opponents, and outsiders. The book, however, considers the full range of possibilities from relatively stable and advanced democracies to transitioning political systems to fragile autocracies. The chapters include

16 S.M. Saideman and M.-J. Zahar both case studies and quantitative analyses, with both the former covering European and African cases while the latter includes data from around the world. In this section, we preview the rest of the volume, delineating how the various chapters contribute to the deterrence/assurance framework. Kalyvas sets the stage by revealing the limitations of the security dilemma approach, showing that deterrence accounts for the behavior of groups in conflict better. He also raises another dynamic – that of revenge. In the absence of deterrence, actors react, not to fears of possible spirals of insecurity, but to past transgressions. Thus, Kalyvas demonstrates the importance of deterrence. Armstrong and Davenport follow up by justifying our interest in the assurance side of the equation, as they demonstrate that governments are the greatest threat to peace – that more people die from governments than from other causes of violence. In their chapter, Armstrong and Davenport argue that the focus on counter-insurgency and rebellion is largely misplaced, summoning a variety of data to prove their point that governments are far more lethal than any other actor. While the two chapters disagree about the relevance of counter-insurgency, they concur on the importance of the dynamics at the center of this volume – the role of the government and of deterrence and assurance. The next question then is what influences the ability or failure of states to deter violence? Murshed provides an economic perspective, arguing that internal and external inequalities significantly influence governments’ ability to deter and assure. Extreme inequalities not only undermine state capacity but also influence potential rebels’ perceptions of the acceptability of the status quo. Levels of development also influence how actors perceive the short and long term, with short-term minded thinking making deterrence and assurance both quite difficult. Aydil and Gates, in their contribution, present the dark side of the state – as genocidaire. They examine the conditions under which the state realizes its capacity to be the greatest threat to its population. What happens when the state not only fails to provide some assurance, but actually attacks its citizens? Aydil and Gates focus on institutions as restraints, arguing that particular dimensions of democracy, not the entire basket, matter in reducing the likelihood of state-driven violence. Petersen focuses on a key actor in the deterrence/assurance equation – the military. He considers under what conditions militaries act as causes of violence or as restraints. He focuses on the demographics of militaries and their societies, arguing that the ethnic composition of each is key. He goes further to focus on recruitment as a means of balancing or exacerbating existing tendencies, finding that the combination of a military’s current composition and its recruitment strategies go a long way in explaining whether the key coercive arm of a government is a force for peace or for conflict. Lecours presents an interesting case – that of the Basques in Spain –

Causing security, reducing fear 17 where violence has continued despite a rather balanced Spanish effort at both deterring violence and assuring its citizens that it will act discriminately. This chapter shows some dynamics when security is actually rather high – that violence can still occur but that it will be rather marginal. That is, the Basque terrorist group – ETA – is small and narrowly supported, even among the Basques. Their use of violence is now mostly symbolic – to show that their cause exists and that their group exists. If Spain had been less competent over the past decade or so in providing a relatively security environment for Basques and other citizens of Spain, one could easily imagine ETA receiving more support and engaging in more violence. Thus, this case provides an interesting contrast – more violent than one would expect from the security situation but less violent than one might otherwise have expected. It does show that some individuals cannot be deterred, but that these are few, and it is difficult to engage in significant violence without more support. Donald Rothchild assesses different institutional strategies for managing the deterrence/assurance dilemma. Specifically, he argues that powersharing mechanisms weaken states’ abilities to deter conflict by providing too much assurance. By paralyzing states via institutional design, powersharing systems limit the ability of the state to respond. By contrast, power-dividing systems preserve the ability of the government to make important decisions while devolving much of the terrain over which policy is fought to lower levels, and thus assuring the populace. The following chapter, by Akbaba, James, and Taydas, shows that the world outside the state in question, beyond just the neighborhood, matters deeply. They document how the existence of external allies for dissident groups upsets the deterrence equation by increasing the chances for success and possibly even reducing the credibility of the government’s coercive threats. Martin-Brûlé considers a series of peace-keeping missions to determine under what conditions outside actors can successfully prevent violence through coercive threats. By applying deterrence theory to the behavior of peace-keeping operations, she finds that the identity of the major troop contributors should not be taken for granted. Finally, we conclude the volume by considering the insights gained by moving away from the security dilemma and focusing more on the government and the dual dilemmas it faces: to deter and to assure. As we review the contributions, we develop the implications for policymakers – regarding institutional design, peace-keeping operations, economic policy, and oversight over the coercive arms of government.

Notes 1 The standard language from NATO documents on its missions in Bosnia and Kosovo.

18 S.M. Saideman and M.-J. Zahar 2 Our focus here is on physical security rather than human security. The latter refers to a broad array of issues and concerns that are clearly important (Axworthy 2001; Mathews 1989; Buzan et al. 1998; Thomas 2000). To focus on everything simultaneously is quite difficult, so we prefer to focus on what we consider to be logically prior – living and dying. 3 Weinstein (2007) takes seriously the challenges insurgent organizations face as they seek and gain more members. 4 For an excellent analysis of the dynamics of insurgency and counter-insurgency, see Kalyvas 2006. 5 We understand that ending a conflict may not be completely parallel to the start of a conflict, but these theories tend to shape attitudes about both causes and potential resolutions. Further, there is far more literature on civil war, failed states, contentious politics, and ethnic conflict. Rather, we highlight certain prominent approaches that frame our effort. 6 Or Selectorate, which refers to the individuals or institutions that are crucial for selecting and removing leaders (Roeder 1991). 7 For the psychology of prospect theory, see Tversky and Kahneman 1981. For its application to International Relations see Levy 1992, 2000; McDermott 2004. For a recent application to the topic at hand, see Masters 2004. 8 See also Ballantine and Sherman 2003. 9 See also Rose (2000), Kaufmann (1996), and Snyder and Jervis (1999). 10 Gurr 2000; Fearon and Laitin 2002, 2003; Toft 2003; Saideman and Ayres 2000; Saideman et al. 2002. 11 Quantitative analyses have shown that inter-group strife (communal conflict) and group-state violence (referred to as rebellion in the “Minorities at Risk Dataset”) have similar but not identical dynamics. See Saideman et al. 2004. 12 Even in Afghanistan, where the NATO operation is using force and taking significant casualties, some troop contributors, especially Germany, France, and Italy, are handicapped by inhibitions of using force or being attacked. 13 For a similar but not identical discussion of deterrence and assurance, see Stein 1991. 14 This is similar but not identical to Stedman’s concept of spoilers. 15 In this volume, we limit our focus to indigenous political movements, rather than transnational ones, i.e. Al Qaeda. 16 Of course, when thinking about deterrence, one generally assumes that an actor would engage in the undesired course of action if there were no constraints. This gets us back to the type of opponent discussed above. 17 Indeed, there has been much debate both preceding the invasion of Iraq and afterwards about whether the US provided enough troops on the ground to do the job, see Ricks 2006. 18 This is how Fearon and Laitin (2003) see GDP/capita. 19 A distinction should also be made between “conflict resolution” and “conflict management.” “Resolution” refers to the elimination of the causes of the underlying conflict, generally with the agreement of the parties. “Management” refers to the elimination, neutralization, or control of the means of pursuing either the conflict or the crisis. (Zartman 1989, 8) 20 Any discussion of resolve starts with Schelling (1980). For a survey, see Huth 1999. 21 The repression literature is quite vast. For a good introduction, see Davenport et al. 2004. 22 Indeed, deterrence at the domestic level is significantly more complicated than

Causing security, reducing fear 19

23 24 25 26

at the international level as the assurance problem is not so severe in the latter, at least in the advent of nuclear weapons and second strike capabilities. Insurgents also need the discipline to be selective, but often fail to do so (Weinstein 2007). See Petersen and Staniland (Chapter 6, this volume) for an analysis of the military’s role in the deterrence/assurance dynamic. Wilkinson (2004) finds that in systems with more politically relevant parties, politicians are likely to defend minorities. Saideman et al. (2002) find that proportional representation is associated with lower ethnic conflict. See Martin-Brûlé (Chapter 10, this volume).

2

Fear, preemption, retaliation An empirical test of the security dilemma Stathis N. Kalyvas

The security dilemma is one of the best known and most influential conjectures about how violence is initiated and escalates in the context of civil war. A concept imported to the study of civil wars from the realist analysis of interstate conflict by Barry Posen (1993), this insight has been used, modified, extended, or revised in several papers focusing on civil war onset and escalation (e.g. De Figueiredo and Weingast 1999; Saideman 1998; Hardin 1995). The Yugoslav civil wars are the main case used to illustrate the dynamics of the security dilemma. This chapter empirically evaluates the security dilemma using microlevel evidence. Specifically, I assess the extent to which the security dilemma is a good predictor of the dynamics of violence in the context of an irregular war. Since the security dilemma can be observed directly only with great difficulty, if at all, I test the observable implications of the security dilemma argument. Three such implications are derived: first, violence is most likely where populations are most intermixed – ideologically, religiously, or ethnically; second, violence should be launched by substantial minorities that are weak enough to feel threatened, yet strong enough to be able to launch an attack; and, third, violence is the only possible outcome under conditions of anarchy – as opposed to peaceful arrangements, such as local peace deals. I test these implications using fine-grained, micro-level evidence from the Greek Civil War and the Vietnam War. Although this test can only be preliminary given the nature of the data, I do not find compelling evidence in favor of the security dilemma. Extreme levels of insecurity as measured by demographic intermixing fail to predict the outbreak and escalation of violence; in fact, extreme insecurity serves as a deterrent more often than not, leading to enforceable local peace deals. When violence breaks out, it is driven by groups that have a clear (though not total) local power advantage rather than minorities. In this case, aggression serves to consolidate this advantage rather than reverse an unfavorable power balance. This chapter demonstrates that deterrence does take place – that individuals and groups are sensitive to the costs of their actions and the risks of reprisals. Furthermore, I suggest

Fear, preemption, retaliation 21 that the mechanism underlying the escalation of violence may be related to the retrospective logic of revenge rather than the prospective logic of the security dilemma. Methodologically, this chapter makes the case for a micro-level based analysis that operationalizes large abstract concepts at the micro level.

Theory: the security dilemma The security dilemma purports to explain why individuals and groups engage in violence and why this violence escalates into war.1 A security dilemma is said to occur when the breakdown of order creates a situation whereby individuals coordinating around focal points (primarily ethnic identities) resort to preemptive violence, or align with warmongering leaders who do so, because of security fears, specifically the inability to distinguish defensive from offensive measures. Incentives for using violence are compounded by uncertainty and the extremely high costs of suffering the first blow: “one must kill so as not to be killed.” Although the masses are motivated by survival considerations and may be conditioned by emotions, such as fear, elites may act instrumentally, in pursuit of power. The security dilemma applies at once to the onset of conflict and the dynamics of violence – though this distinction is rarely made in practice; usually, the dynamics of violence are seen as the predictable manifestation of the onset of intrastate hostilities. However, onset and dynamic ought not to be confused. On the one hand, even if the security dilemma accounts for the onset of the conflict, the violence that ensues may follow its own, distinct, logic. On the other hand, the security dilemma may be only one among many possible mechanisms at work in both the onset and the dynamics of the violence. The exact mechanics of the operation of the security dilemma remain fuzzy at best. While the security dilemma is a plausible proximate cause of either conflict onset or dynamics of violence, it is hard to operationalize and test. Furthermore, there is an acute problem of observational equivalence as the empirical manifestations of the security dilemma tend to be equivalent with several other mechanisms. Therefore, pushing the research agenda forward entails the derivation of specific testable implications that can be differentiated from those of other mechanisms, along with the direct observation of the process at work. In this chapter, I focus primarily on the dynamics rather than the onset of violence, trying to assess the extent to which the security dilemma is a good predictor of the dynamics of violence in the context of an irregular war.2 That is, I study the local dynamics of violence conditional on civil war onset. My focus is on individuals and small groups approached at the micro (i.e. local) level. I derive three testable implications from the security dilemma argument. First, if this argument is correct, violence should erupt where populations

22 S.N. Kalyvas are most intermixed, for this is where the danger of intention misidentification, and its direct cost, is the highest. Intermixing does not have to be ethnic (after all, the logic security dilemma applies to generic “groups,” and ideological groups can be as cohesive as ethnic or religious groups). Second, the actor most likely to engage in violence should be a large minority or a group in a position of slight disadvantage; such a group is weak enough to feel threatened, yet strong enough to be able to launch an attack with a credible chance of success. Third, violence should be the only (or at least, main) outcome under conditions of anarchy. The security dilemma argument is incompatible with nonviolent outcomes, such as local peace deals.

Empirics It is not easy to test these observable implications. For one, the anecdotal record is indeterminate and it is possible to find examples that both confirm or disconfirm them. Examples of violence that appears to be plausibly preemptive in a way that “fits” ex post the logic of the security dilemma are not infrequent in descriptions of civil wars. Posen focuses on the wars of Yugoslavia, but one can find many descriptions that extend the logic of the security dilemma to individuals and the mass-level. For Abraham Lincoln, the American Civil War in the borderland states, where the war was not fought conventionally, was a situation whereby “each man feels an impulse to kill his neighbor, lest he be first killed by him” (quoted in Fellman 1989: 85). A Union commander in Missouri pointed out in 1864 that “there is scarcely a citizen in the country but wants to kill someone of his neighbors for fear that said neighbor may kill him” (in Fellman 1989: 62). In the immediate aftermath of World War II in Malaya, a man recalled that “Once the first clashes were known, the communists in the MPAJA alerted the Chinese in the villages and towns all over the country, to be ready for the Malay rampage. The cry was: “The Malays are out to kill – so, kill before you are killed” (Kheng 1983: 219). In addition, recent developments in Iraq suggest that sectarian violence has erupted in areas where Sunni and Shia populations have been intermixed (e.g. Raghavan 2006c). Yet, at the same time, it is possible to find many counterexamples. For instance, violence in the former Yugoslavia was often exercised by armies rather than among individuals and only small minorities of the various populations participated in the violence (Mueller 2004). Often these armies resorted to violence not only to maximize the probability of group survival but also for commitment purposes, i.e. to intimidate group members into participating in the war on their side (Petersen 2002). Minority or majority status did not cause significant differences in levels of tolerance levels for “enclaved” populations, a plausible proxy for violence (Massey et al. 1999); and there is evidence that local peace deals were reached in several circumstances (Bougarel 1996).

Fear, preemption, retaliation 23 Nevertheless, anecdotal evidence does not constitute a test of the security dilemma conjecture. In this chapter I draw from my theoretical analysis of the logic of violence in civil war (Kalyvas 2006), using data from the Greek Civil War and the Vietnam War. The main assumption is that the onset of conflict has already taken place; that is, I examine the dynamics of violence rather than its onset. The set-up is as follows: there are two actors whom I call incumbents and insurgents fighting for local control via irregular warfare; the main analytical tool I rely on, consists of a measure of local control (or sovereignty) that can be applied to analyze the dynamics of irregular war at the local level. This measure apportions a given territory into five discrete zones, ranging from 1 to 5. Zone 1 is an area of total incumbent control and zone 5 is an area of total insurgent control. These are places where the state and a counter-state operate pretty much unimpeded even though a civil war may be raging elsewhere in the country. Incumbents and insurgents, respectively, enjoy a monopoly of violence and can effectively exclude the rival power from the territory they control. In between those two zones of total control lies contested territory. This territory can be disaggregated into three zones (2, 3, and 4) where control varies. Zone 2 is primarily (but not fully) controlled by the incumbents (dominant incumbent control) and zone 4 is primarily (but not fully) controlled by the insurgents (dominant insurgent control). In those areas, each side has a definitive advantage but cannot exclude its rival from intruding. Lastly, zone 3 is “shared” by the two rivals, that is, they are able to exercise equal levels of control. These are places usually described as areas where “the state governs by day and the insurgents by night.” Obviously, zones of control are subject to temporal variation as well. One way to test the security dilemma thesis at the micro level is to hypothesize that violence will be most likely where the two sides are most intermixed. “Intermixing” captures the intuition of maximum vulnerability coupled with maximum offensive capacity. In the terminology of control used in this chapter, intermixing is the highest in zone 3. This is precisely where we would expect group vulnerability to be most acute given a military capacity to avert it. This is to say, that even though a group’s vulnerability to, e.g. insurgents, would be higher in zones 5 and 4, respectively, the ability to counter and reverse this vulnerability by military means would be limited in these zones. It follows from this reasoning that the security dilemma argument would predict violence to be most likely in zone 3. Furthermore, the same logic would also leads us to expect violence to be launched by groups at a slight disadvantage. If this is the case, we should expect to observe violence launched by incumbents in zone 4 and insurgents in zone 2, for this is where these groups are at a disadvantage and have the most to fear. Last, we should not expect to observe non-violence in zone 3. If anarchy is defined by the absence of a monopoly of violence, then areas where the competition for power (and hence for the prized

24 S.N. Kalyvas monopoly of violence) is most acute, would be violent rather than peaceful places. In sum, the security dilemma conjecture would predict (a) violence in zone 3 and/or (b) violence by insurgents in zone 2 and incumbents in zone 4 and/or (c) lack of peace in zone 3. Note that predictions (a) and (c) are not tautological in the sense that “peace” is not just a default condition but one actively sought out by the parties involved. I test these hypotheses using a dataset that includes all homicides that took place in 1943–44 in the 61 villages of two of the three counties of the Argolid, a region of Southern Greece that underwent a civil war during that year, overlapping with German occupation. This was an ideological war between the Right (which was forced into collaboration with the German occupiers) and the communist Left.3 Like most of Greece, the Argolid was (and remains) a predominantly rural region, dominated by family farms, ethnically and religiously homogeneous. However, despite this homogeneity, the Argolid underwent a very high level of violence. Between September 1943 and September 1944, 725 civilians met violent deaths in the 61 villages I studied, altogether 1.61 percent of their population. In contrast, only 45 combatants from the Argolid were killed in various battles (31 rebels and 14 collaborationist militiamen). Among the individuals killed, 370 (51.03 percent) were killed selectively and 355 (48.97 percent) indiscriminately. By selective violence I refer to a process of personalized targeting and by indiscriminate violence a process of collective targeting.4 The Germans and their local allies killed 353 people (48.69 percent) and the insurgents 372 (51.31 percent). The killings peaked in mid-May and June 1944 but are spread in four distinct time periods which begin in September 1943 and end in September 1944 with the departure of the Germans. By holding time constant (in four discrete time periods), I disaggregate the dynamic process of escalation. In addition I present preliminary evidence from the Hamlet Evaluation System (HES), a large dataset compiled by US agencies during the Vietnam War that includes all villages of Vietnam in 1969.5 Although the data come from a specific region of a specific country, they are extremely fine-grained and, therefore, allow the derivation of tentative conclusions about the micro-mechanisms of civil war and violence that macro studies typically assume but cannot possibly examine. Overall, the data are inconsistent with the security dilemma conjecture. This becomes quite clear in the bivariate scatterplots of both indiscriminate and selective violence for two out of four time periods of the conflict (these periods include most instances of villages located in zone 3) (Figures 2.1–2.4). These scatterplots show that, with the exception of selective violence during the first time period, an even intermixing of the two sides does not correlate with high levels of violence.6 Even in the first time period, selective violence is by no means concentrated in zone 3. In fact, the distribution of violence is bimodal with an incumbent concentration in zone 2 and an insurgent concentration in zone 4 (Figure 2.5).

5 +

c(t 1)

*

*

***

*

**

*

*

*

*

0 + +

0

IVt

1

+ 78

Villages where indiscr iminate violence w as used at t1 b y both actors (n umber of victims per village targeted)

Figure 2.1 Control zone and indiscriminate violence at t1.

5 + *

* c(t 4) * *

* *

0+ * +

0

IVt

+ 4

Villages where indiscr iminate violence w as used at t4 b y both actors (n umber of victims per village targeted)

Figure 2.2 Control zone and indiscriminate violence at t4.

5

5 +

* *

c(t 1)

* *

*

*

* *

*

*

*

*

0 + * +

0

SVt

+ 5

1

Villages where selectiv e violence w as used at t1 b y both actors (n umber of victims per village targeted)

Figure 2.3 Control zone and selective violence at t1.

5 +

*

*

*

*

*

*

*

c(t 4) * *

* 0 + *

* * * +

+ 0

SVt

4

Villages where selectiv e violence w as used at t4 b y both actors (n umber of victims per village targeted)

Figure 2.4 Control zone and selective violence at t4.

26

Fear, preemption, retaliation 27 1.00

Incumbent violence Insurgent violence

Propor

tion of villages in a z one of control exper iencing selectiv e violence

0.90

Selectiv

e violence , Argolid September 1943–September 1944

0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 1

2

3

4

5

Control z one

Figure 2.5 Distribution of selective violence in the Argolid (n = 244 villages/time periods).

Further weakening the security dilemma conjecture is the fact that the perpetrator of selective violence in zones of control where one actor has an advantage is almost always the stronger rather than the weaker actor.7 In sum, where the two sides are equally strong, the most likely outcome is, quite surprisingly, peace rather than violence. In contrast, violence is most likely where and when one political actor, whether the incumbents or the insurgents, enjoy a slight advantage. From the point of view of the political actors involved in the conflict, violence was an instrumental tool to convey the credibility of their coercive capacity and, thus, prevent civilians from providing information and support to their rivals. From the point of view of the individuals involved in the conflict, violence was a way to settle all kinds of accounts, both political and personal, bringing the full might of the organization they were associated with to bear on a variety of conflicts. The dynamics of violence are escalatory: the intensity of violence in villages that experience violence is initially low and goes up with every subsequent “round.” In turn, these “rounds” are determined by shifts in control, brought by (exogenous) military operations. The likelihood of selective massacres, where individuals are sorted out and the victims are selected from a larger list goes up with each round of violence. Indiscriminate massacres, on the other hand, are likely in the mid-point of the conflict and are related to

28 S.N. Kalyvas incumbent “mopping-up” operations that were abandoned later on, when their counter-productive effects became obvious. Mass preemptive violence just did not take place. This is surprising both at the level of villages and families. The logic of the security dilemma would lead us to expect rational actors to anticipate the escalation of the conflict and to try to eliminate or expel their enemies with an initial massive blow (e.g. at t1). Apparently, this logic did not apply in the case examined here. There are at least two possible reasons for why this is the case. First, irregular wars tend to be gradual processes of slow escalation where individuals enter with short-term horizons that do not include an anticipation of control shifts. Assuming that control will be stable, the expectation is that low levels of violence will suffice. Second, striking a huge blow at the outset appears to transgress a set of norms that seem to regulate the process of violence. These norms, clearly related to the norms of blood feuding, dictate a process of gradual reciprocal escalation and exclude unprovoked massive violence. These findings are reinforced by the qualitative violence I collected in Greece, mainly from interviews with surviving participants and ordinary people alike as well as from judicial archives from trials that took place after the war. This evidence further supports the conclusion that violence was not undertaken because of misperceptions about the intentions of the other side. What is particularly interesting is the logic behind the relative absence of violence in areas of shared control (zone 3). It turns out, that mutual deterrence, the fear that one side would be able to retaliate immediately and effectively if violence was launched by its rival, is behind implicit peace deals that emerged in these areas. These deals were made by the local representatives of the rival parties and were robust enough to check their armed wings. In other words, when armed groups pushed for violence, their local collaborators effectively resisted it and protected their local enemies because they knew that if they did not they would pay a heavy price. Interestingly, the logic of mutual deterrence is reminiscent of that of the Mutual Assured Destruction (MAD) logic of nuclear competition, applied to the local level. Obviously, any test with data from a single civil war faces an acute problem of external validity. The problem is that very little data coded at the local level exists. It is even more unlikely to find data (or proxies) for local control. I was able, however, to retrieve a remarkable dataset that does exactly that, the “Hamlet Evaluation System” (HES). A massive project undertaken during the Vietnam War, it coded all Vietnamese hamlets on a variety of dimensions, including control and Vietcong violence. Figure 2.6 shows the distribution of Vietcong selective violence in 1969, the earliest year for which data are available (and, hence, the part of the survey that captures the highest levels of Vietcong strength which kept declining after 1965); this year’s data includes 61,701 observations which correspond to about seven bi-monthly surveys for about each of all 10,000

e

Fear, preemption, retaliation 29 0.09

P ercentage of villages with selectiv violence per time per iod

0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0.00

1

2

3 Zones of control

4

5

Figure 2.6 Viet Cong selective violence, 1969 (n = 61,701 villages/time period) (source: Hamlet Evaluation System).

Vietnamese hamlets. Unfortunately, the HES had only a measure of Vietcong selective violence and excluded indiscriminate violence and incumbent (South Vietnamese or US) violence. Though the percentage of villages that experience selective violence is much smaller than in the Greek case (which reflects the difference between bi-monthly surveys independent of control shifts and my method of coding time only after a major control shift), the shape of the distribution is surprisingly similar. Again, zone 3 does not seem to be particularly violenceprone – much less so compared to zone 4. The same is the case for zone 2, where levels of Vietcong violence are low. In other words, the Vietcong focus their violence where they enjoy a slight advantage, i.e. in those villages where their control is in a phase of consolidation rather than in those villages where they either have no or low control, share control equally with their enemies, or enjoy full and undisputed control. Clearly, this is inconsistent with the security dilemma conjecture. The same mechanism that I uncovered in the Greek case has also been reported about Vietnam: one example comes from the contested Vietnamese village of Binh Nghia, where in 1965–67, a detachment of Marines and a local South-Vietnamese militia ruled by day while the Vietcong ruled by night. Both sides were able to collect taxes: “Our acting village chief, Mr. Trao,” one of the marines, F. J. West, recalls (1985: 254), “says his opposite number has a list of who should pay and how much.” Although the local Vietcong did not dare visit their home in the village on a regular basis, West (1985: 5; 219–20) remembers that

30 S.N. Kalyvas Their families were immune from the violence. The relatives and children of both sides were equally vulnerable to reprisals, so no man dared to strike the family of another, lest his own family suffer ten times over. . . . The PFs [militias] and the Viet Cong had certain rules to their war, understandings which were kept because, and only so long as, they were mutually advantageous. What often has been called accommodation frequently has been nothing more than a precarious balance of power, perceived as such by both sides. Deterrence is a better word than accommodation to describe a situation wherein each side is unwilling to undertake certain acts while the other side retains capability to retaliate in kind. . . . The ultimate step in escalation – the murder or wholesale slaughter of PF families – was unlikely in Binh Nghia because the VC families acted as hostages. Suong [the leader of the militia] had declared that he would kill ten of their children for each member of a PF family killed. Vulnerability to retaliation set limits on the actions either the PFs or the Viet Cong were willing to take in the struggle for Binh Nghia. In sum, if one is to look for analogies from International Relations theory, then it would seem that the classic deterrence theory is a much better analogy compared to the security dilemma. An additional indirect test of the security dilemma conjecture is the close observation of the logic driving the violence. Even when we observe patterns of violence that appear observationally equivalent with the security dilemma conjecture, a close examination of their internal logic suggests a different mechanism at work. A quick look at the recent patterns of sectarian violence in Iraq motivates the discussion. In Iraq, 2006 was the year in which the so-called sectarian violence between Shia and Sunni fighters (and against predominantly Sunni and Shia non-combatants) became the dominant form of violence, perhaps even supplanting the Sunni anti-American insurgency in Western Iraq. It appears that sectarian violence is more likely in areas where the population is intermixed (Raghavan 2006c). In a sense, this is hardly surprising, since potential victims are easier to locate in precisely such areas. Does this pattern suggest that the security dilemma is at work in Iraq? Three possible mechanisms may be at work, besides the security dilemma. A first mechanism is the instrumental use of violence to elicit identity formation via emotions (Petersen 2002); a second one is related to the dynamics of revenge a process that is retrospective rather than prospective (Kalyvas 2006). Last, is organizational fragmentation which also begets violence. Looking at Iraq, it is clear that sectarian violence is often motivated by revenge – this is the motive that dominates journalistic reports (e.g. Raghavan 2006a; Daragahi 2006; Tavernise 2006; Oppel 2006). The logic of revenge is associated to patterns of traditional, tribal blood feuding, following gradually escalating “cycles” of “attack and

Fear, preemption, retaliation 31 counterattack” with militias showing up at funerals, offering to carry out revenge attacks (Tavernise 2006; Raghavan 2006a). Organizational fragmentation is also associated with violence (Raghavan 2006b) and it is possible to find evidence that suicide bombings were instrumentally used by Sunni insurgents to provoke Shias into revenge attacks and foment a civil war in logic akin to that posited by Petersen. My data from Greece also suggest that revenge was an important part of the logic of escalation. Because revenge is retrospective, i.e. it represents a response to past violence rather than a way to preempt future violence, it is incompatible with the logic of the security dilemma which is prospective. In short, while deterrence is associated with the prevention of violence, revenge is associated with the escalation of violence. Between the two poles of deterrence and revenge, there is little room for the security dilemma. A question arising here is the extent to which this empirical test misses an important, though implicit, dimension of the security dilemma: ethnic polarization was absent from both Greece and Vietnam. The answer to this question hinges on the exact work done by ethnicity. If ethnicity implies total separation and low communication between groups at the local level, then both Greece and Vietnam would be inappropriate tests. So would also be Yugoslavia, which occupies a prominent place in the security dilemma literature. As is well known, the levels of inter-ethnic interaction and communication in pre-civil war Yugoslavia were very high. Conversely, political and ideological polarization in both Greece and Vietnam was very high, though defection to the other side remained possible – unlike in Yugoslavia but like in many other ethnic conflicts (e.g. Kashmir or Chechnya). Another variable, the type of warfare, may be potentially more relevant: unlike irregular wars, civil wars that erupt following processes of state implosion display dynamics that may be compatible with different logics (Kalyvas and Kocher 2007).

Conclusion This chapter relied on fine-grained micro-level data to assess the empirical relevance of the security dilemma conjecture – a conjecture that explains the onset and escalation of violence in civil war contexts as a result of a group’s decision to launch preemptive violence. Although this test is preliminary, no compelling evidence in favor of the security dilemma emerges. Extreme levels of insecurity as measured by group intermixing fail to predict the outbreak and escalation of violence; in fact, extreme insecurity serves as a deterrent more often than not, leading to enforceable local peace deals. In other words, this chapter demonstrates that deterrence is a key mechanism behind nonviolence in civil war. Conversely, this chapter suggests that when violence is indeed observed, it may be much more compatible with the retrospective logic of revenge rather than the prospective logic of the security dilemma. Between these two poles – deterrence and

32 S.N. Kalyvas revenge – little room is left for the security dilemma. Lastly, this chapter demonstrates the need of moving from macro to micro analysis as the best way to uncover the mechanisms driving violence in civil war contexts.

Notes 1 Though war is defined as mass collective violence, not all political violence constitutes war; from violent protest to mass riots, there are many ways in which violence can erupt without escalating into a war (Kalyvas 2006). 2 I use the term “irregular war” as equivalent of “guerrilla war.” Analytically, the distinct character of irregular war is marked by the lack of clear frontlines between the parties to the war and the unwillingness of the insurgents to directly face-off the incumbents in the context of set-piece battles. Irregular wars can be fought among indigenous incumbents and insurgents, foreign incumbents and indigenous insurgents, or a combination of foreign and indigenous incumbents and indigenous insurgents (see Kalyvas 2006 for details on this definition). 3 The occupation became a civil war after 1943, insofar as the greatest part of the fighting was Greek against Greek; the immense majority of the victims, as well as a large part of the perpetrators, were also Greek. Though ideological, the conflict had a pronounced “ethnic” character involving foreign occupiers and indigenous insurgents. 4 In this particular case, collective targeting was decided primarily on the basis of location. 5 For details see Kalyvas and Kocher (2005). 6 To allay concerns about the endogeneity of control to violence I coded control as temporally preceding violence. 7 I also estimated a series of multivariate OLS regressions that include a number of additional independent variables on top of the control variable (Kalyvas 2006). The results show that zone 3 has no significant effect on the violence, either selective or indiscriminate (and the coefficient’s sign is consistently negative). Note that the weaker actor may commit acts of indiscriminate violence in those areas, but these are usually raids and attacks undertaken mostly by the incumbents to “clean” guerrilla-controlled areas rather than the kind of preemptive violence caused by the security dilemma.

3

Six feet over Internal war, battle deaths and the influence of the living on the dead David A. Armstrong II and Christian Davenport

Within the past ten years, there has been an increased amount of attention given to human security – namely, the ability of individuals to remain safe from political violence. Accordingly, researchers, policymakers and activists have spent a great deal of time discussing the diverse ways that the world’s population are killed – noting relevant actors, actions, trends and the political-economic circumstances involved. Unfortunately, however, these efforts have been somewhat imbalanced. For example, most individuals studying conflict-related deaths focus on actions that challenge political authorities (e.g. terrorism, guerrilla warfare and civil war), ignoring the fact that governments can also serve as a threat to human security, employing torture, disappearances, massacres and genocide/politicide. Of course, most are now aware of organizations such as Amnesty International and Human Rights Watch, who do focus on such things and, similarly, many are now aware of state-sponsored political violence in Sudan (Darfur), Rwanda, Kosovo and Abu Ghraib but these are the exceptions. Within academic research, the evidence of skewed prioritization is clear. Perusing social science journals and major presses over the last few years, there has been approximately 80 quantitative analyses of civil and interstate war – respectively, but over the same period of time there have been only a handful of quantitative investigations of genocide/politicide. These differences exist despite the fact that data is readily available on all three types of conflict (at relative degrees of quality) and the fact that the statistical models employ the same explanatory variables.1 Such an imbalance in focus has significant implications for research, policymaking, activism and popular levels of awareness/understanding of the problem. Take for example, the debate surrounding conflict “lethality” (i.e. the number of people killed during contentious activities). For several years, scholars have been attempting to understand whether conflict has become more or less lethal over time but the work has been largely inconclusive. For some (e.g. Sarkees et al. 2003), the number of fatalities has been relatively stable over the past 50 years, communicating that major political-economic developments such as the growth of democracy and

34 D.A. Armstrong II and C. Davenport globalization have been largely ineffective at reducing lethality. For others (e.g. Lacina et al. 2005), the number of victims has decreased, suggesting that “the post-World War II international system embodies a largely successful project of democracy, economic interdependence, and rule-based interstate order” (p. 31). While this is extremely important work, we believe that the conclusions reached by this research are misleading for they are based on a very specific definition of conflict fatality that does not explicitly consider the victims of state action. To understand lethality as well as human security, however, it is critical to consider the full range of actors that engage in relevant behavior. Simply, one must consider the full range of contentious politics undertaken by the living in order to understand the diverse processes of collective mortality (i.e. the dead).2 In an effort to address this broader concern with competitively assessing the greatest threats to citizens as well as why these exist, we wed distinct research traditions together and utilize several databases to examine the relative importance of authorities and dissidents in generating fatalities. From our research, we confirm that governments have produced greater numbers of deaths and conclude that political authorities are far more threatening to human life than government challengers. The research also discloses that although economic development and military expenditures consistently decrease lethality, these influences are non-existent until the highest values of these variables are reached; low to moderate development or expenditures have no influence on the number of individuals killed during conflict. Finally, we find that the importance of some variables, traditionally believed to carry a lot of explanatory weight, are not very important. For example, when one considers deaths from genocide/politicide at the same time with those from interstate and civil war, democracy decreases lethality but the influence of this variable pales in comparison to that wielded by population size. Below, we discuss the different orientations used to understand human losses during conflict (i.e. battle-deaths). Following this, we use a model, developed by Fearon and Laitin (2003), that focused on civil war to explain fatalities across diverse forms of large-scale conflict (i.e. civil war, genocide/politicide and revolution, as well as a combination of civil war and genocide/politicide). Within the next two sections, we present our data and research methodology, as well as our statistical results. We conclude with an assessment of exactly how our research challenges existing work and how it should guide additional investigations of the topic in the future.

On death and dying For quite some time researchers have been interested in the lethality of political conflict. Indeed, this is one of the standard characteristics, along with onset, type, frequency and duration, identified by the founders of

Six feet over 35 modern conflict studies (e.g. Sorokin, Richardson and Wright) and continuously focused upon up until the present (e.g. the Correlates of War project and the research provided by Upsalla and PRIO). Researchers have not been concerned with all conflict-related deaths, however; some have died without much attention at all, while others have received the full weight of our collective gaze. Two distinct areas are discussed below. Against political order: interstate and non-state threats to human security Identifying political authorities as the primary agents of order and stability, most interested in conflict fatality have focused on the deaths that occur when states fight each other (interstate war) or when states are challenged from within (civil war and increasingly terrorism). The threshold of exactly how many individuals need to die before relevant events are considered varies across research projects,3 but the focus is on a particular type of contentious interaction – when the state’s power is challenged and when individuals die during such circumstances. In this view, political authorities are the arbiters of peace and protectors of the status quo.4 As such, when governments hold coercive power monopolistically, it is expected that there would be fewer instances of political violence and hence fewer conflictrelated deaths. When these entities were challenged, however, from outside or within, it is expected that deaths would be increased. Within this framework, concerned individuals have identified when challengers to authorities take place and how severe they were (i.e. how many people die when they occur). As a major objective of social science is or should be the alleviation of human suffering, such information is essential to compile for it facilitates the identification of trends and causal influences (e.g. Singer 1976). These, theoretically at least, can be used to identify as well as direct intervention and post-conflict redevelopment. To account for variation in fatality, researchers have systematically examined information compiled on the different characteristics above. The standard explanatory model used by this work considers diverse domestic political-economic characteristics (e.g. democracy, economic development and population size) as well as some international factors (e.g. trade dependence and alliances). As found, poorly developed, large and autocratic societies with relatively little power in the international system as well as generally isolated from trade/investment are the main culprits of political violence – killing off large numbers of their citizens. These findings are robust across studies, time and methodologies. In defense of political order: domestic state threats Other researchers do not accept, by definition, the legitimacy or pacific nature of the state assumed above and they identify that these political

36 D.A. Armstrong II and C. Davenport entities can and have been a major source of threat to human life (e.g. De Jouvenal 1945; Rummel 1997; Van den Berghe 1990; Korey 1998; Totten and Jacobs 2002). Specifically, these individuals argue that many die when those in government employ coercive power against those under their territorial jurisdiction, engaging in genocide/politicide. Similar to the previous area of research, the threshold here is a point of much discussion (“exactly how many deaths does it take to make a genocide/politicide”). But, different from the earlier work, the behavior is identified by the intent of the state and not some body count – although some capacity for effectively carrying out political violence must be exhibited (Harff 2003, 58). With these interests in mind, concerned individuals have identified when large-scale, state-sponsored violent activity occurred and how severe they are (e.g. Mitchell and McCormick 1988; Poe and Tate 1994; Stanton 2002; Harff 2003; Cingranelli and Richards 2004; Davenport 2004; Davenport and Armstrong 2004; Valentino et al. 2004). These efforts have been fewer in number and less frequently employed by researchers in assessing the prospects of human life, but there has been some attention given to the topic. Interestingly, the same explanatory factors are found to wield an influence. Again, it is found that poorly developed, large and autocratic societies with relatively little power in the international system, as well as generally isolated from trade/investment, are the main culprits of political violence – killing off large numbers of their citizens. These findings are robust across studies, periods and methodologies. Different from research on civil and interstate war, however, the influence of international factors is mixed. Several studies reveal that these explanatory factors are quite weak or inconsistent in their effects (Hafner-Burton 2005; Hathaway 2002).

A comparative assessment of lethality How do the two forms of conflict compare to one another? Over the 1946-to-2000 period, a side by side evaluation of new battle-death data (provided by Lacina and Gleditsch (2005)) and genocide/politicide data (that we compiled from diverse sources) is startling. Observing the number of deaths due to both forms of conflict, it is clear that genocide/politicide far exceeds civil and interstate war by an average magnitude of eight. Now, clearly there are some changes within the pattern of genocidal/politicidal lethality over time. The number of victims drops significantly in 1988–1989, following the collapse of the Soviet Union, only to increase again temporarily for Rwanda in 1994. By and large, however, the number of victims was relatively steady during the post 1988–1989 period, and following 1989, the number of deaths due to genocide/politicide is still higher than that attributed to civil and interstate war – albeit a much lower difference. The implications of this trend are significant. First, if one is interested in the influence of conflict on human mortality,

Six feet over 37 3,500,000

Genocide/democide Battle

Deaths

2,500,000

1,500,000

500,000 0 1950

1960

1970

1980

1990

2000

Y ear

Figure 3.1 Trends over time in battle deaths and deaths from genocide/democide.

then it seems essential to focus on the state – specifically genocide/politicide, in addition to other forms of contentious politics. There is simply no other reasonable conclusion given the sheer magnitude of the difference. Accordingly, we strongly disagree with the argument of Gleditsch et al. (2002, 617) when they suggest that “[r]ather than expanding the (existing) conflict list, we think that there is a need for a separate list of genocides and extremely serious human rights violations.”5 Second, when genocide/politicide is considered, the issue of post-World War II pacification maintained by several in the literature becomes suspect. During the same time that democratization, economic interdependence and cultural globalization are growing, one sees a relatively steady amount of state-sponsored, large-scale killing. The dominant forces in the globe appear to have brought with them, simultaneously, a liberal-democratic dream within part of the world and an autocratic nightmare in the other.6 Again, the need for considering state-sponsored violence is revealed. Third, and complicating matters significantly, it is apparent that there is no clear-cut separation between the diverse forms of conflict that researchers of battle deaths study (interstate and civil war) and the type of conflict that human rights scholars focus upon (genocide/politicide). Even within the figure, there appears to be some overlap between the two, which is especially revealed when one considers the countries involved. For

38 D.A. Armstrong II and C. Davenport an illustration, let us consider the case of Rwanda from 1960–1961 (independence) to the present. Following a withdrawal of Belgium and social revolution in the Congo, Rwanda underwent a major political transformation in the early 1960s. During this time, a small group of Hutu replaced a small group of Tutsi as political leaders, reversing an authority pattern that had existed for many years. The majority of the population was still without substantive political and economic change but the modification in leaders was important. This was especially the case for how different citizens were treated. For instance, as the Hutu-affiliated regime was coming into power, there were numerous instances of repression being enacted against ethnic Tutsis throughout Rwanda, resulting in the mass migration of this group to surrounding countries: e.g. Uganda and Tanzania. From this Tutsi diaspora formed a rebel organization – the Rwandan Patriotic Front (hereafter RPF), who decades later (in 1990) began a civil war with the Hutu-oriented Rwandan government. Following failed attempts at negotiation (e.g. the Arusha Peace agreement) and democratization, violence in Burundi during the early 1990s that provoked mass migration of several hundred thousand Hutus into Rwanda, as well as the assassination of the presidents of Burundi and Rwanda on April 6, 1994, mass violence ensued. In the wake of this violence, a small group of Hutu extremists assumed power of the government and initiated politicide and genocide – of relatively equal lethality against Hutu and Tutsi (Davenport and Stam 2004). At the same time, the RPF renewed its effort to take over the country. Rather quickly, the RPF emerged as the military victor. In response, members of the Hutu regime, as well as large numbers of the population, fled into neighboring countries (especially Congo) due to widespread fear about the conflict in the country and what the returning group of Tutsi might do once in power. The transition between Hutu extremist defeat and RPF victory is important to highlight because it is widely acknowledged that while assuming power, the latter engaged in numerous reprisal killings that reached a significant number (perhaps upwards of 150,000–200,000).7 The assumption of power is also important for it is after this time that the RPF put in place a military regime, applying a significant amount of repression against the Rwandan population (Reyntjens 2004). It is clear that this behavior never reached the levels of violence encountered during 1994, and most activities involved restrictions being applied against political and civil liberties, but there was significant restrictions on civil liberties and some state-sponsored violence (e.g. Kippenberg 2005). Additionally, there were numerous efforts extended into neighboring Congo (during 1996–1998) in an effort to systematically eliminate the remaining former incumbents. There were also efforts to provoke civil war in Congo and gain control of the country by Rwanda through the installation of Kabila. All told, this violence resulted in an estimated four million fatalities between the years 1996 and 1999 (Human Rights Watch 1999).

Six feet over 39 With civil war, genocide/politicide, state repression and interstate war overlapping in many ways, it is clearly the case that multiple forms of conflict, as well as multiple actors, need to be addressed at the same time. Of course, not all contentious interactions are as difficult to disentangle as Rwanda during the 1990s but such a case should give us pause when we attempt to understand the influence of conflict on lethality by parsing diverse actions/actors and the relative threats to human life wielded by them. Indeed, such a case should compel us to more carefully consider how different forms of contention concatenate to influence fatalities. Exactly how could this be achieved? We address this below. Generating deaths While research over the last five to ten years has significantly advanced our understanding of civil war (a particular form of highly lethal political violence), perhaps none has been as important as the acknowledgment that it is the relative strength between combatants (i.e. authorities and dissidents) that holds the key to understanding this type of conflict (e.g. Fearon and Laitin 2003; Sambanis and Zinn 2002). In one of the clearest examples of this argument, Fearon and Laitin hypothesize that civil war is most likely to emerge when a government’s capacity to identify, capture and kill challengers is diminished (Fearon and Laitin 2003, 79–80) and when the growth of “insurgency” – a technique whereby “small, lightly armed bands [practice] guerilla warfare from rural base areas” (Fearon and Laitin 2003, 79), is enhanced. Of course, the struggles between authorities and dissidents are generally not balanced (with each actor wielding comparable levels of coercive power); in fact, they are normally quite lopsided in favor of the state. As conceived, the ability of the government to engage in repressive behavior leaves a situation where authorities dominate their territorial domains and where insurgents are comparatively weak.8 Under most political, economic, demographic and geographic conditions, therefore, governments possess well-functioning military organizations capable of identifying where challenges are located, deploying the appropriate measures to combat these challenges, and either bringing insurgency to an end or preventing it from getting “out of hand” (Fearon and Laitin 2003, 79–81). Under exceptional circumstances, challengers are favored. This tells us several things about lethality. First, when capable states fight each other (associated with situations of interstate war) lethality should increase as those that are the most capable are the most brutal. Alternatively, lethality might decrease as those that are most capable are the most efficient thereby preserving human life. Second, when a state’s capacity is low but dissident capability is high (generally associated with situations of civil war and revolution), conflict

40 D.A. Armstrong II and C. Davenport lethality should be low. This is explained by the fact that if states have the monopoly on the use of coercion but they are largely unable to employ it, then overall levels of violence should be limited. Alternatively, this could be exactly when conflict is most lethal. On the one hand, a weak state might be in a desperate position and resort to violence as a way to stay in power (Dallin and Breslauer 1970). On the other hand, facing a relatively weak opponent, dissidents might escalate their violent activities – attempting to push the situation into a crisis. Third, when state military capability is high but dissident capability is low (generally associated with situations of genocide/politicide), then lethality should be high. In this context, states have the monopoly on the use of coercion as well as the capability to employ this behavior without significant resistance from the population. Lacking opposition, however, it is possible that conflict would be less lethal for it is not necessary. Fourth and last, we acknowledge that the civil war literature has generally assumed that military capability influences actual state behavior (e.g. Fearon and Laitin 2003), but this is not necessarily the case. For instance, one argument maintains that states with military capability are likely to be the most active in their use of repression because they can engage in this behavior; consequently, they are the most lethal. Here, repressive behavior is applied out of a position of strength (e.g. Duvall and Stohl 1988). Others suggest that states without military capability are the ones most active in their use of repression and in these situations high levels of casualties are produced (Dallin and Breslauer 1970). Here, repression is applied out of a position of weakness (e.g. Arendt 1951). Given the difficulties with separating distinct forms of conflict and accurately parsing out casualties to different political actors (evident within the Rwandan case), we believe that the consideration of specific combinations is particularly relevant. For instance, while civil war and genocide/politicide are frequently identified at the same time – generally noted for their high degree of lethality, it is not exactly clear how this relates to the discussion above about military capability. It may be the case that when a regime is engaging in genocide/politicide and authorities along with rebels are engaging in civil war, military capability is unimportant for understanding conflict fatalities. All that is important here is that conflict has essentially exploded and two different forms of largescale contentious activity are occurring at the same time. Alternatively, it may be the case that when a regime is engaging in genocide/politicide and authorities along with rebels are engaging in civil war, military capability is extremely important for understanding conflict fatalities because it is only within states that have the wherewithal to engage with dissidents and engage in mass killing at the same time that one would likely see high casualty rates. Opposite, we display the diverse possibilities (Figure 3.2).

Six feet over 41 Genocide and civil w

ar – 2

Genocide and civil w

ar – 1

Lethality Genocide – 2

High

Civil w ar and rev olution – 2

Genocide – 1

W ar – 1

Repression – 2

Repression – 1

Lo w

High Civil w ar and rev olution – 1

Capability

W ar – 2

Lo w

Figure 3.2 Contention and casualties.

Our approach here is similar to the argument made by the editors of this volume for we see state repression as “deterring” others within the relevant territorial domain and abroad by sending a signal/message that they are able and willing to use the coercive power at their disposal. In fact, we are perhaps less sanguine about the role of the state than the editors of this volume as we focus on behavior that far exceeds the parameters of deterrent signaling: genocide/politicide seeks to wholeheartedly eliminate a group of people. We also identify that it is best to evaluate deterrent capability through the simultaneous consideration of state and non-state behavior. Repressive and dissident behaviors both send signals to the same audiences regarding not only the state’s capability to deter but also the ability of challengers to overcome this capacity. An assessment of lethality thus directs us to provide a more comprehensive understanding of the security situation. Data and methodology To investigate the interplay between conflict and capability as they influence lethality – accounting for distinct political actors, we employ data from a variety of sources. First, our measure of inter- and intra-state warrelated fatalities comes from Lacina and Gleditsch (2005). As discussed above, these data focus exclusively on those deaths resulting from inter-, intra- or extra-state wars. Data on mass state-sponsored killings not covered by Lacina and Gleditsch (2005) are obtained from three different

42 D.A. Armstrong II and C. Davenport sources. In one project, Stanton (2002) has compiled, through Genocide Watch, a list of genocides/politicides and their magnitudes. Additionally, we use Rummel’s (1997) data on democides and Harff’s (2003) data on genocide/politicide casualties.9 Specifically, we annualize these by simply dividing the battle deaths equally across the years of the conflict. Where only one figure exists per country-year, we employ that number. If more than one number exists, then we use the average.10 Table 3.1 shows descriptive statistics for battle deaths and genocide/democide deaths in different contexts. Data concerning lower-level conflict events are obtained principally from the Banks’ Cross-National Time-Series (CNTS) Data Archive (2001). This includes data on the number of guerrilla wars ongoing within a country. We use the square root of the number of years a country has had an ongoing guerrilla war11 as our first measure of lower-level conflict. The square root accounts for decreasing marginal returns of another year of guerrilla war on fatalities. The thought here is that the deadliest guerrilla wars are probably the shortest. Guerrilla wars that produce mass casualties will likely be decisive on one side or another. We also include a variable indicating whether a country is involved in a revolution12 for that particular year (also derived from the CNTS). We include a variable measuring the number of terrorist incidents13 coded from data provided by the National Memorial Institute for the Prevention of Terrorism (2004). These are truncated at 25 to reduce the influence of outliers. In an effort to capture state activity below the level of genocide/ politicide, we need to be a bit creative. As conceived, we wish to model repression’s effect on battle deaths, but the two concepts are often conflated in their measurement. Repression measures, such as Gibney’s (2005) Political Terror Scale (PTS), include killing, and therefore would be related to the genocidal/politicidal deaths by definition. Fortunately, Cingranelli and Richards (2004) have recently disaggregated the four variables that sum to the PTS – killings, forced disappearance, torture and political imprisonment. We add the latter three together for our measure of repression and standardize the variable such that it has a mean of zero and unit variance. Finally, we rely on a number of demographic factors to model the Table 3.1 Descriptive statistics for battle deaths and genocide/democide deaths in different contexts Min Conventional battle deaths Civil war no genocide/politicide Civil war with genocide/politicide Genocide/democide deaths Genocide/politicide no civil war Genocide/politicide with civil war

8 12 7.7 7.7

Mean

Median

Max

3,352 9,102

377 1,478

7,500 350,000

50,500 28,180

1,000 2,885

1,055,000 1,055,000

Six feet over 43 number of deaths generated from political violence. Specifically, we use the GDP and Population figures from Fearon and Laitin (2003). We measure Regime Type with the interpolated Polity variable from Polity IV (Marshall and Jaggers 2005). We include in the model, the squared Polity variable to allow for an “inverted U”-shaped relationship between regime-type and casualties. Finally, as a measure of military capability, we use the Correlates of War project’s data on military expenditures, divided by the country’s annual GDP. To model the relationship between our independent variables and deaths, we use a Negative Binomial regression (Long 1997) since it allows for overdispersion – a condition where the variance of the dependent variable is greater than its mean and one which our data exhibit. In addition to this, in an effort to assess the importance of specific factors (e.g. military expenditures relative to GNP) we examine interactive influences as well.

Findings Table 3.2 presents the findings of the negative binomial regression outlined above. In the first two columns are the results for all country-years in the dataset. Here, the results are largely as expected. As per-capita income increases, the number of deaths decrease, which may be simply a function of the lower likelihood of richer societies being involved in conflicts, a point discussed in more depth below. In contrast, countries with larger populations tend to have greater numbers of deaths. This could be simply an indicator of supply – countries with greater populations have a greater number of people who can die when conflict erupts. Though more likely, Table 3.2 Negative binomial regression results – conventional battle deaths for all observations and only country-years in conflict All observations

(Intercept) log(GDP/capita) log(Population) log(Military Expend/GDP) Repression Years of Guerrilla War Terrorism Revolution Polity2 Polity22 log(military expend/GDP)  repression log(military expend/GDP)  guerrilla log(military expend/GDP)  terrorism

■

Only conflict

Estimate

p-value

Estimate

p-value

–6.1197 –1.8310 0.9172 0.8771 3.3748 0.5125 0.2377 1.6504 0.0245 –0.0211 –0.5571 0.0020 –0.0360

0.000 0.000 0.000 0.000 0.000 0.292 0.007 0.000 0.306 0.000 0.005 0.990 0.177

5.6066 –0.9118 –0.0140 0.5702 2.1208 0.6095 –0.0875 0.5001 –0.0298 –0.0144 –0.4758 –0.1402 0.0624

0.000 0.000 0.851 0.000 0.000 0.017 0.122 0.023 0.064 0.000 0.000 0.095 0.002

44 D.A. Armstrong II and C. Davenport the effect of population is almost entirely owing to the increased likelihood that states with higher populations (and hence a greater drain on resources) will engage in higher-level conflicts, such as civil wars. Preliminary evidence for the latter is presented below. Polity, or more specifically, its square, is significant. This suggests that the hypothesized parabolic relationship between regime-type and deaths holds. By way of support for a military capability increases lethality argument, the coefficient for the log of military expenditures divided by GDP is positive indicating that the more money that is spent on the military, the greater the expected number of casualties. This coefficient can only be interpreted in the absence of guerrilla war and terrorism and at mean levels of repression due to the interaction terms. In contrast, those countries with ongoing revolutions can expect increased deaths opposed to those countries with stable regimes. Admittedly, the effects of the conflict variables are somewhat harder to interpret because of the interactions. The coefficients for the main terms of repression, terrorism and guerrilla war can be directly interpreted, with the former two being significant, when military expenditures divided by GDP equals zero. However, this does not provide much real insight. It is exceedingly rare that a country would spend no money on the military. The nature of multiplicative effects suggests that guerrilla war, repression and terrorism – all have different effects at different levels of military expenditures. Further, the standard error for these effects varies as a function of the different levels of military expenditure. Thus, the significance of the effect (coefficient/standard error) also varies as a function of military expenditures. The coefficients from Table 3.2 suggest that the square root of the number of years a country has been involved in a guerrilla war is significant but only for middle-levels of the log of military expenditures per GDP. In other words, if a regime is expending resources on the military and if guerrillas have been engaged in conflict for several years, then this conflict is likely to be quite lethal.14 As for the relationships between terrorism, repression and military expenditures, when military expenditures are low, countries with higher levels of repression or terrorist incidents have significantly higher levels of deaths. However, when military expenditures are high, these differences disappear. Thus, countries that spend a lot on their militaries can expect a relatively constant number of deaths regardless of the number of terrorist incidents or the country’s level of repression; in contrast, countries that do not spend a lot on their militaries can expect to find relatively high numbers of casualties suggesting that capability reduces lethality. One clear explanation for this concerns the fact that countries with large military expenditures communicate information to challengers that if they engage in dissident activity, they are likely to prompt a response from a well-funded organization. Realizing this, challengers are less likely to engage in violence unless they feel that they can avoid the state’s response and/or outright defeat political authorities.

Six feet over 45 Columns 3 and 4 present results for the same regression as above, but only for countries in which an interstate or civil war is ongoing.15 We do this in order to understand the extent to which the relationships uncovered above are more a function of the binary process of conflict/no conflict or whether they are really distinguishing between various levels of conflict activity. The results here are somewhat mixed. As found, GDP/capita again decreases repression, but its coefficient is half that of the one for all years. Amidst conflict, the influence of economic development is diminished. The relationship between population and deaths completely evaporates when the data include only those countries in conflict. This suggests that population is a better predictor of the binary conflict process than of the variation in deaths within conflicts. The relationship between regime type and deaths also seems to change, though once the main- and squaredeffects are taken into account, the relationship is largely the same. Revolution’s effect is also still significant, but its magnitude is only a third of its previous level. Military expenditures are still positively related to deaths given the same caveats as above. Both the main terms of repression and years of guerrilla war are significant, but the main term for terrorism is insignificant in this context. Here, repression has the same relationship. As military expenditures increase, the coefficients on repression decrease until they are indistinguishable from zero when the log of military expenditures/GDP is about 3. When relatively little money is spent on the military, repression increases the expected number of deaths. However, when a lot of money is spent on the military, expected deaths are essentially constant with respect to repressive behavior. The results for terrorism are particularly interesting. Amidst periods of civil and interstate war, for low levels of military expenditures, increasing amounts of terrorist events decrease battle deaths. Already taxed with two distinct forms of political violence, lethality is not enhanced by terrorism when the capability of authorities is limited. The relationship is reversed as terrorism increases deaths when a relatively large amount of money is spent on the military. Guerrilla warfare is also interesting. In all years, the effect of this variable on lethality is almost constant across all values of military expenditures. When we subset the data, however, results disclose that when little money is spent on the military, guerrilla warfare increases the number of deaths and this finding holds until middle-range values of expenditures. Above this level, there is no systematic relationship. Well funded military organizations do not lead to more deaths when civil war ensues. It has been noted by various scholars that genocides/politicides only happen in civil wars. As a result, civil war deaths should capture the deaths from genocide/politicides as well; in short, genocide/politicides might be an unremarkable case of civil war. Other scholars suggest that genocides and politicides are so aberrant and abhorrent that they should be a subject of study in their own right (Totten and Jacobs 2002). We

46 D.A. Armstrong II and C. Davenport agree that genocides/politicides are likely governed by similar processes as other high-level conflicts but also acknowledge that disregarding the unique characteristics of these events will detract from our ultimate understanding of genocides/politicides in particular and high-level conflict, as well as who is responsible for threats to human security in general. Below, we present evidence in support of this line of thinking. Table 3.3 presents results similar to those above, but for two different sets of countries. In the first and second columns are results for those countries in civil war without genocide/politicide and in columns 3 and 4 the results are presented for countries with both a civil war and genocide/politicide (identified with the data discussed earlier). The results from this analysis are informative. When genocide/politicide and civil war are considered simultaneously, income has a greater pacifying effect than when there is only a civil war. Larger countries appear to have lower casualty levels in countries with a civil war and genocide/politicide. The coefficient is not significant in civil war in the absence of genocide/politicide. The main effect for military expenditures is significant in civil wars without genocide/politicide, but it is insignificant for those with genocide/politicide. This is possibly due to the difference in military technology used in each scenario, though this is pure speculation at this point. Another key difference is found with regard to regime type. In countries experiencing only a civil war, the parabolic relationship between democracy and deaths continues to hold, but in genocides/politicides, this relationship disappears. If there is any relationship between democracy and deaths in this latter case, it is linear and not quadratic. Table 3.3 Negative binomial regression results – conventional battle deaths for civil wars with and without genocide/politicide Civil war no genocide/politicide

(Intercept) log(GDP/capita) log(population) log(military expend/GDP) Repression Years of guerrilla war Terrorism Revolution Polity2 Polity22 log(military expend/GDP)  repression log(military expend/GDP)  guerrilla log(military expend/GDP)  terrorism

Estimate

p-value

4.233 –0.360 0.151 0.400 0.862 –0.008 0.051 0.602 –0.072 –0.012 –0.079 0.029 –0.021

0.000 0.039 0.055 0.024 0.030 0.953 0.473 0.010 0.000 0.006 0.560 0.580 0.391

Civil war and genocide/politicide ■

Estimate

p-value

12.225 –1.550 –0.596 0.046 1.451 0.508 –0.039 0.078 –0.084 –0.007 –0.034 –0.037 0.006

0.000 0.000 0.001 0.833 0.000 0.046 0.625 0.843 0.064 0.492 0.778 0.549 0.822

Six feet over 47 Regardless of the level of military expenditure, guerrilla wars do not have any systematic impact on casualty levels when there is only a civil war. However, this changes when a genocide/politicide is also ongoing. Under these circumstances, guerrilla war increases deaths in the middlerange of military expenditures. The relationship of repression, conditional on military expenditures, on deaths is roughly the same across both sets of countries although there is a difference in the magnitude of the effect. Repression increases battle deaths over the middle range of military expenditures when there is only civil war and over almost the entire range of military expenditures when genocide/politicide is also happening. The conditional relationship of terrorism also changes across the two contexts. When there is only civil war, terrorism increases deaths at upper-middle levels of military expenditures. There is no systematic relationship between terrorism and deaths in countries where genocide/politicide is happening along with a civil war. This makes perfect sense because these actions are motivated by states against their own citizens and not necessarily as a function of a violent trigger. Together, these results suggest that while civil wars, where genocide/ politicide is also happening, are governed by a similar process as civil wars without genocides/politicide, there are non-trivial differences between these two circumstances, thereby suggesting that the two should be examined individually. Within the last analysis, we only considered whether a genocide/politicide was taking place along with a civil war. We did not incorporate, however, information about the number of fatalities attributed to genocide/politicide. In Table 3.4, we add deaths attributed to these events to the count of deaths compiled by Lacina and Gleditsch (2005). We then estimate the same model as above for the following three sets of observations (1) all observations (columns 1 and 2), (2) countries experiencing genocide/politicide and civil war (columns 3 and 4), and (3) countries experiencing only genocide/politicide and no civil war (columns 5 and 6). The results show interesting differences across the three analyses. For example, we find that income, in the form of GDP/capita, universally lowers deaths, not only in these three models, but also in all of the others estimated within the study. This is the most consistent finding within our research; more economically developed societies kill less people even amidst diverse forms of conflict. Population increases deaths in the first and third case (all observations and countries with genocide/politicide but not civil war) and decreases battle deaths, though not significantly, in the second subset (countries with genocide/politicide and civil war). Military expenditures are significant and increase deaths in the conflict subsets (2 and 3). While income tends to decrease deaths, if much of that income is being spent on the military, then deaths will actually increase, in the particular circumstance of mean levels of repression and no guerrilla war or terrorism.

(Intercept) log(GDP/capita) log(population) log(military expend/GDP) Repression Years of guerrilla war Terrorism Revolution Polity2 Polity22 log(military expend/GDP)  repression log(military expend/GDP)  guerrilla log(military expend/GDP)  terrorism

–4.915 –1.013 1.122 0.411 1.760 0.102 0.180 1.283 0.010 –0.007 –0.183 0.026 –0.034

Estimate

All years

0.001 0.012 0.000 0.070 0.010 0.886 0.162 0.021 0.751 0.318 0.231 0.894 0.181

p-value 8.734 –0.864 –0.163 0.435 1.735 0.180 –0.053 0.352 –0.050 0.014 –0.378 –0.097 0.013

0.000 0.000 0.051 0.004 0.000 0.451 0.392 0.106 0.003 0.005 0.000 0.185 0.454

Genocide/politicide and civil war ■ Estimate p-value

Table 3.4 Negative binomial regression results – conventional battle deaths  genocide/politicide deaths

■

–7.209 –0.865 1.208 0.984 1.381 –2.304 0.212 0.015 –0.126 0.010 –0.432 0.410 –0.049

Estimate

0.005 0.039 0.000 0.001 0.154 0.161 0.152 0.981 0.004 0.356 0.053 0.169 0.251

p-value

Genocide/politicide only

Six feet over 49 Regime type has differing relationships across the established models. Specifically, our research discloses that regime type does not matter at all when we include all observations. It has the predicted parabolic effect when investigating its relationship within genocidal/politicidal civil wars; however, its effect is linear in genocides/politicides without civil wars. In most studies of violence, democracy is seen to have a robust effect. However, we do not produce the same finding here. The different effects of democracy suggest that democracy and democratic institutions have varying effects on different types of conflict. For instance, as Davenport, Armstrong and Lichbach (2006) suggest, civil wars are often a function of the interaction of state and dissident forces. There is some ongoing instability here – thus the parabolic effect makes sense. It might be that the states with the least are the ones that are the most violent. On the other hand, genocide/politicide without a civil war suggests that citizens are not putting up much of a fight. Here, the real constraining factor is the extent of democratic institutionalization.16 As this decreases, the probability that the government could engage in these types of actions increases. Finally, results disclose that revolutions, while significant in the full dataset, are not significant for either conflict subset mentioned above. This suggests that revolution tends to be a good indicator of conflict versus non-conflict, but not a particularly good predictor of violence or lethality within conflicts. This finding is not particularly surprising. Some revolutions happen quickly with relatively little violence and others may be considerably more violent. As for the interaction terms, there are no levels of military expenditures where guerrilla war or terrorism is significant. When we add in genocidal/politicidal deaths, these lower level forms have no impact on the magnitude of killing. This is different from the analysis provided above. However, repression does have a significant impact, revealing that only the activities of political authorities matter when genocide/politicide deaths are added. For the full dataset, repressive behavior increases deaths at low levels of military expenditure. For civil wars with genocide/politicide, repression increases deaths at low levels of expenditures and decreases deaths at high levels. Well-funded militaries thus reduce the sheer magnitude of killing when they engage in genocidal/politicidal repression, while those with poor funding tend to be the most brutal. For genocide/politicide without civil war, however, repression only decreases deaths when military expenditures are high. In contrast to the civil war and genocide/politicide situation, poorly funded militaries are not more inclined to take human life than those in the middle range of expenditures.

Conclusion: the dead inform the living Our research was directed toward a particular limitation that we perceived within existing literature concerning the neglect of specific political actors

50 D.A. Armstrong II and C. Davenport in assessments of relative threats levied against those residing within the territorial jurisdiction of the state. Specifically, we highlight this problem within the context of the debate around conflict fatalities – i.e. the number of individuals that die during contentious activities. We focused on these activities not only to provide a more comprehensive examination of threat and lethality but also in an effort to assess the general effectiveness of various political and economic factors in reducing the magnitude of violence experienced when contentious activities take place. The former is important for if we are to understand how conflict influences human life, then we must explore all of the different ways that conflict can manifest itself; especially forms as brutal as genocide/politicide. The latter is important for if we are to understand what influences conflict lethality, then we must explore causal factors across the full range of relevant activities and actors. What have we learned? Essentially, we have learned two things. First, the number of individuals killed during episodes of genocide/ politicide exceeds those of all other forms of conflict by a statistically and substantively important magnitude. If one were to focus on the most significant factor to decrease the security of individuals, they would have to focus on genocide/politicide as these activities and the actors involved within its implementation – states – are the most deadly. This is crucial to understand because before we take up the cause of deaths as a dependent variable and bring the considerable knowledge and ability of quantitative political science to bear on what is undoubtedly a (if not the most) pressing issue confronting us, then we should be sure that we have defined the scope of inquiry properly. We can all acknowledge that studying deaths to decrease their frequency is important, the question really becomes, which deaths should be studied? Lacina et al. (2005) suggest that conflict-related deaths have been decreasing over the last 50 years, signaling the dawn of a more pacific century. However, the trend in deaths from genocide and politicide has been much less comforting – implicating states directly. It is not obvious that these massive killings will become any more or less frequent in the coming decades (activities in Congo, Sudan and other areas were not included because of data limitations), but this clearly reveals a persisting problem. As the world grows smaller, countries become more intertwined and the long arm of international law becomes somewhat stronger, it seemed less likely that we will see the types of conflict that generated the huge number of casualties in the middle of the previous century. Unfortunately, however, individual states turning on their citizens have not seemed to decrease with the end of the Cold War. Indeed, it is these conflicts that will be the biggest threats to human security in the future. Governments thus frequently overreact in their deterrent efforts far exceeding any level deemed reasonable. Second, the evidence presented here suggests that genocide/politicide deaths may well share some causal mechanisms with more conventional

Six feet over 51 battle deaths. Both are affected by income and democracy, though in differing magnitudes and different functional forms. There were also some differences. For example, only the most militarily capable governments were found to decrease the lethality of repression during genocide/politicide without civil war but all other levels of expenditure were indistinguishable. In the context of genocide/politicide, unless militaries are extremely well-funded, they tend to increase the magnitude of death generated; poorly funded militaries are thus associated with the most brutal forms of political violence suggesting that deterrent efforts go awry in contexts where authorities are under-prepared and overmatched. In the case of genocide/politicide with civil war, however, we find that across a large range of military where greater military expenditures decrease the lethality of state repression. Poor states that cannot fund their militaries to levels that would discourage predation from within or without are the most likely to experience the high-level civilian casualties brought on by genocide and politicide. This does not bode well for Africa and central Asia. The key implication of these findings – that states pose the greatest threats to their citizens – is that we need to focus more attention on efforts to constrain governments from preying upon their societies. While Saideman and Zahar focus equally on deterrence and assurance, the results here indicate that policymakers need to stress the latter. Given the dangers posed by governments, we need to determine how to limit their ability to harm their societies. Given the various interactions military spending had with other key variables, simply restricting such spending may not be all that helpful. Given that democracy and wealth only matter at the highest levels, we cannot expect either to be a magic potion to solve the problems of Africa and much of Asia. This chapter, instead, suggests that we need to think more about the limits of existing efforts and to focus more attention on the threats posed by states to their own societies. The choice to consider diverse political actors/actions at the same time including deaths from genocide/politicide and civil war in calculations of lethality are consequential because if they were excluded we would be led to have an overly optimistic conclusion about conflict-related deaths in the future. Considering a more comprehensive operationalization of fatalities, however, we are led to a somewhat more pessimistic conclusion about where we may be going. Given the large presence of autocratic regimes within specific regions of the world (Africa and the Middle East), the difficulties with developing many of these countries economically and the low likelihood that the international community will intervene in most of these countries before or during large-scale conflict behavior, it seems that many individuals throughout the globe (including some of the most populous countries) will continue to live in an environment where state-initiated violent behavior will take their lives. If we are to address human fatalities from conflict, therefore, it is to this part of the world and those political

52 D.A. Armstrong II and C. Davenport actors that we must turn. Such an acknowledgment is at once difficult and important to make for [w]hen the community no longer raises objections, there is an end, too, to the suppression of evil passions, and men perpetuate deeds of cruelty, fraud, treachery and barbarity so incompatible with their level of civilization that one would have thought them impossible. (Freud 1915, 278) Toward this end, our objection, our suggestion, is simple: we must not maintain a narrow conception of conflict fatalities, we must focus on all actions and actors that take human life and attempt to understand the processes involved with these diverse ways of killing. Only in this manner will we be able to structure a useful investigation. Only in this context, will the experiences of the dead assist us with protecting the living.

Notes 1 Consideration of research dedicated to protest, strikes, riots and terrorism relative to protest policing, counter-insurgency and human rights violation only reveals more of a bias – although the last has grown as of late. 2 This relatively inclusive approach to the study of conflict is reminiscent of the work initially conducted in the field – under the label “internal war” (e.g. Eckstein 1965). 3 See Sambanis (2004) for good discussion of civil war. 4 This perspective has been reinforced at numerous junctures in history (e.g. after the Enlightenment (e.g. Nagengast 1994) and following World War II (e.g. Walt 1991)). 5 We argue that there is no reason for constraining any list of conflict fatalities. This is especially the case if conflict fatality research is going to make comments about global trends and the changing prospect of human life throughout the world. For example, they state that Rummel (2000) has compiled an extensive list with approximate numbers of the total casualties in what he terms “democide,” a slightly wider category than genocide. However, his data are not annualized, and the same goes for other lists of human rights violations. These data are therefore less suitable for research based on country-years or on hazard analysis (Gleditsch et al. 2002, 619). 6 The latter is occasionally linked to the very same pacific democracies as in the cases of Guatemala (the United States) and Rwanda (the French and Belgians). 7 It should be noted that there are several databases that do conform to the country-year format required by the authors (e.g. global data: Mitchell and McCormick (1988); Poe and Tate (1994); Harff (2003); Cingranelli and Richards (2004); country-level data: Ball et al. (1999); Davenport and Stam (2004)). This estimate was given to me in private correspondence. 8 Researchers also note that a state’s extractive capability can fund counterinsurgent activities, including the provision of material rewards and political inducements. While these researchers may also note that political systems often enjoy some degree of political obedience and/or support from the population, repressive capability carries the bulk of the explanatory weight. 9 Accordingly, our research moves beyond that provided by Aydin and Gates for

Six feet over 53

10

11 12 13

14 15 16

we use multiple sources to assemble a more complete listing of genocide/politicide. In the unlikely case where each year roughly the same number of people dies, this measure would be approximately correct and thus using this measure has no negative implications whatsoever. However, it is more likely that the distribution is not uniform across all years. To an extent this is true, our estimates are probably not as precise as we would like. It is possible that moves in the independent variables would help us understand the differences in magnitude across years in the same conflict. Because of our annualization of aggregate data, we are more making an argument that the mean level of the independent variables during that conflict and is causing the per-year value of deaths we see for that conflict. Thus, our inferences are more about average battle deaths among different conflicts rather than about changing casualty figures within conflicts. Any armed activity, sabotage or bombings carried on by independent bands of citizens or irregular forces and aimed at the overthrow of the present regime (Banks 2001). Any illegal or forced change in the top governmental elite, any attempt at such a change, or any successful or unsuccessful armed rebellion whose aim is independence from the central government (Banks 2001). Terrorism is violence, or the threat of violence, calculated to create an atmosphere of fear and alarm. These acts are designed to coerce others into actions they would not otherwise undertake, or refrain from actions they desired to take (National Memorial Institute for the Prevention of Terrorism 2004). A figure can be useful to understand these effects. In the interest of space, these have been omitted from the text, but are available on the author’s website. Most of the cases are civil war. See the chapter by Aydin and Gates for which institutions matter most in democracies.

4

Inequality, indivisibility and insecurity S. Mansoob Murshed

Civil war and organised collective violence is a complex phenomenon. Not only does it produce human tragedies on a colossal scale, but it creates humanitarian crises that are of concern to the international community, as well as contributing to global and regional insecurity. To economists, especially development economists, civil war is important as it is now recognised as a major cause of underdevelopment and the persistence of poverty (see Murshed, 2002a; Collier et al., 2003). The number of countries embroiled in a civil war seems to have waned after 1994 (Hegre, 2004). But the average duration of civil wars, standing at 16 years in 1999, does not seem to exhibit a significant downward trend (Fearon, 2004). The number of fatalities in civil war may be declining recently, but the numbers of refugees and internally displaced persons is rising (Human Security Report, 2005). For all of these reasons ending conflict or reducing its intensity must be a very high policy imperative within the development and international security agenda. As is well known, the contemporary rational choice literature on the origins of conflict and civil war offers two possible explanations for the origin of conflict, see Addison and Murshed (2002a) for a review. They are, respectively, grievance and greed. The former notion refers to historical injustices and inter-group inequalities that could be both economic as well as involving unequal political participatory rights. The latter concept emphasises the role of rents, which are occasionally lootable, in producing inter-group rivalry for their control; a competitive process of rent seeking that can descend into outright war. Here, the role of natural resource rents is crucial, as some types of resource rents are more easily appropriated.1 In practice, both motivations co-exist simultaneously; it is difficult to motivate groups to fight one another without historical grievances even when valuable resource rents are at stake. This is because the collective action problem in the sense formulated by Olson (1965) needs to be overcome so as to coalesce groups to fight each other. Similarly, wars motivated mainly by grievances can also degenerate into greed, once war produces new avenues for profit for the few. Thus, greed and grievance are inextricably intertwined. Furthermore, societies with well established

Inequality, indivisibility and insecurity 55 mechanisms for peaceful conflict resolution tend not to experience outbreaks of war. In this connection, it has to be pointed out that per-capita income levels tend to be the single most important factor in explaining civil war across nations. In other words the poorer and less developed a nation, the greater the risk of civil war (see, for example, Collier et al., 2003). Greed and grievance are secondary factors to per-capita income in explaining the risk of onset of civil war on an average across all conflict cases. A country’s economic status or relative affluence dominates all other factors in predicting the risk of civil war onset. This is because poorer countries tend to have correspondingly inferior institutions of conflict management, greater short-termism in decision making and less to lose from war. In other words, per-capita income and governance standards are strongly and positively correlated. This does not mean, however, that we can dismiss factors related to greed and grievance. Poverty also plays a pivotal role in this regard, as it makes soldiering a less unattractive option, and predation a more compelling survival strategy, see Addison et al. (2002). The state is an important actor in the discussion surrounding the causes of conflict and its persistence. Its role is to provide law and order, minimal security and a modicum of other essential public goods. Its economic policies are central to whether growth takes off or not, and recall that at higher levels of average income all forms of violent conflict are much less likely. The state’s activities are also crucial in maintaining non-violent means of conflict resolution. Above all, it has to be seen to be just; fair to minorities, allocate its relatively vast resources fairly across diverse groups, setting an example in this regard; and, be able to devise innovative sharing rules for the seemingly indivisible. The decline of state capacity is a major cause for conflict, often allowing other fractious groups or warlords to step into its shoes producing conflict, and ultimately requiring external intervention. In that respect, this chapter analyses two causes of conflict, setting the scene for the consideration of the role of the state and outside actors. My intention, in this piece, is to go beyond greed and grievance and focus on two issues that promote insecurity leading to collective violence: inequality and indivisibility. The former idea is elaborated on in the first section, and contains international and national dimensions. On the international side, the rising inequality between nation states may be contributing to international insecurity. During the high period of globalisation post-circa 1980, divergence in income levels between rich and poor countries has been progressing at a more rapid pace than at any time after the Second World War, contradicting the predictions of neoclassical growth theory which states that poor nations should grow faster than rich countries because of higher returns to capital in poorer countries. Inequality within a nation state can take on a variety of forms and I will concentrate on inter-group or horizontal inequality. Incidentally, this notion is also

56 S.M. Murshed central to the grievance explanation for contemporary conflict. The second section is concerned with indivisibilities. One aspect of indivisibility concerns the difficulties in arriving at fair divisions among contending parties that ensure peace. These factors help to explain the intractability of contemporary civil war and why conflict resolution is so difficult. Furthermore one could argue that indivisibility, and not just poverty or lowincome, underpins the coordination failures, misperceptions and short time horizons that contribute to the persistence of war, as well as being a reason for the absence of credible commitment to peace. Very often this is because of an excessive emphasis on the present, and the heavy discounting of the future costs that present profit and power entail. In some societies the future cannot be foreseen or is substantially discounted. This implies the non-recognition of the fact that the future and the present are indivisible or inter-connected. Finally, the third section attempts to synthesise these ideas about inequality and indivisibility and their connection to insecurity by way of a conclusion.

Inequalities As indicated above, the kinds of inequalities that promote insecurity are at two levels, one pertaining to the growing income disparities between nations and the other relates to group inequalities within nation states. This section considers these two types of inequalities in turn. The former type of inequality may create frustrations among the relatively deprived in poorer countries, as the glittering affluence of richer countries is all too apparent over the broadcast media and the internet. They may decide to turn on their ‘failing’ state. The latter type of inequality is important in promoting and coalescing inter-group rivalry. It is important to emphasise that this type of horizontal inequality is very different from the normally measured Gini coefficient of vertical inequality which arrays all citizens homogenously by income, and not ethnicity. Growing inequality in the presence of poverty lowers the opportunity cost of conflict. Global inequalities Globalisation is meant to be beneficial for the world’s poorer nations as participation in international trade and reforms aimed at attracting foreign finance are meant to narrow the gap between rich and poor nations according to conventional neoclassical economic theory. This process is called ‘convergence’, and means that the real income per-capita between richer and poorer nations moves closer to one another. Greater international trade is meant to be the engine that achieves this. Globalisation as a phenomenon describing a high degree of international economic integration in terms of trade and finance is not a new phenomenon. The era

Inequality, indivisibility and insecurity 57 before the first world war (1870–1913) was equally, if not more, globalised as our present age, see Murshed (2002b). Both episodes of globalisation involved economic interaction between developed (the North) and developing (the South) countries, and there are many similarities between the two periods. Is the globalisation game played on a level playing field? Or does greater power confer an advantage? Who has the greater voice in determining the rules of the game? My argument is that nineteenth-century globalisation in a colonial context set the stage for the second act, which is our contemporary experience of globalisation, notwithstanding the fact that a few actors or countries experienced role reversals (there is entry and exit from the developed-developing nation categorisation). The rules governing globalisation are fundamentally unfair to the developing world. Nineteenth-century globalisation was preceded by an industrial revolution in the UK and some other parts of North-Western Europe and the colonialisation of the present-day third world. Colonialisation was accompanied by the de-industrialisation of the then industrialised part of the South: India and China, see Baldwin and Martin (1999), and the colonial contract ensured trade policies favourable to the export of manufactured goods from the colonial power to the colony, stifling any nascent indigenous manufacturing capacity. In short, globalisation in the nineteenth century did not benefit the South which became poorer as the North grew richer. It, however, assisted the convergence of incomes towards higher levels for the Atlantic economy: countries in NorthWestern Europe and North America. Globalisation, therefore, increases the income gap between richer and poorer nations. To see this, it is worth examining historical income gaps between the richest and poorest nations. UNDPKO (1999) reproduces figures to show that this gap was only 3:1 during the dawn of the industrial revolution in 1820, rising to 11:1 by the end of the first episode of globalisation in 1913. More recently, it grew to 35:1 in 1950, rising slightly to 44:1 by 1973. After the commencement of the present round of globalisation, this figure has acquired a staggering magnitude of 72:1. This

Table 4.1 GDP per capita (1995 constant US$) growth rates Area/country

Annual average GDP growth (%) 1960–1970 1970–1980 1980–1990 1990–2000

All developing countries East Asia and Pacific South Asia Latin America and Caribbean Sub-Saharan Africa Source: World Bank (2002).

3.1 2.9 1.8 2.6 2.6

3.3 4.5 0.7 3.4 0.8

1.2 5.9 3.5 –0.8 –1.1

1.9 6.0 3.2 1.7 –0.4

58 S.M. Murshed is the most conclusive evidence of the process of marginalisation of developing countries during the two great phases of globalisation. Globalisation at present (the post-1980 period) has also marginalised much of the third world and low-income developing countries. Table 4.1 attests to that fact. Apart from East and South Asia, all the world’s lessdeveloped regions grew faster during the relatively less globalised era of the 1950s and 1960s. Yet all regions have expanded their exposure to international trade. While it is true that some middle-income developing countries as well as the most populous countries, India and China, are doing well out of globalisation, the benefits of globalisation are far more widespread in the South (Murshed, 2002b). In Africa, in particular, the era of globalisation is associated with huge development failure. Not only have incomes declined, but also other indicators of inclusion and wellbeing have deteriorated. This includes the return of old diseases such as tuberculosis, the AIDS pandemic, stagnating maternal mortality and literacy rates. In this connection it is worth stating that we can ascribe some of the lack of growth to factors other than globalisation (see Murshed, 2004) such as poor institutions, incorrect policies, disadvantageous geographical (tropical) location and excessive mineral/fuel endowments. Yet, trade or globalisation, remains central to explaining the convergence or divergence between rich and poor nations. Three different concepts that may be used to measure inter-country inequality (see Milanovic, 2005). All three methods arrive at the Gini coefficient, the most commonly used measure of income inequality. The Gini coefficient ranges from perfect equality (0) to perfect inequality (1), or in percentage terms from 0 to 100. The population whose relative inequality is being measured is categorised in several groups of equal size, five groups (quintiles) or ten groups (deciles) and so on. The first concept used to measure international inequality treats all countries, large or small, equally. This is known as category 1 inequality. According to this concept all countries for which data is available are arrayed according to per-capita income in comparable purchasing power parity (PPP) dollars. If, for example, there were a total of 150 nations, we would be effectively measuring the inequality across 150 different individuals. This is because each nation consists of a representative individual, whose income is that country’s average or per-capita income. Category 2 inequality is the same as category 1 inequality with the important difference that each national per-capita income is weighted by its population size (national population relative to the world). Thus, China is given a weight of about 0.2 as it accounts for a fifth of humanity. This may appear to be a reasonable procedure, but a serious flaw in category 2 type inequality measures is that most changes are accounted for by the alterations in populous countries such as China and India, and downwardly bias events elsewhere. Moreover, when nation states are the unit of analysis, each state should be treated equally, as each nation is an independent entity and

Inequality, indivisibility and insecurity 59 represents a unique policy experiment. Equal treatment for all nations means that each unit’s income should not be population weighted; rather there should be equal weighting, which implies no weighting at all. Both category 1 and 2 type inequality indices do not, however, take into account within-country inequality. In any nation there are income differences between inhabitants of town and country, the capital and places in the hinterland. Within larger states such as China and India there are huge regional disparities in economic performance and socio-economic conditions, consequently income levels differ. It might be better to focus on individuals rather than nations as the unit of analysis, even for assessing international inequality. This measure, known as category 3 international inequality, however, represents a tall order in terms of comparable international data collection. Since the late 1980s, however, household surveys have become more common across the globe. What do the three types of international inequality measures show us? Category 1 measures indicate a rise in inequality in the globalised era. From about 46 in 1978, the Gini coefficient has risen to over 54 by 2000. In contrast, during the less globalised period of 1960 to 1978 the Gini remained fairly stable between 48 and 46 (Milanovic, 2005). Clearly, this shows that globalisation produces winners and losers and does the converse of achieving income convergence. It also captures the collapse of output and national income, lowering income per-capita in Eastern Europe and the former Soviet Union following the demise of socialism. The decline in income in those regions may be likened to the fall in output brought about by a prolonged and intensive war. Moreover, while within region income differentials, within the OECD, nations continued to decline in the 1978–2000 period, it rose in all regions in the developing world except Latin America and the Caribbean. Recall that category 2 inequality is the same as category 1 except that it is population weighted. Category 2 measures will show a fall in international inequality in the highly globalised era, because of the impressive growth in China’s real per-capita income. Using this method the Gini for the world declines from 54.4 in 1978 to 50.1 in 2000 (Milanovic, 2005), indicating a decline in inter-country inequality. But category 2 measures are based on per-capita income as the unit of analysis. Thus, not only are category 2 indicators biased by what happens in China and India (large countries), but also any country’s income growth success in overall terms masks within country income variations along spatial or socioeconomic lines. The category 3 measure is based on household surveys, with households as the unit of measurement. Category 3 measures do not, unfortunately, allow us to go back farther than 1988. We would expect globalisation to have already produced winners and losers by 1988, intensifying inequality. Milanovic (2005) shows a rise in inequality (Gini coefficient) during the globalised phase from 62.4 in 1988 to 66.0 in 1993, falling back slightly

60 S.M. Murshed Table 4.2 Number of countries in different ‘social’ groups Type

1960

2000

Rich (OECD) Upper-middle Lower-middle Poor

41 22 39 25

31 8 25 67

Source: Milanovic (2005).

to 64.6 in 1998. Moreover, these figures would be higher, implying greater inequality, if, instead of PPP dollars, ordinary dollars, based on market exchange rates, were used. It therefore, captures the huge rises in inequality among citizens inside the former communist bloc. The urban–rural divide in inequality inside China and India are also explained, and indeed this particular factor greatly explains the gap in the measured Gini coefficient using category 2 and 3 methods. It can be argued that the category 1 Gini is still the best measure of inter-country differences in income, and especially whether there is convergence between incomes of poor and rich nations. This has patently not occurred; see Table 4.2 based on Milanovic (2005). Whereas in 1960 there were 22 upper-middle-class nations contending to join the wealthy group with two-thirds or more of the average income in the poorest OECD or rich country; by 2000, there were only eight such nations. The number of countries with between a third and two-thirds of the poorest OECD country income declined from 39 to 25 during the same period, indicating that the lower-middle classes, too, have been squeezed. The numbers of rich (OECD type) countries also fell, from 41 to 31; the rich man’s club is much more exclusive nowadays. Most importantly, the number of really poor nations (the fourth world or the lower classes) defined as having an average income less than a third of the lowest OECD national average income rose from 25 in 1960 to 67 in 2000. This is conclusive evidence that the world is becoming polarised into rich and poor nations since the beginning of moderate globalisation in 1960. We are, therefore, living in a world where there is a vanishing middle class in the sense of inter-country differences (Milanovic, 2005). Is rising global inequality, or inequality for that matter, cause for concern? Or should we only worry ourselves with absolute levels of poverty whether based on national standards or the international dollar a day measure of abject poverty. Clearly, this depends on our notion of justice. The current development-donor focus is on poverty alleviation. While this is a noble objective, citizens of the globe, including those residing in poor nations, are more aware of the differences in their own circumstances and capabilities compared to those of fellow human beings in rich nations. This is all the more so in a digital age, where satellite television

Inequality, indivisibility and insecurity 61 and the Internet is widespread. A Gini coefficient of 66 (category 3 inequality) means that the expected difference between two random individuals income is $9,200 based on an average world income of $7,000.2 It also implies that accidents of birth, in terms of nationality not social class, can cause a $9,200 earning difference. The same figure would be $7,560 with a Gini coefficient of 54 (category 1 inequality). Such levels of inequality are truly staggering; they would be intolerable in most Western democracies. Countries with national Gini coefficients around 60 such as Brazil and South Africa are ridden with strife, usually taking on the form of criminal violence. I have argued elsewhere (Murshed, 2002a) that a viable social contract with agreed upon rules for redistribution, is necessary to contain dissent and open conflict in developing countries. This concept of the social contract has its international counterpart too, one which we might refer to as the development contract, that is needed to sustain world peace. Growing global inequality since 1980 and the end of the Cold War has undermined this development contract. This is all too apparent to the public in poor countries, and has encouraged states to adopt a more repressive stance with its citizens rather than an accommodative approach. Thus, deterrence (repression) has come to dominate assurance (accommodation) in the actions of the state in many developing countries. Horizontal inequalities within the nation state Relative deprivation – the perception by one or more parties that they are unjustly treated – can be an important cause of civil war. Besançon (2005) has demonstrated that growing inequalities contribute to rebellions, but declining inequalities may contribute to ethnic conflicts. Of course, many conflicts have an ethnic, as well as a rebellious element. Many conflict societies are characterised by large inequalities in access to the productive assets necessary for livelihoods and in public spending on economic and social infrastructure and services. The importance of horizontal inequalities between groups, classified by ethnicity, religion, socio-economic class, etc., as sources of conflict is potentially important, see Murshed and Gates (2005) and Stewart (2000), for example. This subject is seriously underinvestigated by rational-choice researchers because of the paucity of data. Horizontal inequality can be much more crucial compared to vertical (or income class-based) inequality because it helps to overcome Olson’s (1965) collective action problem. Three dimensions of horizontal inequality are noteworthy: Discrimination in public spending and taxation. Discrimination in the allocation of public spending, and unfair tax burdens, lead to serious unrest. Grossman (1991) gives us a theoretical model of insurrection against the state by the peasantry reacting to over taxation, where the state is a tax-farmer interested in maximising the income of the rentier class.

62 S.M. Murshed Discrimination in the allocation of public employment is particularly resented in societies in which public employment represents the principal avenue for personal advance. In addition, the over taxation of smallholders encourages insurrection, and indigenous peoples often face discrimination in access to schooling, health care and public-sector jobs; many of these factors are present in Nepal’s current civil war (for example, see Murshed and Gates, 2005). Where there are inter-group fiscal transfers, which may take the form of spending on education and health for disadvantaged groups or including them in government employment, commitment to the transfer by those in power may be imperfect. This lack of credibility of the transfer can eventually lead to civil war. High asset inequality. Agrarian societies with high income inequality – for example, El Salvador, Guatemala, Nepal, the Philippines and Zimbabwe – have high asset inequality and are very prone to conflict. In these societies, agrarian elites use their collateral to further leverage their existing wealth through a financial system that they control by means of family/business cross-holdings. Asset redistribution such as land reform to lessen inequality is more difficult than public finance reform. Besançon (2005), however, points out that purely ethnic conflicts, as opposed to revolutions and genocides, are more likely when a greater degree of income equality3 has been achieved between contending ethnic groups. Inclusion in the political process is more crucial to preventing this type of conflict, which do not usually take the form of civil wars4, as the state is not involved. Economic mismanagement and recession. In Africa, Latin America and the former Soviet Union, conflict ridden countries have also suffered prolonged economic mismanagement and growth collapse. Successive IMF and World Bank supported adjustment programmes in DRC–Zaire, Somalia, Russia and elsewhere not only proved incapable of promoting economic recovery, but given the level of corruption within the state, themselves became targets to be captured by elite groups. Economic mismanagement is often associated with an uneven and unfair distribution of the burdens of subsequent adjustment; public spending benefiting the elite and the military is protected, often favouring particular ethnic groups, with the burden of adjustment placed on expenditures of value to the poor and disadvantaged groups. Despite the existence of horizontal inequalities violent conflict is unlikely to take hold if a country has a framework of widely-agreed rules, formal and informal, that govern the allocation of resources and the peaceful settlement of grievances. Such a viable social contract can be sufficient to restrain, if not eliminate, opportunistic behaviour such as large-scale theft of resource rents, and the violent expression of grievance. Conflict-affected nations have histories of weak social contracts, or a once strong social contract that has degenerated. This weakness is in many instances a legacy of colonialism which institutionalised mechanisms

Inequality, indivisibility and insecurity 63 favouring settlers over indigenous peoples (Guatemala, Zimbabwe, South Africa); divide and rule favouring one ethnic group over another, as in Rwanda; market controls to create rents for settlers to the cost of locals (Mexico, Brazil, Zimbabwe); and the expropriation of land and resource rents (Angola and the Belgian Congo). Pre-colonial ethnic rivalry over territory and assets, the case in resource-scarce countries such as Afghanistan, Somalia and Sudan, and the failure of long-standing independent states to strengthen mechanisms of political representation, notably Ethiopia, Haiti and Liberia, also lie behind weak social contracts (see Nafziger et al., 2000). A single ethnic group, or a subset, often assumed power in the immediate post-independence era, subjugating others and concentrating the fruits of state power – public employment, other public spending and resource rents – into its own hands. A final complexity in fatally weakening social contracts was the interaction of these domestic factors with external events, notably the Cold War, which provided financial and ideological succour to ruling elites and rebels. The net result of these processes is the accumulation of grievances within the context of a disintegrating social contract that would otherwise have provided the rules of the game to govern the distribution of the social pie and to achieve peaceful conflict resolution. A degenerating social contract is often a product of deliberate policies pursued by those who control a fractious non-benevolent state, as they are interested in the self-aggrandisement of certain groups. Combined with palpable horizontal inequalities, a declining social contract can encourage populations to revolt against a repressive state that merely attempts to deter these grievances.

Indivisibilities As indicated in the introductory section, the duration of civil wars shows little signs of diminishing. Also, as Walter (2001) and Wood (2003) have suggested, peace agreements that end civil wars are notoriously unstable in that they are often not implemented, or break down, as was dramatically the case in Angola. This is much more the case than in inter-state wars. One reason that such conflicts continue to persist could be certain indivisibilities in perceived shares of power and income in the peace settlement, as well as the inability to correctly infer the value of path dependence (the future depends on present actions). The former problem mainly relates to the problems of sharing the post-war economic pie or the peace dividend; the latter concerns credible commitment problems to the peace agreement itself. Both of these concepts pertain to the durability or fragility of peace agreements. This section considers the two factors mentioned above in turn. It should be noted that both these problems may encourage armed factions, the state and rebel groups, to favour belligerence or deterrence strategies in contrast to policies favouring assuring the other side, which require longer time horizons and compromises over seemingly indivisible

64 S.M. Murshed issues. Strategies of deterrence as well as uncompromising attitudes focus on the present, in contrast to strategies of assurance which require a very long time horizon, something that may be absent from the calculus of political actors in low-income countries. Indivisible shares Fearon (2004) points out, that of all types of civil wars those with secessionist tendencies and ‘sons of the soil’ dynamics are the most difficult to resolve, and tend to last the longest. This could be because of an overlapping interest and attachment to the inviolability of land and territorial sovereignty by both parties to the conflict. Certainly other causes such as the ready availability of easily lootable narcotic or gemstone revenues that help finance conflict, or misperceptions about the chances of outright military victory, are important in prolonging conflict. But the indivisibility of war aims, symbols or land can also make certain civil wars intractable. Wood (2003) highlights indivisibility as a major impediment to peace deals. This arises when territory, symbols or revenue in a post-conflict situation cannot be divided up so as to achieve peace. The problem can be most acute when religious sites such as Harem El Sharif or Temple Mount in Jerusalem are involved. Also, considerable difficulties arise when it is problematic to achieve compromise over a war aim such as land reform (Nepal and Colombia), or deep constitutional change (future of the monarchy in Nepal). There can also be seemingly irresolvable disputes over post-war power sharing and the allocation of offices in a post-conflict government. Secessionist wars where territorial sovereignty is contested can also be tricky to resolve. Compared to these, disagreements over sharing resources may require less challenging solutions. The theoretical literature on sharing and division offers us several insights. For example, Brams (2005) points out several allocation rules for a single divisible good, many divisible goods and several indivisible goods. All of these have implications for durable peacemaking involving compromises over issues and post-war economic stakes. If a peace agreement and the divisions and compromises it entails are perceived to be unfair, then the deal itself will not be robust. Sharing in this regard must be equitable as well as efficient. That is why envy-free allocative outcomes are so important. In an envy-free outcome each participant does not regard the allocation achieved by another player to be superior to what he/she has achieved. All the various allocative mechanisms considered by Brams (2005) require monitoring or intervention by an outside agency, a mediator and/or external power. This is all the more so in the case of allocations in a post-war situation. In the case of a single divisible good the analogy with cake cutting is applicable. This may, for example, concern the division of the post-war peace dividend. The application of the envy-free criterion may entail

Inequality, indivisibility and insecurity 65 several slices or divisions that may be inefficient and in excess of the number of parties to the conflict. This will be all the more true if what is being divided up is not homogenous. One can visualise situations to do with the division of the expenditure of post-war aid and the dividing up of land that may require a great deal of parcelling. A second situation considered by Brams (2005) entails several items to be divided, each of which is in principle divisible. Peace negotiations usually involve several issues, including regional autonomy, sharing of resource rents (such as oil revenues in the Sudan), constitutional changes, power sharing in the federal government and so on. Typically these issues will involve a long period of extended bargaining. The procedure behind the settlement, if reached, is described as the adjusted winner mechanism. Negotiations on the issues may involve placing upper and lower bounds on the values of each issue, bearing in mind that assigning pecuniary values is more amenable in quantifiable matters such as resource rents rather than for non-monetary matters involving status such as who should be President. Each side will allocate weights on the different issues at hand, and given that each side has a similar number of bargaining chips, each party will win on some of the disputed issues. These will tend to be in areas most highly valued by the concerned protagonist. So if regional autonomy is more highly prized by a rebel group compared to resource rents, they will put a higher weight on it and secure that goal under the adjusted winner mechanism. But one side can end up with wins on many high valued issues, and the consequent allocation could be inequitable to the other side. So this mechanism requires an equitability adjustment. Basically, this means sharing on high-valued issues where the two sides preferences are close, or the weights assigned to them, out of their bargaining chip allocation, are similar. So if the government and the rebels assign a close and high weight to resource rents, they must share these. In other words, if the government and the rebels both value resource rents highly, one side cannot equitably be allowed to be a sole winner. There has to be a revenue sharing mechanism on this issue. Other issues, where values diverge considerably, tend to be winner take all, based on which side places the higher value. This adjusted winner mechanism gives both sides an allocation which is roughly equal and more than 50 per cent of the assigned weights from the bargaining chip pile. The problem with applying this equitability included adjusted winner mechanism is that many issues are not easily divisible, such as, which side gets to first occupy a rotating post-war presidency. A further difficulty can arise if the two sides do not have similar bargaining power, something that external actors need to engineer. Third, and most importantly, Brams (2005) considers allocating several indivisible issues. Once again external intervention or mediation is required. The allocation of indivisible goods requires the application of the envy-free principle for any allocation to endure. And, a unique envy-free allocation may not be Pareto-efficient. Pareto efficiency means that one

66 S.M. Murshed side cannot be made better off without making another side worse off.5 One can make an envy-free allocation Pareto efficient by improving the utility of one side without lowering the utility of the other. But such allocations may not remain envy-free as one side could have a lower allocation of relatively more highly-prized items (yielding the same utility) that are being allocated, and consequently resent the other side’s allocation. A similar argument can be made about a maximin allocation being envypossible. Consider an application of the envy-free principle to the elections held in Iraq in January 2005. A criticism of the method adopted in that Iraqi election, for example, could be that the electoral mechanism (one person-one vote instead of representative bodies of each community) is not envy-free for the minority Sunni community, and has, therefore, not enlisted their full cooperation. Generally allocations involving indivisible items that are more qualitative are more difficult to achieve. The answers, in the more intractable cases, must lie in sharing, equal user rights and other ‘federal’ arrangements. Wood (2003) considers non-cooperative strategies of actors in a conflict, and whether their strategies to fight or compromise are self-enforcing without third party mediation. The decision to compromise is based on the pay-off in the peaceful state, as well as beliefs about the strategy which will be adopted by one’s opponent. There also has to be bargaining over the share of the post-war pie that each side gets. The Nash equilibrium can involve either fighting or compromise. If each side’s expected post-war share is greater than what they can get from fighting, feasible compromise equilibria exist. But that depends upon beliefs about the other side’s strategy. The feasible compromise equilibrium and the sharing it involves may not coincide with beliefs about the opponent’s strategy. In general, there will be an optimal share of the post-war pie for each side which will maximise the robustness of a peaceful settlement (that is the agreement lasting or being self-enforcing) given beliefs of the two sides about each other. Within each group there may be factions with more pessimistic views about their opponent’s strategies. This will depress the value of any share of the post-war pie. Indivisibilities regarding the issues contested, and the post-war pie, also devalue the expected worth of any share of the post-war settlement, making self-enforcing compromises difficult. In more extreme cases this may require external intervention in the form of bribery to increase the total size of the potential peace dividend as in Sri Lanka, so there is simply more to be shared. In the case of sites and symbols, steps have to be taken to encourage sharing and envy-free access through con-federal structures. This may require diplomacy and, in some cases, coercive intervention by external powers.

Indivisible periods and commitment problems Another form of indivisibility arises when the future is heavily discounted, and when the future costs of current actions are similarly undervalued.

Inequality, indivisibility and insecurity 67 This may lead to problems of commitment to negotiated settlements even when they are Pareto optimal, that is when each side is better off in a state of peace. In most situations, war is irrational and inefficient (non-Pareto optimal), as pointed out by Skaperdas (2002). Why is, therefore, the credibility of the commitment to peace treaties so fragile? There could also be misperceptions about the benefits of war or an overestimation of the prospects of military victory. To deal with misperceptions first, the most obvious candidate that prevents peace in this category of explanations for civil war persistence is an overestimate of the probability of military victory, see Collier et al. (2004) in this connection. The same authors also emphasise that the state of war may also be highly profitable for one or more of the belligerent groups. As indicated earlier, this is likely in the case of contraband substances and lootable minerals such as alluvial diamonds. Fearon (2004) has also pointed out that secessionist wars with ‘sons of the soil’ dynamics are notoriously difficult to resolve as noted above because of indivisibilities of objectives. The commitment problem to an agreed peace treaty is also a serious concern. This difficulty arises when it is in the interest of one or either side to renege on the promise of peace, and the actions that peace involves. In that situation, commitments lack credibility. Azam (2005) considers one example where the intentions and incentives of protagonists are only known imperfectly to external peacekeepers. Rothchild (see Chapter 8) points out to the fact that power sharing arrangements, which guarantee peace temporarily, may in the long-run undermine the polity by creating new incentives to deviate from these arrangements in the future. This gives rise to time inconsistency problems involving high discount rates of the type I consider below, and formalised in Addison and Murshed (2002b). Sometimes agents or groups cannot commit credibly because there are no institutions or mechanisms upon which to anchor promises. For governments, this is more likely in the context of weak state capacity, as it is difficult for a state to guarantee pledges when its own legitimacy and power base is fragile. An aspect of the commitment problem that has received scant attention is the very high discount rates, or the short time horizons of the parties involved (Addison and Murshed, 2002b). In situations of poverty and high uncertainty, agents strongly prefer a dollar today to a dollar tomorrow. Although the absolute value of future peace may be much higher than that of continued warfare, the present value may be much lower when the discount rate is very high and there is an impatience to consume. High discount rates may also undermine power sharing arrangements as described by Rothchild (see Chapter 8). The same argument can be applied to reputation, a factor that is key to the credibility of peacemaking. Breaking an agreement damages future reputation, but with a high enough discount rate it might pay to renege because the cost comes in the future. Each

68 S.M. Murshed failure of the peace process raises the discount rates of the belligerents, thereby increasing the difficulty of making peace. Given the tarnished reputations of belligerents it is even harder to establish credible peace. The problem is particularly apparent in Africa where most indicators of political risk are substantially greater than elsewhere in the world. Solutions lie in directly increasing the cost of reneging on peace agreements and devising commitment technologies through institutional innovation, particularly at the international level. Improving the quality of peacekeeping forces is an urgent need, as is increased commitment to bringing war criminals to trial. With regard to commitment and commitment technologies there are three other factors that can be considered: the separation of economic life and politics, time horizons and institutional settings. Economics and politics. There may be situations when conflict and business entrepreneurs are one and the same, as in many cases in Africa. This makes the commitment to peace less likely to hold, compared to societies with a relatively stricter dichotomy between those who rule (politicians) and those who conduct economic affairs. This is because in the former case the political and economic interests are one, and clearly prowar. Economic interests in this instance centre around war-contracts and the harnessing of resource or illicit drug rents. In the latter case there is some room for competition between different interests; business activities such as the exporting of manufactured goods from Sri Lanka or Nepal may be disrupted by the war. Even when there are links between the two groups, the greater the institutional separation through parliament and the political process, the better are the chances for lasting commitments to peace. Time horizons. This turns out to be a crucial feature in individual decisions. When a future is seen to feasibly exist, this results in more peaceful attitudes. Generally speaking, investment, which only bears fruit in the future, requires a long time horizon. More secure and affluent societies tend to have a longer time horizon and will adopt strategies of assurance against potential rivals. By contrast, severely war torn, insecure and poorer societies have shorter time horizons, with a very strong preference for a dollar today compared to an uncertain prospect of more than a dollar in the future. Short-term income may be readily obtainable in a war situation, even if war destroys future earning prospects. In the language of economics, this is referred to as a high discount rate applied to future income, as opposed to the high value put on present consumption. All of this means discounting the future cost of conflict, as well as undervaluing the tarnished future reputation which arises from an excessive zeal for short-term profit. Furthermore, societies with faulty and degenerating institutions of governance and democracy tend to have a high discount rate, as the future is uncertain. New and fledgling democracies are often characterised by these high discount rates, as the future is uncertain due to

Inequality, indivisibility and insecurity 69 the fact that the political system may collapse. The state apparatus in this situation runs the risk of descending into kleptocracy. The important point here is that many groups in these situations are also characterised by similar, short-term mentalities, making them often prefer current profits in a war situation when compared to investing for a far greater income than peace might bring in the future. Also, investment in trying to bring about future peace can have substantial present-day costs in terms of foregone profit. Institutions of commitment. Even when all parties agree to and recognise the benefits of peace they need to credibly commit to peace and the conditions stipulated therein. Generally, this requires institutions that help parties to credibly anchor their commitment to the peace treaty. The fear of reversal in the context of poor commitment technologies, leads to a peace treaty being imperfectly credible. And if it is not credible, the peace agreement will not last. Leaders of various groups and factions will then tend to behave like roving bandits with no concern for the country, rather than akin to stationary bandits who have an encompassing interest in nurturing the tax base from which they obtain rent (see Olson, 1996). A poor environment for commitment often arises when the government or the rebel leadership’s power base is weak and/or lacks legitimacy. Solutions here lie in devising better mechanisms to engender credible commitment via institutional improvement. This includes better constitutional safeguards, greater respect for the rule of law and superior regulatory capacities. In this respect externally enforced commitment technologies via internationally enforced peace treaties may help, but consideration also has to be given to constitutional systems that lead to the separation of powers and multiple majorities rather than simple majoritarian outcomes that may exacerbate ethnic divisions in the future (Rothchild, see Chapter 8). This may also deter the state or party in power from carrying out acts of violence on a large scale as considered by Aydin and Gates (see Chapter 5).

Conclusions I have argued that inequality both between countries, and within groups at the level of the nation state, can induce an insecure environment that may lead to open conflict. I also posit that indivisibility in terms of post-war stakes, envy-free allocations and the inability to connect the future to the present perpetuate wars and prevent lasting peaceful settlements. In doing this I am going beyond the greed versus grievance debate on the causes of conflict. There are signs that the rational choice literature is also inching away from these two explanations, as competing theories of conflict, to a position where these motives complement each other under conditions of poverty and underdevelopment. The current emphasis in the cross-country empirical (econometric) literature on conflict is on income per-capita as the

70 S.M. Murshed most robust explanation for conflict onset. Low-income countries do tend to have poorer governance indicators and institutional quality. And those among them with a high endowment of lootable gemstones and illicit narcotics have a ready made source for financing civil wars which are, on average, seemingly getting longer. These generalised propositions based upon cross-country regressions are, however, too broad to inform the policy maker, particularly when the unit of measurement for conflict is often suspect in a cross-country regression. In emphasising the role of inequality and indivisibility, I am searching for plausible explanations for conflict and its persistence over time, respectively. Neither of the two phenomena can be adequately proxied in cross-country regressions. True, there are adequate measures of global inequality. But data on horizontal inequality is characterised by its paucity. At best we have regional and spatial data on countries which allows us to approximate group inequality because some communities are more geographically concentrated (see, for example, Murshed and Gates, 2005). More work needs to be done in gathering data on horizontal inequality. Measuring indivisibility will be well-nigh impossible in a crosscountry framework except on the basis of some ad hoc and subjective coding practice. It is worth stressing the fact that good econometrics needs to be informed by theory. In the case of the conflict literature this practice is very ad hoc as far as economic theory is concerned, but superior when it is related to theory in political science. I believe group inequality and the indivisibility of post-war shares are compelling theoretical arguments for conflict onset and the continuation of conflict. They are best studied at the level of an individual conflict, not just through the method of thick description, but also where use is made of all available economic statistics over time. The message to policy makers is clear. Most conflict ridden countries have weakened state capacity, something that does not allow the state to be regarded as an impartial actor in all of the state’s traditional conflict prevention functions. The decline of state capacity also implies a weakened social contract between groups, as the state cannot engage credibly in redistributive functions between groups and help resolve disputes peacefully. It is tempted, more often than not, into acts of repression or deterrence rather than strategies of assurance which implies accommodating grievances. So some external intervention is required to restore peace and security, including re-building state capacity. Tackling horizontal inequalities are key to preventing and ending conflict. This means poverty reduction, growth, greater political participation and improved governance. Making peace settlements last implies that the indivisible has to be made divisible through external intervention in the form of diplomacy and coercion which encourage sharing, as well as greater economic aid to make the peace dividend more palpable. It also means strengthening institutions and

Inequality, indivisibility and insecurity 71 creating mechanisms for the separation of power that help anchor commitments. Furthermore, post-war allocations need to be envy-free in order for them to endure.

Notes 1 For example, it is easier to appropriate mineral-based commodities rather than agriculturally-based goods, see Murshed (2004) and Addison and Murshed (2002a). 2 This figure is obtained by multiplying the Gini coefficient of 0.66 by twice the mean world income, $7,000. 3 Note that income equality is different from asset equality, which concerns wealth. 4 Such as Hindu–Muslim riots in India, or Christian–Muslim violence in Nigeria or Indonesia. 5 This is, however, consistent with one person having everything and another person nothing in a two person world.

5

Rulers as mass murderers Political institutions and human insecurity Aysegul Aydin and Scott Gates

Why do rulers target the ruled? Mass killings orchestrated by rulers can be traced back to the beginnings of history. Even in the twentieth century – an age of carpet bombing, blitzkrieg, and two atomic bomb attacks – the dominant threat to human security has not been the perpetrators of interstate violence, but governments engaged in attacks on their own civilian population. The dreadful litany of cases of genocide and politicide in the second half of the twentieth century includes inter alia: Suharto’s Indonesia 1965 (Dwyer and Santikarma, 2003); Mao’s China 1958–1961 and 1966–1969 (Chang and Halliday, 2005); Rios Montt’s Guatemala 1981–1983 (Grandin, 2003); and Pol Pot’s Kampuchea (Midlarsky, 2005). More people have died at the hands of their own state than those killed in either civil or international wars (Rummel, 2005).1 Davenport’s chapter on the lethality of different forms of conflict clearly demonstrates that mass killings for political and ethnic reasons have been the most deadly form of state behavior in modern history when compared to civil and interstate wars. Despite this gruesome fact, there has not been adequate scholarly attention on the topic. The adverse implications of human insecurity on the social fabric of states call for more extensive research in this area to better understand its origins. We attempt to do so in this study. The idea that the state is the dominant threat to its people appears to stand in stark contrast to Hobbes’ view of human security as deriving from the social contract between ruler and the ruled. Hobbes’ Leviathan has absolute political power because the individual transfers her rights to the sovereign after careful consideration of the economic and social consequences of the state of nature. Intermediary associations such as parties, ethnic or religious groups, and parliaments – unless they are authorized by the sovereign power – are unlawful: they hamper the proper functioning of the state by claiming allegiance from the individual and thus manipulating her free reasoning (Boyd, 2001). Though Hobbes’ gloomy view of pluralism seems to preclude the limitation of political power, the social contract recognizes that the key threat that would emerge once the establishment of

Rulers as mass murderers 73 a state settles the “war of all against all” is the state itself. Therefore, the social contract contains important guarantees against acts of the Leviathan that may threaten human security. Hobbes recognizes that the individual has a “right to disobey” if the Sovereign issues orders to the effect that one’s physical existence, economic survival, or individual freedom are threatened. According to Steinberger (2002), if the existence of the Leviathan contradicts the end that it was created for, the social contract is null; the state dissolves and the individual has the right to protect herself against those that no longer constitute a state. Thus, Hobbes views human security as the backbone of state legitimacy and the exclusive goal of the central authority. Cases where civilians are targeted by their own governments indicate that the deep-seated fears of early theorists are not misplaced. Rulers have, all too often, knowingly chosen policies that spawned economic and political crises and produced public “bads” such as corruption, human rights abuses and human insecurity. Indeed, “government is potentially the key threat to any group, as governments are usually the actors that commit genocide. Whoever controls the government is protected from extermination and those who are excluded from government are vulnerable to destruction” (Saideman and Zahar, Chapter 1: 8). It is worth asking why some rulers govern for the well-being of their society whereas others deliberately seek policies to destroy their people. When and why do rulers turn against civilians and formulate policies that cause the death of their people? These are the questions that we explore here. To understand the origins of mass killings orchestrated by political leaders – Hobbes’ Leviathan – we focus on the ruler whose strategic objective is to stay in power, and who uses (in)security as a policy tool to solidify her domestic political standing. We argue that aggregate categories of democracy and autocracy by themselves are insufficient to explain why governments engage in mass murder. A more fruitful approach is to analyze how certain dimensions of political institutions affect the preferences of the rulers and their capacity to formulate and implement murderous policies. We contend that the structural relationship between mass killing and broad categorizations of democratic and autocratic institutions is weak. Autocratic institutions do not foreordain mass murder. Hence, a pure structural approach that dichotomizes regime type and standardizes rulers’ preferences across categories cannot explain why regimes with some democratic features, such as illiberal democracies, fail to achieve liberal governance and sometimes turn violent (Zakaria, 1997).2 When Milosevic lent military support to Serb militias that targeted Albanians (1998–99), it was one ruler exploiting liberal institutions that gave him access to political power in order to produce illiberal policies. It illustrates how leaders manipulate and capitalize on nationalist sentiments in newly democratizing countries and target specific groups to sustain political support. A

74 A. Aydin and S. Gates more refined and productive institutional approach should elaborate on different dimensions of political systems and their impact on the quality of governance. Our framework puts special emphasis on the state and statesmen to understand the origins of state-organized violence against civilians. Yet, our model only offers a snapshot of a more complex process that leads to human insecurity. It is built on the assumption that an imaginary core, the state, is the actor that controls and applies all the violence that takes place in this context. This approach discounts the role of other domestic groups in mass killings either individually or collectively in connivance of the state and overestimates the state’s role on the outcome. For instance, Kalyvas in his study of the actors in civil wars observes that; “actors seeking power at the center use resources and symbols to ally with peripheral actors fighting local conflicts, thus making for the ‘joint production’ of action” (2006: 476). Cases such as Indonesian mass killing in 1965 show that violence can be perpetrated by domestic groups as part of an inter-communal feud. Events such as the Rwandan genocide in 1993–94 point to the possibility that states can recruit extremist groups to target fellow civilians labeled as “other.” Midlarsky, focusing on extreme cases of genocide (the purposive policy of exterminating a people as opposed to ethnic cleansing), demonstrates that state insecurity plays a driving role in explaining the incidence of such events. Together, the historical evidence shows that states are not the sole culprit of human insecurity: a more complex interaction takes place among the organizers of mass murder and local groups, which is a topic that we leave to future study. In the pages that follow, we examine the institutional roots of mass killings. We focus on how different political systems affect rulers’ policy choices. We argue that the way in which the political process is organized provides incentives for rulers to choose certain policies over others. Political institutions determine whether political leaders should pursue good politics or if they can survive politically with bad public policy. In order to develop our analytical framework, we apply work that has examined leaders’ incentives to promote economic growth in different institutional settings (Bueno de Mesquita et al., 1999, 2002, 2003; Bueno de Mesquita and Root, 2000; North, 1990; Persson and Tabellini, 2003). We draw on these to understand variation in leaders’ incentives to promote human security. A “political system” should not be treated as a dichotomous parameter or a one-dimensional continuum. Following Gates et al. (2006), we break political institutions into several dimensions including: institutional constraints on political power; the (s)election of the ruler; and the extent of popular participation in the (s)election process. Our model of mass killings also accounts for certain characteristics of a society that might affect leaders’ incentives to target civilians.3 Societal cleavages, mainly ethnic divisions, may provide incentives to rulers to politicize ethnic issues for the pursuit of political power. Hence, one

Rulers as mass murderers 75 should expect the political leadership in ethnically divided societies to be more prone to target civilian groups: the ruler can use the state as an instrument to control and suppress an ethnic segment in order to mobilize the broader public. Interestingly, this may also happen in democracies. Elections may increase the competitiveness of gaining access to political power, but will not in themselves safeguard the civil and political liberties of the individual. On the contrary, electoral competition may create incentives to “play the ethnic card”, which can be rewarded with electoral success in fractionalized societies. In this respect, mass political participation and competition between political elites for power are not guarantees against bad public policy. Mobilizing or capitalizing on societal sentiments may not only occur in ethnically divided societies, but in all systems that encourage political competition but do not sufficiently monitor the ways incumbents use their power. We believe that good institutional designs that impose certain limits on leaders’ access to political power can mitigate the effects of ethnic fractionalization on policymaking. Therefore, we offer an institutional analysis of human security, adopting a more nuanced conceptualization of the institutional environment than those offered so far by the literature. We analyze data on mass killings provided by Harff (2003). The empirical analysis includes 37 country-year cases of genocides and politicides between 1955 and 2001. The data on political institutions comes from Gates et al. (2006) – a study of the institutional matrix along several dimensions. In the sections that follow, we begin with a brief review of previous research that illustrates theoretical weaknesses in the existing literature. In the third section, we offer an institutional analysis of mass killings that fills important gaps in the literature, and formulate testable hypotheses. Then, we account for our research design and present an empirical analysis. Finally, we discuss how our approach contributes to the understanding of leaders’ incentives to target civilian populations.

Strategic versus structural explanations Scholars commonly assume that regime type determines the quality of governance: autocracies produce poor policies and democracies facilitate good policies. A similar dichotomy can be found in the literature on mass killings. In a recent work, Harff (2003) shows that autocratic polities are three and a half times more likely than full or partial democracies to experience mass killings for political and ethnic reasons.4 In Harff’s framework, limitations on the power of political authorities, and leaders’ access to power through competitive elections involving mass political participation are characteristics of democracies that decrease the likelihood of a genocide/politicide. Valentino et al. (2004) follow a similar approach with a three-category polity type and find that full democracies are 28 percent

76 A. Aydin and S. Gates as likely as full autocracies to engage in mass killings.5 Davenport, on the other hand, finds that democratic institutions have a parabolic relationship to the lethality of genocides/politicides that take place within a civil war environment.6 However, when the civil war variable is removed, this relationship is negative and linear suggesting that on average, democratic governments kill less if they are already engaged in violence against their civilians. Rummel (1995) in his study of mass killings emphasizes regime type as a major factor that explains political leaders’ behavior. Rummel argues that unlimited access to political power and the absence of effective veto players in policymaking allow the political leadership to stay in power in spite of – or because of – the insecurity of its citizens. Using factor analysis, Rummel analyzes the death toll in all types of mass killings by governments between 1900 and 1987, and observes that institutions affect leaders’ choice of deadly policies: as the regime types vary from democratic to somewhat democratic to authoritarian to somewhat authoritarian to somewhat totalitarian to totalitarian, there should be a virtual logarithmic increase in the number of people a regime kill. (1995: 6) Rummel examines a broad array of institutional characteristics. Using factor analysis, he constructs continuous indices that measure the penetration of regimes into individuals’ private sphere, and institutional constraints on the exercise of political power. Yet, recent research has largely focused on dichotomous or trichotomous categories rather than expanding on Rummel’s rigorous analysis of political institutions.7 Explanations relying on dichotomous/trichotomous indices of regime type are inadequate for two reasons. First, the dichotomous democracy – autocracy scale or the trichotomous scale that accounts for mixed and pure regime types, introduce an arbitrary cut point that does violence to the presumably continuous institutional incentives for good or bad governance.8 Second, it is not institutions per se that lead to repressive policies – there is not a direct causal link between institutions and good or bad public policies. Rather, the question is how rulers define their interests and preferences within different institutional environments. Hence, there is a tripartite relationship between political institutions, the ruler’s preferences, and poor or failed policies – including mass killings with the ruler as the key political actor. We see the institutional matrix as a continuous scale and multidimensional, and, therefore, examine mass killings for ethnic and political reasons from a disaggregated institutional approach.9 Whether rulers govern through fear or welfare depends on the incentives that arise given a particular institutional arrangement (Gates et al., 2006). In particular, institutional restrictions on leaders’ access to political

Rulers as mass murderers 77 power raise the costs of repressive policies that target civil rights and liberties. However, when authority becomes concentrated, the number of actors that can challenge policies due to their divergent preferences and political support base falls, which also decreases the costs of bad public policies. With this in mind, executive decision-making constraints constitute the key institutional dimension shaping rulers’ incentives. Effective limitation points to the presence of accountability groups in the system, which substantially increases friction in policymaking. Multiple political actors are unlikely to reach a consensus in formulating and implementing policies with human costs. In such an institutional environment, liberal governance leads to cooperation with other groups, expands societal consent and increases the legitimacy of the ruler. Repression of the masses, on the other hand, nurtures the relative strength of opponents and the power of rival ideologies, thereby eventually increasing the costs of policymaking (Drake and McCubbins, 1998). Therefore, we think that in an institutional environment which limits executive political power, the benefits of good governance outweigh those of intimidation, repression, and the use of force. In the following section, we will develop our analytical framework and derive testable hypotheses.

Institutional configuration and the strategy of political survival We start this section with the basic assumption that there is no natural harmony between the interests of the ruler and the ruled: political institutions tend to benefit one side or the other (Acemoglu and Robinson, 2001).10 Rulers in certain institutional designs can only stay in power if they promote the common good, that is, the political system works for the benefit of the ruled. In others, rulers have the incentive to engage in policies that produce private goods for themselves and their support base: the system works to solve the friction between the ruler and the ruled to the benefit of the ruler. Therefore, we focus our attention on the ruler, government, or the leader of the state. This theoretical framework aims to understand the link between different dimensions of the institutional configuration, certain characteristics of a society and the strategic decisions that the rulers make to survive politically. Executive constraints Rulers behave according to which institutional environment they make decisions within. Institutions provide incentives to choose certain policies over others. The quality of governance that the leader needs to achieve to stay in office is closely related to the political rules that define the limits to political power. Unlimited access to political power leads to strong incentives to maintain the uncontested power structure and govern through

78 A. Aydin and S. Gates repression rather than for liberty. In systems where the rulers’ hold on power is absolute and the risk of being ousted from office by a political opposition is low, rulers may target certain groups in the populace to politicize ethnic or political divisions and mobilize the broader public for the continuity of their political rule. Such policies can be used to suppress regime challengers and deny any form of liberty to political opponents. Repressive policies can also be intended to divert mass attention away from the policy performance of the regime and to group divisions. Political system affects preferences in governance by structuring leaders’ reliance on their constituents’ and other political actors’ support. Rulers’ limited access to political power through the presence of groups that can institutionally hold them accountable works as an institutional check – a safety valve against the formulation of repressive policies. In cases where institutions fail to empower a number of political actors in the policymaking process with the authority to block the action of the ruler, bad public policy including mass killings are more likely to occur. Most effective institutional arrangements require that there are multiple agents in the system, which are situated against each other with divergent preferences and check one another’s political ambitions. In this framework, policymaking requires the consent of a qualified number of agents whose power originates from an institutionalized framework (preferably constitutionally mandated) rather than from the mercy of the ruler or the ruling group. Policies aimed at hurting some groups of citizens are more likely to be vetoed and never reach the implementation phase in multiple-agent systems. Consensus and cooperation are harder to achieve in group decision-making, especially if the policies for which consensus is sought – such as mass killings, are likely to give way to situations with grave social, economic and international consequences.11 The institutional requirement of collective action forces’ cooperation on agents, leads to friction in policymaking and increases costs of negotiation, particularly for policies with adverse effects. Even though targeting civilian groups may remain an attractive option in order to eliminate opposition and divert public attention, the transaction costs of seeking the consent of other political actors deter the ruler from doing so.12 Accountability groups constrain the ability of the executives to engage in repressive policies also because they compel leaders to share information with other groups involved in the policymaking process, and provide effective public policy through transparent and formal procedures. Information-sharing and diversity of information sources make it harder to impose narratives that stigmatize specific groups in society. The information effects of group policymaking limit rulers’ ability to create myths about ethnicity and political beliefs, and diminish the benefits of mass killings to political survival. In this respect, leaders’ incentives to promote the common good or the common “bad” are determined in large part by constraints on executive decision-making. Executive constraints appear to be equally critical in

Rulers as mass murderers 79 democracies and autocracies. In order to achieve consolidated democratic governance without the risk of a rise in populist nationalism, institutional limitations of political power must coexist with mass political participation so that leaders do not exploit societal divisions in order to mobilize political support in elections (Gates et al., 2006). This institutional configuration discourages the ruler from undermining the workings of the system because it is a self-enforcing institutional arrangement without loopholes that can be exploited to produce illiberal policies. Burton et al.’s (1992) work on unconsolidated democratic regimes, and Zakaria’s (1997) influential argument on the de facto deviation of democracy from constitutional liberalism, lead us to feature the role of executive constraints as the critical dimension that serves to prevent rulers from abusing their authority and targeting civilians. This view is extended by Weingast (1997) who demonstrates that unconsolidated democracies in divided societies are particularly prone to violence and politics of intimidation. This discussion on the primacy of institutional constraints to human security leads us to our primary hypothesis with regard to genocide/politicide: Executive constraints: Greater constraints on the executive lead to a lower probability that rulers choose to commit mass killings. Political participation and executive recruitment To understand whether the ruler is motivated to serve the people’s interests rather than her narrow constituents for political survival, the extent of mass participation in the (s)election of the executive is another important dimension of the institutional matrix. A small winning coalition would mean that purchasing political loyalty of particular groups would be sufficient to stay in power. In this case, rulers might have the incentive to engage in mass killings to suppress political opponents and maintain the privileged access of these key groups to public resources. In systems with democratic overtones, masses participate in the electoral process, and leaders have to satisfy a qualified majority with successful economic and political policies rather than through coercion and repression. Nonetheless, even in political systems where leaders are answerable to a large group of people and face the threat of removal from office by public choice, there is no guarantee of the protection of a small minority without some form of institutional checks on executive powers. Populist politicians may rely on genocidal activities to curry favor from the majority and mobilize electoral support. Thus, we see that mass political participation will lead an autocrat to engage in repression to rally electoral support. Similarly, an illiberal democracy with high levels of political participation but without executive constraints is also prone to abuses of political power. Therefore, political participation is a dimension of the institutional matrix that would encourage illiberal governance.

80 A. Aydin and S. Gates Political participation: The higher the level of mass political participation is, the higher the probability that the ruler will choose to commit mass killings. We expect to find no relationship between the competitiveness of executive selection and the incidence of mass killings. Though the system might allow individuals to freely compete for the executive position, this does not serve as a guarantee for good governance once the competition is over. Without institutional checks on executive decision-making, there is no guarantee against elites once elected of using state power to target and eliminate their competitors. Executive recruitment: The openness of executive (s)election is not related to the probability that the ruler will choose to commit mass killings. Autocracies Limitations on executive authority are not only critical to the consolidation of democratic governance but they can also serve to prevent mass atrocities in autocracies. Though not as effective as constitutional arrangements, de facto limitations of political power can deter rulers from targeting certain groups in society. For instance, Peceny et al. (2002) argue that personalist regimes are the least constrained type of autocracies as opposed to military and single-party regimes. According to Peceny et al.: “The strategies pursued by dictators to stay in power, . . . render military institutions . . . incapable of doing much more than repressing unarmed civilians” (p. 19). The survival of a personalist regime depends heavily on the virtuosity of the ruler to eliminate, and divide any group and its supporters that can pose a danger to the regime. Therefore, splits within the governing elite are less likely to occur in one-man regimes, which prevent the formation of (de facto) accountability groups through causing human insecurity for supposed or potential opposition groups. Single-party and military regimes also rely on the loyalty of their members for political survival. However, one-party regimes need to mobilize masses to stay in power. In line with our argument on inconsistent democracies, this may give incentives to the ruling elite to exploit or mobilize political and ethnic divisions, thus increasing the likelihood of societal hatred and hostility. Though single-party regimes might resemble multipleagent frameworks where party members can check each other’s political ambitions, the goal of staying in power as a party may override the incentives to oppose and curb the formation of factions within the party, and lead to a consensus for policies that exploit ethnic/political cleavages for political survival. Also, autocratic institutions, unfamiliarity of the masses with multi-party elections, and entrenched political alignments among the

Rulers as mass murderers 81 public in durable single-party regimes, prevent the competitiveness of a second party, further decreasing the benefits of opposition to the party leadership. Military regimes are not answerable to a large winning coalition but the ruler has to satisfy members of the military elite to deter counter-coups. This type of regime needs to sustain cohesion within their small winning coalition. Defection might lead to a counter-coup and the loss of political power. For instance, Peceny et al. (2002) argues that “military regimes face somewhat more constraints because decisions need to be approved by a portion of the officer corps” (p. 17). Contrary to the low benefits of opposition in one-party regimes, if a member of the ruling elite succeeds in swaying the loyalty of the military to her side, political power can easily change hands. Such a system exhibits a winner-takes-all structure, whereby the benefits of a successful coup are so great that dissent is attractive and worth bearing the risk of losing (i.e. imprisonment, torture, execution, etc.). This strategic interaction between the officers and the leader (or the ruling cadre) can take on the qualities of a multiple-agent environment. Members of the military elite can function as de facto accountability groups whose consent the ruler regularly seeks in order to avoid dissent. In this respect, military regimes are autocracies in which multiple agents have to agree to a policy, unlike singleparty regimes in which opposition has no political benefits, and one-man regimes where potential opposition groups are already assimilated. Military regimes: In military regimes, rulers are less likely to choose to commit mass killings. Single-party regimes: In single-party regimes, rulers are more likely to choose to commit mass killings. Personalist regimes: In personalist regimes, rulers are more likely to choose to commit mass killings. Ethnic polarization In ethnically divided and polarized societies, ethnicity is often the most immediate societal cleavage along which political actors can mobilize support. This point is lucidly put by Gunther and Diamond (2001) in their examination of the political dispositions of ethnically oriented parties: The electoral logic of the ethnic party is to harden and mobilize its ethnic base with exclusive, often polarizing appeals to ethnic group opportunity and threat . . . The ethnic party’s particularistic, exclusivist, and often polarizing public appeals makes its overall contribution to the society divisive and even disintegrative. (pp. 23–4)

82 A. Aydin and S. Gates Snyder (2000) argues that political elites foster nationalism to achieve popular support among their supporters, particularly in newly democratized environments. In fact, Snyder convincingly claims that nationalism (which is the root of genocide) is not strong in pre-democratic societies and only develops with democratization as political competitors seek electoral support to access office through exploitation of popular sentiments. The targeting of particular populations on mostly ethnic dimensions occurs in autocracies as well as in democracies when populist rulers are relatively unconstrained. We expect to see that leaders in ethnically polarized societies are more likely to target civilians, all else being equal. As suggested by Reilly (2003), “. . . it is often easier to mobilize support by appealing to ethnic allegiances rather than issues of class or ideology, aspiring politicians have a strong incentive to mobilize support along ethnic lines” (p. 3). As such, ethnicity is a ready and easy-to-exploit societal cleavage: it provides incentives to profit politically from ethnic divisions, and to divert public attention from the overall bad performance of an administration. Also, the organization of daily politics along ethnic lines pushes the locus of political competition toward the extremes as rival parties or opposition groups respond in kind and a process of “outbidding” takes hold (Rabushka and Shepsle, 1972). Therefore, in an ethnically-charged political environment, we expect to see that there would be greater incentives among political actors to politicize identity issues which lead to increasing ethnic tensions and, in some cases, the outbreak of ethnic conflict and/or deliberate killing of civilians by government authorities. Ethnic polarization: The greater the ethnic polarization is, the higher the probability that the ruler will choose to commit mass killings. Based on our former discussion about illiberal democracies and leaders’ incentives to play the ethnic card to mobilize their support base in these institutional environments, we also formulate a hypothesis about the interactive effect of ethnic polarization and institutional limitations on executive power: less constrained leaders in ethnically divided societies would be more likely to produce policies that jeopardize the survival of opposition groups. In other words, good institutions that impose limitations on the executive mitigate the risk of human insecurity associated with ethnic fragmentation; Ethnic polarization: In ethnically polarized societies, the lower the degree of executive constraints is, the higher the probability that the ruler will choose to commit mass killings.

Rulers as mass murderers 83

Empirical analysis: limits of political power and human security Dependent variable and estimator choice We test our hypotheses on cases of mass killings targeting groups due to their ethnic backgrounds or political beliefs, identified by Harff (2003).13 In this dataset, there are 37 governments that have committed genocides/politicides and 570 country-years of mass killings in the time period 1955–2001. The definition of a genocide/politicide comes from Harff: Genocides and politicides are the promotion, execution, and/or implied consent of sustained policies by governing elites or their agents – or, in the case of civil war, either of the contending authorities – that are intended to destroy, in whole or part, a communal, political, or politicized ethnic group. (2003: 58 (italics original)) The dataset is designed as a cross-sectional time-series and the unit of analysis is the country-year. We create a dichotomous response variable marking whether a mass killing has occurred in a country-year. We use a random effects longitudinal logit regression to estimate the relationship between dimensions of political institutions and mass killings.14 Independent variables Our political institutions data on executive constraints, openness of executive recruitment, and political participation come from Gates et al. (2006). In this dataset, countries with a population over 500,000 are observed in consecutive time periods from 1800 to 2000 on three dimensions. Using the Polity IV and Vanhanen datasets,15 Gates et al. create three variables “describing systems along three authority dimensions” (p. 11). Executive constraints is the Polity IV dataset variable “decision constraints on the chief executive” and is a categorical variable ranging from 1 to 7 (Marshall and Jaggers, 2003). Executive recruitment measures the competitiveness of the (s)election of the executive. It is a categorical variable ranging from 1 to 4, that distinguishes between competitive selection, concentrated selection (e.g. forceful seizures of power, ascription, and designation), and dual systems where rulers that are (s)elected through concentrated and competitive processes coexist. Finally, the political participation variable is created using Vanhanen’s data on electoral participation and competition, and measures “the extent to which an election has a decisive impact on the selection of the executive” (p. 14). It is a continuous variable that ranges between 0 and 4.23 in our sample. We explicitly adopt Vanhanen’s objec-

84 A. Aydin and S. Gates tive coding of political participation to escape an inherent problem in Polity’s indicator of political participation. If any violence occurs during an election, Polity regards it as factional, and the regime is relegated to the middle ranges. Anyone wanting to study intrastate violence would then have violence on both sides of the equation. Vanhanen’s coding allows us to avoid this problem. In line with our theoretical discussion, we also disaggregate autocratic regimes in order to more accurately examine leaders’ incentives to kill their subjects within different types of autocracies. To determine regime type, we use Geddes’ (1999) data which include information on all authoritarian regimes for the period 1946–96. Geddes classifies autocracies as either single-party, military, personalist or hybrid types. To test for the possibility that rulers in military regimes might be less prone to mass killings given the de facto limitations on executive power, we rely on the trichotomized autocracy scale. Following Geddes, we include military and militarypersonalist regimes in the “military regimes” category, one-man regimes in the “personalist regime” category, and single-party, single-party hybrids and regimes showing certain characteristics of all three categories (triplehybrids) in the “single-party regime” category.16 We use dummy variables for each category of autocracy. Another key independent variable in our analysis is ethnic fractionalization. We hypothesized that cultural diversity will provide ready-to-exploit societal cleavages to the rulers to mobilize masses and divert attention from the policy performance of the regime. To test the link between ethnic composition of a society and the risk of mass killings we have adopted two different datasets. The first one is taken from Collier and Hoeffler (2004). Using data on ethnic composition, Collier and Hoeffler create a continuous scale ranging from 0 to 93, where higher values indicate high ethnolinguistic fractionalization (ELF) in a society. Second, we have also used Cederman and Girardin’s (2007) index N* of ethno-nationalist exclusiveness. This measure compares the ethnic composition of the elites in power and ethnic groups that are excluded from political power. Cederman and Girardin argue that civil conflict is more likely to occur under circumstances in which demographically important groups are excluded from the decision-making apparatus of the state. Peripheral ethnic groups are better able to challenge the groups in power depending on their demographic size. High scores on the N* index indicate that the size of the excluded group vis-à-vis the ethnic group that controls the state is demographically significant. Drawing from the same logic, we expect to see that higher values on the N* index will be associated with a higher probability of mass killings. This is mainly because, if the group opposing the state is ethnically dominant, elites in power will feel insecure and be desperate in consolidating their power by violently targeting rival ethnic groups. We have also created the interaction of the ethnicity variables with the executive constraints variable to test for our second ethnic polarization

Rulers as mass murderers 85 hypothesis. To do this, we first inverted the scale of executive constraints so that high values would point to less constrained leaders – and then multiplied this variable by fractionalization. Therefore, the highest values on the interactive effect indicate the least constrained leaders in ethnicallycharged societies. We also include three control variables in the analysis. First, Harff (2003) has shown that economic integration into the world economy substantially reduces the incentives to commit mass crimes. Therefore, we include trade openness in the models and adopt the data from Bueno de Mesquita et al. (2003).17 Second, Harff also argues that states which are members of international organizations fear the repercussions of human rights abuses and repression, and are therefore less likely to engage in violence against civilians. We include a binary independent variable in the analysis to control for international organization membership, using data from Harff (2003). Third, we include the natural logarithm of the total population of a country from the Penn World Tables (Hesten et al., 2002) because our political institutions data are collected for countries with a population of over 500,000. The population variable should control for possible threshold effects. Results Table 5.1 presents the random effects logit estimation of the relationship between political institutions, cultural diversity and genocides/politicides. Model I examines the link between three core dimensions of the institutional matrix from Gates et al. (2006) and mass killings, along with control variables measuring population size, and economic and political openness. In Model II, we add the ELF index and its interaction with executive constraints. In Models IV, V, VI, and VII (Table 5.2), we further elaborate on the link between ethnic composition of a society and the likelihood of genocides/politicides. In Model III, we disaggregate autocracies adopting Geddes’ approach. In Model I, the executive constraints variable is negative and highly significant, indicating that limits to political power effectively reduces the incidence of death by government. This finding is usually robust across all models that have been tested, suggesting that mass killings may become a tool for political survival in polities that do not limit the decision-making power of the executive. A closer look at the distribution of mass killings on executive constraints shows that in most regimes where genocides/politicides have occurred, institutional checks on political power are substantially limited (Figure 5.1). As is evident among Harff’s cases of genocide/politicide that have been committed by democratic governments, the distribution of mass killings is so skewed toward regimes in which rulers are relatively unconstrained in their decision-making, that it provides ample support for our argument.

0.877 (.030) 856

Scale parameter N

Notes 1 Robust standard errors are reported for all logit models. 2 Two tailed hypothesis tests; *p < 0.10; **p < 0.05; ***p < 0.001. a Ethnic polarization index uses data from Collier et al. (2004).

–20.586 (3.750)***

–19.029 (3.211)*** 0.857 (0.042) 780

–0.918 (0.288)*** 0.142 (0.438) 0.917 (0.301)*** 1.440 (0.289)*** –0.030 (0.017)* –0.023 (0.081) 0.060 (0.018)*** –0.001 (0.003)

Model II Ethnic polarization a

–0.755 (0.234)*** –0.353 (0.410) 0.765 (0.280)*** 1.733 (0.293)*** –0.018 (0.014) –0.081 (0.075)

Model I Base model

Executive constraints Openness of recruitment Political participation Log population Economic openness International organization membership ELF ELF and constrained leaders Single-party regime Military regime Personalist regime Constant

Cofactor name

Table 5.1 Random effects logit analysis: dependent variable, genocides/politicides

0.889 (0.031) 856

2.345 (10.258)* 2.176 (0.912)** 2.438 (0.961)** –22.609 (3.906)***

–0.718 (0.248)*** 0.253 (0.541) 0.911 (0.314)** 1.730 (0.336)*** –0.018 (0.016) –0.121 (0.082)

Model III Disaggregating autocracies

0.866 (0.041) 780

Scale parameter N

Notes 1 Robust standard errors are reported for all logit models. 2 Two tailed hypothesis tests; *p < 0.10; **p < 0.05; ***p < 0.001. a ELF indices from Collier et al. (2004). b ELF indices from Cederman and Girardin (2007).

–29.893 (5.275)***

–0.972 (0.288)*** 0.611 (0.454) 0.869 (0.328)*** 2.404 (0.434)*** –0.022 (0.016) –0.194 (0.086)** 0.055 (0.012)***

Model IV a Ethnicity

Executive constraints Openness of recruitment Political participation Log population Economic openness International organization membership ELF ELF and constrained leaders Constant

Cofactor name

0.857 (0.042) 780

–0.918 (0.288)*** 0.142 (0.438) 0.917 (0.301)*** 1.440 (0.289)*** –0.030 (0.017)* –0.023 (0.081) 0.060 (0.018)*** –.001 (0.003) –20.586 (3.750)*** 0.790 (0.101) 364

–23.506 (7.787)***

–0.767 (0.365)** –0.669 (0.690) 0.663 (0.415) 2.031 (0.636)*** –0.019 (0.026) 0.043 (0.113) 7.370 (2.558)***

Model V a Model VI b Ethnicity and constraints Ethnicity

Table 5.2 Random effects logit analysis: ethnicity and genocides/politicides

0.780 (0.114) 364

–0.767 (0.361)** –0.618 (0.720) 0.654 (0.413) 1.971 (0.670)*** –0.018 (0.026) 0.040 (0.115) 9.104 (7.060) –0.611 (2.264) –22.823 (8.116)***

Model VII b Ethnicity and constraints

88 A. Aydin and S. Gates Interestingly, there are 69 observations where a genocide/politicide took place even though there were substantial limits on executive authority.18 The fact that constrained leaders at times have resorted to violence in order to achieve their political goals runs counter to our expectations. One possible explanation for this is that Polity IV codes constitutional as well as non-constitutional accountability groups as limitations on executive authority. For instance, the authority of a legislative originates from the constitution whereas the authority of a council of nobles or military officers is not a constitutional practice in the sense that it might depend on tradition and inter-elite bargaining. To illustrate with cases from our dataset, Chile, Indonesia, and Pakistan were consistently autocratic states according to Geddes (1999) in the genocide/politicide episode, whereas Polity has assigned relatively high executive constraints scores to these countries for the same periods. Non-constitutional limits to political power, however, may at times be ineffective at safeguarding collective decision-making because there can be no basic law securing the rights and autonomy of the accountability groups. Given the probabilistic nature of any inference, there are several unconstrained autocrats that refrained from genocides and politicides. For instance, Cuba had the lowest possible score on the executive constraints variable for each year in the 1955–98 period according to our dataset but experienced neither genocide nor politicide. There are several such examples where completely unconstrained executives did not commit mass murder. This is expected because like all other theories, ours is not a 199

Country years of genocide

200

145

150

100

51

50

41 27 15 1 0 0

2

4 Ex ecutiv e constraints

Figure 5.1 Genocide/politicides and limits to political power.

6

8

Rulers as mass murderers 89 deterministic theory claiming that institutional constraints on the executive power affect the probability of mass killings. Therefore, we do not suggest that autocrats kill; we only suggest that, all else being equal, unlimited power can increase leaders’ incentives to solidify their domestic standing through violent means. As such, why leaders resort to such extreme political measures to stay in power is a multi-causal process in which political institutions are only one of many possible answers. This is why ethnicity is an important factor in our model in terms of identifying the possible threats that leaders perceive. In this regard, the institutional framework provides an opportunity structure in mass killings, whereas ethnic polarization provides the motivation to actually undertake it. In sum, both Model I and Figure 5.1 provide support for our theoretical expectations: unconstrained executives are more likely to commit mass killings. We conclude that in polities where policymaking requires the consensus of multiple agents, rulers’ access to political power is limited through the presence of accountability groups. Constitutional limits to power seem to increase transaction costs and diminish the utility of violent policies with the burden of negotiations as rulers seek a bargaining space within a multiple-agent framework. Limited power also breaks information monopolies because political actors with constitutional authorities are independent from the will (or mercy) of the ruler. The institutional autonomy from the head of the state protects the right to disagree and prevents against the state’s being used as a propaganda apparatus to mobilize masses (Snyder, 1991). Our expectations with regard to political participation are confirmed in most of the models: mass enfranchisement seems to increase the likelihood of mass killings. Broad political participation may be a source of trouble rather than a buffer against rulers. The finding suggests a plausible explanation of the ostensibly paradoxical observation that mass killings occur in illiberal democracies. Extensive mass participation in executive (s)election does not serve as a guarantee of human security. On the contrary, it encourages leaders to seek popular support, jockey for political advantage, and suppress their challengers through all available means. In such cases, electoral support may justify turning divisive campaign promises into actual policy, abusing the privileges of office, and using the state as a repressive machine that targets small minorities. The third dimension of the institutional matrix, competitiveness of executive recruitment, does not appear to be significantly related to mass killings. This result suggests that openness of office neither works as a safety valve against abuses of political power nor particularly encourages rulers to cause human insecurity. This is in line with our theoretical expectations. Increasing competition may lead elites to develop strategies to defeat their rivals which can be moved to the extremes with deliberate killings of supposed members of the opposition. However, political institutions condition the impact of competition on the strategies with which

90 A. Aydin and S. Gates political elites choose to settle accounts with their rivals. In this respect, the adverse effects of competition for political power on human security are mitigated through the political participation and executive constraints dimensions of the institutional matrix. Model III disaggregates autocratic regimes and thus expands the analysis of the relationship between polity type and mass killings. In the third section, we have argued that in single-party regimes the low probability of electoral success of a second party and the desire to stay in office and reap its benefits may discourage party members from dissenting. These characteristics effectively reduce single-party regimes to a one-man framework and eliminate de facto accountability groups in the form of political factions within party. In this respect, single-party regimes share similar institutional characteristics with personalist regimes given the low negotiation costs in policymaking and the ease of holding back information from the public. We have also claimed that military regimes are less likely to reach a consensus on repressive policies, which would substantially decrease the likelihood of genocides/politicides. The results of the analysis show that consistently autocratic regimes are more likely than institutionally inconsistent regimes and democracies to commit mass atrocities. Interestingly, the effect of a military regime on mass killings is positive and significant, which runs counter to our theoretical expectations. The explanation may be that institutional checks on executive authority alone may be insufficient. What may be more relevant is the connection between the institution checking executive authority and the general citizenry. Another way to illustrate the importance of consistently autocratic regimes is to examine the overlap between the genocide/politicide episodes and the periods in which a country has experienced a Geddes-type regime. Table 5.2 does just that, and presents strong support for our findings in Model III: there is nearly a perfect match between autocratic periods and the incidence of mass killings. In the majority of occurrences, rulers have made decisions to target their subjects within autocratic environments. Contrary to our theoretical expectations, military regimes appear to be no less likely than other autocratic forms to generate human tragedies. Interestingly, in the cases of Burundi, Indonesia, Iraq, Rwanda, Sudan, and Uganda, the change in the characteristics of the autocratic regime did not make these rulers less likely to target civilians. The consistently positive effects of different autocracy types suggest that de facto limitations on political power do not constrain rulers from targeting civilians. This is fundamental to the logic of our argument on institutional design – executive constraints in particular – and also Zakaria’s (1997) strong emphasis on constitutional liberalism as the guarantor of good governance: constitutionally designed constraints on political power rather than unstructured negotiation and competition between political elites are the factors that give rulers incentives to respect human rights. The economic and political openness variables appear to be unrelated to

Rulers as mass murderers 91 Table 5.3 Autocracies and genocides/politicides Country

Genocide/politicide episode

Afghanistan

1978–92

Autocracy type

Personalist Single-party Algeria 1962 – Angola 1975–94, 1998–2002 Single-party Argentina 1976–80 Military Bosnia 1992–95 – Burma 1978 Triple-hybrida Burundi 1965–73, 1988, 1993–94 Military/Single-party Military Cambodia 1975–79 Single-party Chile 1973–76 Military/Personalist China 1959, 1966–75 Single-party Congo (Kinshasa) 1964–65, 1977–79 Personalist El Salvador 1980–89 Military/Single-party Equatorial Guinea 1969–79 – Ethiopia 1976–79 Military/Personalist Guatemala 1978–90 Military Indonesia 1965–66, 1975–92 Personalist Triple-hybrid Iran 1981–92 – Iraq 1963–75, 1988–91 Personalist Single-party/Personalist Personalist Pakistan 1971, 1973–77 Personalist Philippines 1972–76 Personalist Rwanda 1963–64, 1994 Single-party Military Yugoslavia (Serbia) 1998–99 – Somalia 1988–91 Personalist South Vietnam 1965–75 – Sri Lanka 1989–90 – Sudan 1956–72, 1983– Military Personalist Military Syria 1981–82 Triple-hybrid Uganda 1972–79, 1980–86 Personalist Personalist

Regime duration 1973–78 1979–93 1976– 1976–83 1962–88 1966–87 1987–93 1975–79 1973–89 1949– 1965–97 1948–84 1974–91 1970–85 1949–65 1967– 1963–68 1968–79 1979– 1971–77 1972–86 1962–73 1994– 1969–90 1958–64 1969–85 1989– 1963– 1971–79 1986–

Notes The data on genocide/politicide episodes came from Harff (2003) and include all available cases. The autocracy data are adopted from Geddes (1999). Columns 3 and 4 report only the polity periods that overlap with a genocide/politicide period. In this respect, most countries have longer autocratic experiences which are not included in Table 5.1 because Harff does not report any mass killing cases for these periods. Iran, Equatorial Guinea, and Sri Lanka are not in Geddes’ dataset. Though South Vietnam had a personalist regime between 1955 and 1963, the politicide targeting supposed supporters of Viet Cong has not taken place before the personal rule was over. Also, Geddes classifies Yugoslavia as a single-party regime between 1945 and 1989 which again does not correspond with the timing of the ethnic cleansing of Muslims and is, therefore, not included in the table. a Triple-hybrid refers to polity types which contain characteristics of all three autocracy types.

92 A. Aydin and S. Gates mass killings. The dissociation between economic integration and genocides/politicides is surprising given our expectations that rulers have an incentive to avoid domestic violence and political instability that would increase risks to traders and investors. We attribute this null finding to the mixed relationship between socio-political instability and trade openness. On the one hand, a ruler in a closed society may seek to solidify her domestic standing by resorting to violence knowing that international pressure would be inconsequential to her political survival. A ruler, on the other hand, may seek to improve external economic relations and attract foreign trade and is, therefore, less likely to weaken the economic and social fabric of the society through illiberal policies. We think that it is these competing relationships that explain the null effect of trade openness on the incidence of mass killings. With regard to the effect of international organization membership, its non-significance confirms Harff’s conclusion. In Models IV, V, VI, and VII, we have looked closer into the relationship of ethnic polarization and genocides/politicides. We have used two different ELF indices from Collier et al. (2004) and Cederman and Girardin (2007). In both Models IV and VI, the individual impact of ethnic diversity is significant and positive though the marginal impacts differ somewhat. The findings suggest that rulers are more likely to exploit identity issues and engage in mass killings in ethnically mobilized societies. This finding is robust across two indices of ethnic polarization and fits previous research and our expectations. We believe that in all polity types, rulers have to mobilize support among masses for the solidification of their rule. Even in personalist regimes, leaders are better off if they can back their rule with mass support. Therefore, mass mobilization is critical to the survival of a regime regardless of the characteristics of political institutions. In this respect, identity issues may provide clear-cut and easy-toexploit societal sentiments for greedy politicians. The exploitation of latent communal cleavages may sometimes escalate communal unrest to the level of human insecurity.19 The interaction variable, however, is not associated with mass killings.20 Cultural pluralism does not, by itself, indicate the breakdown of trust and cooperation, communal conflict, and ethnic exclusion, as long as political institutions are adequately designed to prevent their politicization. In this respect, political institutions determine the opportunity structure for mass killings and the tools that leaders would use once their power is challenged, whereas, societal characteristics, such as ethnic fabric of a society, shape the nature of challenges that leaders might actually perceive.

Discussion: rules, rulers, and the ruled Our research examined the relationship between political institutions, ELF, and the incidence of mass killings in all countries from 1955 to 2001 using Harff’s (2003) data. We found strong evidence suggesting that death by

Rulers as mass murderers 93 government is directly associated with the lack of institutional limits on executive power. Regimes with substantial constraints on executive decision-making were less likely to engage in mass atrocities. The nonsignificant effect of the openness of competition for office, and the significant positive association between political participation and mass killings, further suggests that institutional limitations on executive power are central to understanding why leaders may or may not provide human security. This was confirmed when autocratic regimes were disaggregated into single-party, personalist and military regimes. All autocratic polity types were more likely than inconsistent regimes and democracies to commit mass atrocities, despite possibilities for inter-elite bargaining within autocracies and the restraints on policymaking that certain autocratic designs provide. Finally, we observed that ethnic-linguistic diversity is not associated with mass killings in regimes which impose group decision-making. Overall, the strong negative impact of executive constraints on mass killings suggested that rulers have an incentive to pursue good governance as a political survival strategy if the polity type effectively limits rulers’ access to power through constitutionally designed accountability groups. Political elites are much more likely to resort to coercion as a tool to forestall threats to the regime or to solidify their domestic standing in systems where the ruler cannot credibly commit herself to protect a variety of citizens rights, including the right to live and prosper. The same conclusion also explains the origins of human tragedies in ethnically divided societies. Killings are primarily observed in countries lacking constitutional checks on the political power of the sovereign. In these systems, governance rests in the hands of small groups that avoid internal dissent, or make concessions to a dominant actor in order to retain power. Such systems suffer from the absence of autonomous political actors with authority to monitor the quality of governance. Inadequate monitoring affects not only the decision-making but also the policy implementation process. Hence, there are stronger incentives to escalate killings where the absence of opposition reduces the political costs of formulating and implementing coercive policies. How to create incentives for rulers to govern for human security is an important policy question. Our analysis has aimed to suggest some answers. The theoretical and empirical analysis in the present study has indicated a close link between rulers’ incentives to target civilians and the institutional environment in which their actions are embedded. One lesson that may be drawn from this exercise is that constitutional practices that limit leaders’ access to political power tend to ensure that they govern for the security of their citizens.

94 A. Aydin and S. Gates

Notes 1 Source: R.J. Rummel’s personal website accessed at www.hawaii.edu/ powerkills/20TH.HTM, on March 26, 2005. 2 Following Gates et al. (2006), we refer to such “mixed” regimes which contain both democratic and autocratic elements, as institutionally inconsistent regimes. 3 For example, rulers in economically integrated societies may be less willing to cause domestic violence and increase political risks to traders and investors. Also, membership in international institutions may create outside pressure on rulers for good governance. 4 Harff uses the Polity 0–10 point democracy – autocracy scale. Full democracies are polities that have a score of 7 to 10; partial democracies, 1 to 7; and the rest is coded as autocracies. 5 Valentino et al. (2004) uses the +10/–10 Polity democracy – autocracy scale. The three categories in this study are; full democracies (+7 or higher); midrange polity (between +6 and –6); and, full autocracies (–7 or lower). 6 Similar to Valentino et al. (2004), Davenport also adopts the Polity democracy – autocracy scale, but uses this index as a continuous variable rather than as discrete categories. This approach still does not differentiate between the relative influences of different dimensions of regime type on human insecurity. 7 Also see Eck and Hultman (2007) for a similar structural approach to the relationship between regime type and death by the ruler. 8 For instance, rulers do not wait for fully autocratic environments to engage in poor governance. In this respect, 6 on a 10-point scale which scholars identify as the cut point for democracy/autocracy presents an arbitrary choice and does not seem to be theoretically plausible to understand the incremental shifts in leaders’ incentives to govern for in/security. 9 Rummel’s (1995) factor analysis in many ways allows for a similar continuous multidimensional analysis. 10 Following much of the literature, we make a second assumption. We assume that the goal of the leader is political survival (e.g. Bueno de Mesquita et al., 2003). 11 As one of the important international consequences of state-sponsored murders, political actors would try to avoid military intervention that thirdparty states would undertake to stop the killing, which would substantially increase the costs of targeting civilians. Hence, empirical evidence shows that decisive international action can manage mass killings. For instance, Krain (2005) shows that perpetrators of genocides/politicides are more likely to yield to international pressure in the face of outside interventions that either assist the targeted group or challenge the perpetrators. 12 See Valentino et al. (2004) for a thorough examination of the strategic logic behind mass killings that aims at eliminating opposition groups and the techniques with which these deadly policies are implemented. 13 These data can be accessed at the State Failure Task Project website www.cidcm.umd.edu/inscr/genocide/. 14 This type of logit analysis is especially suitable for cross-sectional time-series datasets. The random effects approach allows us to account for the fact that we have multiple observations for each country; this is important, given that the underlying probability of observing a genocide/politicide varies by countries so that for the same values of the institutional variables, the unobserved heterogeneity problem leads to different probabilities of mass killings across units. In using a random effects model, we avoid imposing a correlation structure on the serial observations for each country: we assume that unobserved heterogeneity

Rulers as mass murderers 95 is a random variable and there is no relationship between the unit effects and the independent variables in the analysis. 15 Version 1.0 of the Vanhanen data accompanied an article in Journal of Peace Research (37–2, March 2000). The current version of the dataset is Version 2.0 (www.prio.no/page/Project_detail/Replication_Datasets/9649/42472.html). 16 For a detailed description of these categories, see Geddes (1999) and Peceny et al. (2002). In a nutshell, In military regimes, a group of officers decides who will rule and exercises some influence on policy. In single-party regimes, access to political office and control over policy are dominated by one party, though other parties may legally exist and compete in elections. Personalist regimes differ from both military and single-party in that access to office and the fruits of office depend much more on the discretion of an individual leader. (Geddes, 1999: 7) 17 Trade openness is measured as the ratio of exports plus imports to the GDP. Bueno de Mesquita et al. (2003) use the Penn World Tables to create the openness index. 18 Of 570 observations, 91 cases represent transitional periods according to Polity IV and therefore, are not included in this classification. If we take the midpoint of the executive constraints (4) as an arbitrary cut point to classify regimes into high–low categories with respect to limits on political power, there are 410 cases that fall within the low category (including those regimes that score 4 on the scale), and 69 regimes that are in the high category. The countries and time periods in the high category are: Regimes with a “7” executive constraints score; El Salvador 1984–92, Uganda 1966, Sudan 1956–89, Pakistan 1973–76, Burma 1961–62, and the Philippines 1987–98. Regimes scoring 6; Guatemala 1996. Regimes scoring 5; Chile 1973, Argentina 1973–76, Burundi 1963, Sri Lanka 1983–98, the Philippines 1969, and Indonesia 1956–59. 19 See Rabushka and Shepsle (1972) for their influential argument that in ethnically divided societies, the formation of one ethnic party triggers the formation of others and mobilization of masses along ethnic lines. However, the “intense and non-negotiable” character of ethnicity reduces the value of bargaining with other parties and leads to democratic breakdowns by eliminating competitive politics (p. 72). 20 The Cederman–Girardin N* index appears to be insignificant when we introduce the interaction variable into Model VII. This, we think, is a matter of the small sample size of the dataset. N* data are generated for a country sample and therefore, are not able to handle demanding models that require more information.

6

Resentment, fear, and the structure of the military in multiethnic states Roger Petersen and Paul Staniland

Introduction A fundamental purpose of this volume is to apply deterrence theory to the domestic context. In pursuing this issue, Saideman and Zahar highlight the central role of the state in producing security. Correspondingly, the editors’ operating questions for this volume ask scholars to identify the conditions when the state can deter violence and when the state can assure its citizens that the state itself is not a threat. But it is not just the state at issue here. Rather, it is the state and its ability to project force, an ability usually tied to the military. In multiethnic states, domestic deterrence issues often involve three types of actors: the state (the government and bureaucracy), the military, and the politically salient ethnic groups. The way these actors align critically affects deterrence and assurance. Assuming that group-based conflicts arise in multiethnic countries, state-military relationships can determine the evolution and outcome of these conflicts. Consider five basic relationships: 1

2

3

One ethnic group dominates both the state and the military. In this case, the state, in league with an ethnically similar military, can likely deter any predatory or problematic challengers to state power. However, it will be difficult to assure ethnic groups in the state that the state itself will not become a predator. The state and the military are both multiethnic. Here, the state is less likely to be driven by specific group grievances and also less able to coordinate violent action against ethnic groups. When the military is composed of many ethnic groups, it is more likely to resemble a unitary actor maximizing the organization’s, rather than one ethnic group’s, interests during times of political change. Problems of both assurance and deterrence are decreased. The state is dominated by one ethnic group while the military is multiethnic. In this case, the military may act as a brake on the state’s predatory tendencies or aims. “Evil leaders,” raving nationalists, inflammatory rabble-rousers may wish to initiate violence against an

Resentment, fear, structure of the military 97 ethnic target, but they may not put this program into place if a multiethnic and professional military stands as an obstacle. The state will likely find it difficult to involve a multiethnic military, especially a professional military, in violence against a specific group of citizens. Problems for both deterrence and assurance are most likely to arise in the following situations: 4 5

The state is dominated by one ethnic group while the military is dominated by another ethnic group. The state is multiethnic while the military is dominated by one ethnic group.

First, a state with an ethnically skewed military may heighten consciousness of ethnicity in the first place. When the guns are mostly in the hands of members of one ethnic category, it is difficult not to think in group terms. In effect, “they” have the guns and “we” do not. There will be a tendency to conceive of ethnicity in security terms. Second, the existence of an ethnically skewed military reduces the credibility of a status quo state’s threats and promises. Because an ethnically skewed military may capture the state on behalf of a group and then advance that group’s agenda in the future, groups cannot fully count on the security guarantees of the state in the present. Without confidence that the state can provide protection, groups may begin to arm themselves, creating or enhancing a spiral toward internal war. Territorially concentrated ethnic groups may seek to separate in order to guarantee their own protection thus setting off secession-based conflicts. With an ethnically diverse military, the chances of state capture are greatly reduced and the necessity for groups to provide their own security is reduced. In effect, when the military is ethnically diverse and less capable of acting on behalf of one group, the chances of security dilemma dynamics arising are blunted because windows of opportunity and vulnerability are at least partially closed. The most dangerous circumstance, however, is the transformation of a domestic system from situation #1 to situation #4. In these cases, the visible positions of power in the state are being transferred from one ethnic group to another. There is a reversal of political status while the primary means of violence remain in the hands of members of the newly subordinated ethnic group. It is no exaggeration to state that one of the major security issues of our times is how to deal with reversals of ethnic status hierarchies. Some ethnic hierarchies are bound to experience change and status reversal. South Africa could not go on forever in apartheid form. With the collapse of Communism in Yugoslavia, the relative positions of Serbs, Croats, Slovenians, Albanians, Muslims, and others were bound to undergo change. The Arusha Accords confronted the social system in Rwanda with impending status change. These three cases illustrate the

98 R. Petersen and P. Staniland range of outcomes that might occur with changes and reversals in ethnic hierarchy. Peaceful change, ethnic war, and genocide are all possible. Which outcome obtains can depend upon the actions of the military. If there is one institution capable of preventing the slide toward violence, it is the military. If the military acts on behalf of a neutral or autonomous state rather than splintering or siding with one particular group, the chances for ethnic killing might be greatly diminished. If the military itself is swayed by ethnic resentments, it can side with one group to attack or punish another. Larger societal and demographic changes may produce processes leading to status reversals and strong ethnic resentments, but the military can act as either a brake or a catalyst on these processes. This chapter asks two questions: (1) What conditions are most likely to see the military act as a brake on ethnic violence and which conditions are most likely to make the military a catalyst? (2) Are there ways to institutionally structure the military in multiethnic states so it is more likely to act as a brake on ethnic violence rather than a catalyst? The chapter proceeds in three sections. First, the problem of ethnic status reversal is specified through a discussion of its component processes of resentment and fear formation. Second, the broader context of military structure is examined. What variables help create multiethnic or dominated militaries? How do these variables also help create the resentments and fears that in combination with a skewed military lead to potentially violent outcomes? A third and final section explores possible institutional solutions through a comparison of two cases.

Resentment, fear, and the military: paths to violence or peace during ethnic status reversals In most multiethnic societies, social relations are often tinged with overtones of domination and subordination.1 While status can be complex, status relations among ethnic groups can generally be tied to the following indicators all connected to state policy or ethnic composition of the state’s offices: the composition of elected officials, the composition of the state’s bureaucracy, the composition of the police, symbols such as street names and flags, the language of day to day government. In this definition, resentment is a political, not an economic, phenomenon. Some ethnic groups may be wealthier than others, but when individuals are forced to speak the language of others in everyday business, when they are under the eye of ethnically different police, when they cannot advance in the ranks of the state bureaucracy or the military, then they occupy a lower level on the status hierarchy. Resentment is the feeling of being politically dominated by a group that has no right to be in a superior position. It is the everyday experience of these perceived unjust status relations that breeds the emotion. At any given point in time, an ethnic hierarchy based on these observ-

Resentment, fear, structure of the military 99 able political factors is usually in something of an equilibrium. This equilibrium, however, can be upset through a variety of structural changes. In Eastern Europe in the past, and in Iraq today, foreign occupation has worked to reorder status relations. Reform or negotiated change, such as seen in the last decade in South Africa, will change ethnic hierarchy. The decline or collapse of the state will inevitably change both power and status relations. Decolonization has been a common process producing ethnic status change. These changes can bring equality to previously hierarchical systems, introduce hierarchy to previously equal systems, or reverse the status orderings of a hierarchical system. Resentment motivates violence. There is no better way of establishing and publicizing the subordinate status of an opponent than through violence. Violence and visible force can quickly re-establish the “just” social order. Resentment can be compared to the narrative underlying fear-based conflicts. In general form, fear narratives hold that structural changes such as the collapse or weakening of the political center eliminate institutional constraints and guarantees to produce a situation characterized as anarchy or emerging anarchy. Under these conditions, fear heightens the desire for security. Under fear, the target of ethnic violence will be the group that is the biggest threat; under the sway of resentment, the target will be the group in the new and “unjust” status position. Ethnic resentments and ethnic fears can independently lead to violence. One group can attack another group simply because it resents the effects of political status reversal. They can attack the higher status group even in the absence of threat. Alternatively, one group can attack a lower status group that presents a threat. There are other possibilities, though. One particular group can both present a threat and occupy an unjust status position. Resentment and fear can operate simultaneously. As illustrated by the arrows in Figure 6.1, resentments can lead to fear. These emotions may operate in sequence. In moving from situation #1 to #4, both fears and resentments may come into play and influence the military’s action. For the sake of illustration, consider an ethnically bipolar country – two groups (A and B), a high general sense of ethnic hierarchy, an ethnically disproportionate military in which group A is overrepresented. Now imagine that the country is facing a situation in which group B is becoming dominant in the bureaucracy, B’s symbols will be seen throughout the state, B’s language will become the official language in most of the state, and so on. Resentment, fear, and violence can line up in various ways: 1

Resentment develops, does not lead to fear, but violence still results. If B has no force potential, and little hope of gaining force potential, the military and group A face no threat. If violence occurs, it occurs in response to resentments only, not fear. In this case, the military may commit violence against the status-based target, or tacitly allow or even encourage pogrom-like violence.

100 R. Petersen and P. Staniland Gener al v ariab les 1 Le v el of dev elopment 2 Type of go v ernment (militar y v ersus civilian) 3 Colonial histor y 4 Geog raphical region 5 Siz e of state 6 Number of ethnic g roups 7 Existence of dominant g roup

Resentment (Changes in status hier arch y, rev ersals)

Militar y structure (composition, recruitment)

Fear (Secur ity dilemma conditions)

Ethnic violence

Figure 6.1 Fear, resentment and military structure.

2

3

4

Resentment develops, does not lead to fear, and no violence results. Although dominated by one group, the military may remain committed to the integrity and laws of the state. A reversal occurs, but as the military retains a monopoly of force, group A’s resentment cannot easily translate into violence. No group need fear another group, at least in the short run. Resentment leads to fear and violence. As in the first scenario, in the pre-conflict equilibrium, ethnic group A is dominant over B. A status reversal takes place in which B gains rights in language, state jobs, official symbols, but the military remains identified with group A. However, group B shows the clear ability to raise military force and also the likelihood of increasing force potential over time. In this situation, fear is joined to resentment and the military/group A has every incentive to strike first. Ethnic war is possible. Resentment leads to change in military structure which in turn generates fear and leads to violence. The military may be an independent actor committed to the integrity of the state in the period previous to structural changes. However, in this case, the military may come to

Resentment, fear, structure of the military 101 identify with group A’s resentments during the process of status change. The military then may purge members of B from its ranks creating a more homogenous military even more dominated by members of A. Purged military from group B then forms a counterforce capable of setting security dilemma dynamics into motion. Ethnic war is possible. Moving from situation #1 to #4 can create a variety of possible paths and outcomes, but which is likely to obtain? The answer depends upon a number of general variables, the structure of the military, and the interaction between these general variables and military structure.

The broader context of military structure Why are some military structures dominated by one ethnic group in the first place? Are the variables that help create ethnically skewed militaries also likely to drive these militaries to react to status reversal with violence? This section examines these questions in light of the set of general variables listed in Figure 6.1. A first issue regards the conception of military structure. For the present purposes, the most relevant questions concern who is in the military and how did they get there. Correspondingly, military structure can be broken down into two components: composition and recruitment policy. The authors have compiled data on about 30 multiethnic states and their militaries, as they stood, around 1980.2 These data create a helpful snapshot of the variation found in world militaries. In dealing with composition, proportionality seems to be the most relevant issue. Are certain ethnic groups underrepresented or excluded? The following scale was used to assess proportionality in a state’s military: 0  Total exclusion or near total exclusion of certain ethnic groups. The underrepresented groups are 0–10 percent of expected participation rates based on their percentages in the population as a whole. 0.5  Token representation of some ethnic groups; 10–20 percent of expected rates. 1.0  Higher than token representation, but still a gross imbalance; 20–40 percent of expected rate. 1.5  Some ethnic groups at half of expected enlistment rates; 40–60 percent of expected numbers. 2.0Easily recognized imbalance; 60–80 percent of expected participation. 2.5  Some skewing; 80–90 percent of expected figures. 3.0  Total proportionality or near total proportionality; there is less than a 10 percent deviance from expected figures. The military is not a monolithic organization. There are divisions between the ranks and the officer corps and between the services (army, air

102 R. Petersen and P. Staniland force, and navy). Even the most cursory glance at the militaries of multiethnic states reveals that these divisions operate in numerous ways to both exclude and balance native ethnic groups. In some states, one ethnic group dominates the officer corps while another fills the ranks; in other states, one ethnic group may dominate the army while another will be found mainly in the air force; in other states, there will be equality across the board. In order to measure the proportionality of composition of the recruitment policies involved with the ethnic boundary, these divisions must be taken into account. Accordingly, separate scores were assigned for the army ranks, army officer corps, and the ancillary services (air force and navy). These scores are then added to derive a total measure of the proportionality of composition ranging from 0–9.3 For the scale on the whole, a zero score would indicate that the composition of the military is totally unrepresentative of the state’s population in general. A score of 9 would indicate a military with a composition that is totally proportional to the state’s population. The scores in between, of course, indicate varying shades of proportionality. The second component used to measure military structure addresses recruitment policy. This measure tries to capture the type and amount of organizational effort involved in recruitment. Again, a 0–3 scale was developed.4 0  Market forces play the biggest role in recruitment. No organized effort is evident in patterns of recruitment. 1  Informal networks. Recruitment originates from ethnic circles, family groups, or localities. Patterns found at sub-state levels. The state tacitly accepts or approves of the relationship between these informal networks and the military. 2  State-centered strategies. Evidence is available that the state is imposing patterns of recruitment from above, although the mechanisms will be non-legal. 3  Legal quotas or other legal devices have been implemented. As with composition, differences between the ranks, the officers’ corps, and the services exist in recruiting policy. Commonly, the nature of the ranks in a certain state is determined by market forces while the officer corps of that same state is determined by informal networks or state-centered strategies. In addition, the recruitment policy varies across services. As with composition, scores for three components are added to create a 0–9 scale. A 0 score will indicate an overall recruitment policy with market forces predominating over conscious state policy. A score of 9 will represent a recruitment policy where the state employs formal, legal means to decide how individuals are placed into or excluded from the armed forces.

Resentment, fear, structure of the military 103 General connections Figure 6.2 plots the scores of composition and recruitment. The chart should be read as follows: the upper left corner represents cases which exhibit state-intrusive policies with disproportionate compositions; the upper right corner contains states with state-intrusive policies but proportionate compositions; the lower left corner contains states with less intrusive policies and disproportionate compositions; the lower right represents less intrusive policies and proportionate compositions. In an earlier study, the senior author ran a regression analysis that suggested the power of three of the general variables in producing this observed configuration: number of groups, colonial background, and level of development.5 The findings on number of groups is by far the strongest. With a smaller number of salient ethnic groups, the state seems compelled to play an active role in shaping military structure. Stunningly, all ten states in the sample with only two ethnic groups have recruiting policies in the range 6.25–9.0. None of the eleven states with five or more groups have scores in this range. The mean score for states with just two groups is 8.05 while states with three to four groups and states with five or more groups have means of 5.19 and 4.39, respectively. While the relationship between “number” and recruiting policy was the only one to meet the standard test of statistical significance, several other results were suggestive. The number of groups seems related to the composition variable as well as the recruiting variable. Seven of the ten states with only two groups had composition scores of 0–3 while the remaining Belgium Switzerland

Guyana

Sri

8

d ka un Lan ur B

7.5

u S rinam

6.5 6 5.

Iraq

4.5

Singapore

Kenya Syria

ria

5

Ghana Uganda Burma Thailand

Nige

an ca da uS d th Afri an Rw Sou iji, F i,

7

Recruiting policy

Canada

Israel

ia av k lo oh s ez c C

9 8.5

Yugoslavia

4 M a laysia o J rdan

3.5 3

India

2.5 1.5

Ethiopia

Indonesia

Lebanon Pakistan

2 1 0.5 0

0

1

2

3

4

5

6

7

Composition

Figure 6.2 Ethnicity and militaries: composition and recruitment.

8

9

104 R. Petersen and P. Staniland (a score of 9 represents a completely proportionate military while a score of 0 represents a highly disproportionate military). The data again show a sharp break between states with two groups and those with a higher number of groups. The only other variable with possible substantive meaning for both recruiting policy and composition was colonial background. Like “number,” this variable was coded into three categories with a projected linear relationship: non-British former colonies would possess the most state-intrusive and unequal policies, while former British colonies would have intermediary scores on both measures, and states with no colonial history would have the highest scores on each measure of military policy. Thus, the values suggest that for each move in category, from non-British to British or from British to non-colonial, there is about a one point increase on both scales that can be accounted for by “colonial.” As revealed by the regression analysis, only one other relationship appeared to be significant enough to be addressed at this point: the effects on per capita GNP (level of development) on composition. The logic behind this variable was that higher stages of development should produce more proportional militaries, and indeed the data suggests that this is the case. Four of the five countries exhibiting high proportionality are found in the class of states with per capita GNPs of more than $2,000; this is clearly a distinctive characteristic that suggests that levels of development may be affecting military composition.6 States with per capita GNPs of $750–$2,000 had nearly an identical distribution as states with a per capita GNP of below $250, suggesting that the effect of wealth is mainly operable at the highest level of GNP. The other independent variables – population, type of government, dominant group, level of development on recruiting policy, and region – did not appear to have consistent and convincing effects. In summary, the statistical analysis suggested the power of the number of groups variable on both recruiting policy and composition. The most commonly cited parameter, level of development, seems to be influential only on composition and even then it seems important only in terms of the very richest of countries versus the mass of poorer states. Examining clusters of cases With so few cases, statistical analysis may not be as illuminating as a simple examination of the clustering of cases seen in Figure 6.2. As is shown on the chart, five distinct clusters of states can be identified. Cluster #1: Israel, Guyana, Burundi, Surinam, Sri Lanka, Fiji, Rwanda, South Africa, and Sudan. Cluster #2: Canada, Belgium, Switzerland, and Czechoslovakia. Cluster #3: Burma, Ghana, Iraq, Thailand, Uganda.

Resentment, fear, structure of the military 105 Cluster #4: Yugoslavia, Singapore, Kenya, Nigeria, and Syria. Cluster #5: India, Malaysia, Jordan, Indonesia, Pakistan, Lebanon, and Ethiopia. The nature of these clusters suggests how some key variables affect domestic processes of deterrence and assurance. Cluster #1 is composed of nine states possessing state-intrusive recruiting policies and highly disproportionate military compositions. In all of them, situation #1 (one ethnic group dominates both the state and the military) prevailed in 1980. Obviously, if the most dangerous situation is movement from situation #1 to #4, the cases must be in situation #1 to begin with. These are the potentially dangerous cases. Several of the variables emphasized in the regression analysis, specifically, number of groups, colonial policy, and GDP, would seem to explain this particular clustering of cases. First, seven of these nine states have two major ethnic groups. In the remaining two, Surinam and South Africa, competition between ethnic groups often collapses into a dichotomous game as well (recall that the data are circa 1980). Second, almost every state in this cluster underwent a colonial “divide-and-rule” policy involving ethnic skewing of the military. In Burundi, the Belgians promoted the minority Tutsi tribe over the majority Hutus. This imbalance has survived until the present as Tutsis have employed very rigid state policies to maintain an all-Tutsi force (here is a case of the military preventing status change). In Rwanda, Belgian policies again favored minority Tutsi over Hutu only this time the Hutus managed to overthrow the formerly favored Tutsi and exclude them from the military. In the Fiji Islands, British colonial policy brought in large numbers of Indian laborers and then employed native Fijians in the armed forces to control them. Native Fijians worked to retain this former colonial imbalance in order to maintain power over the Indians who actually form a majority of the population (Fiji also provides a clear example of the military preventing status changes in the state through military coup). In Sudan, a similar story can be told as the Arabs have acted to maintain inequalities in the armed forces that were created during the British colonial days. Sri Lanka also provides an example of how the mixture of low number and colonial policy act to produce the general qualities found in cluster #1. The British recruited almost exclusively from the Tamil population for the military and much of the civil service. After independence, the majority Sinhalese began to assert themselves. The 1956 election split entirely upon ethnic party lines. Sinhalese bitterness over the privileged treatment previously received by the Tamils led to a series of laws regarding language and religion that threatened the very existence of the Tamil culture. In 1962, a group of military officers, partially motivated by ethnic factors, attempted a coup. The coup failed and quite predictably the Sinhalese-dominated government worked to change the ethnic composition of the military as well. A large number of

106 R. Petersen and P. Staniland Tamil officers were immediately retired and a new set of stringent state policies were instituted to place Sinhalese in dominant positions throughout the armed forces. The remaining four states in this cluster might be explained more by the effect of number rather than that of colonial policy. Israel, South Africa (discussed at length below), Surinam and Guyana are all states that witnessed a severe polarization of ethnic groups. In many of these cases, the nature of two-group competition sharpened ethnic consciousness and heightened the sense of injustice over the legacy of colonial policy. Following Figure 6.1, a set of general variables (low number of ethnic groups, colonial history, and low GDP) produced interactions among resentment, military structure, and fear that have led to either an unstable situation #1 or to violence and coups. Cluster #2 (Canada, Belgium, Switzerland, and Czechoslovakia) is composed of four highly developed states found in the right hand corner of Figure 6.2. These states therefore represent intrusive policies and equal composition. The cluster relates to the findings of the regressions in two clear ways. First, the power of the number variable on recruiting policy is again very evident. Three of the four states have only two groups and the fourth, Switzerland, also often exhibits a dichotomous political situation in which German-speakers often face off against French- and Italian-speaking cantons. The result has been “affirmative action” programs aimed at maintaining proportionality. Second, this cluster helps clarify the relationship among number of groups, level of development, and military structure. Both cluster #1 and cluster #2 are almost entirely composed of states with only two major ethnic groups; the factor which almost certainly separates the two clusters on the proportionality score is development. All four of the states in cluster #2 rank in the highest levels of per capita GNP in the sample, while only one case in cluster #1 can be found at this level (Israel). The effects of development on composition can be clearly seen by comparing the events in Sri Lanka to those that have occurred in Belgium.7 In both states, the fact that there are only two significant groups has helped to produce a zero-sum mentality in which every institution is seen as an ethnic battleground. Despite this similarity, Sri Lanka produced a highly disproportionate military while Belgium has instituted policies that have led to a high degree of proportionality. For example, Belgium required all officers to be bilingual and set up programs to help accomplish this goal. This action helped prevent the military from being seen as a Belgian institution rather than Walloon or Flemish institution. This policy was possible mainly due to the fact that Belgium’s highly developed educational system had already spawned a relatively high percentage of bilingual speakers and was wellequipped to meet higher demands for bilingual education. The Sri Lankan educational system was simply not able to perform such a role. In addition, in Belgium, there has been a broad pool of possible middle-class recruits existing across ethnic categories that allows the state more options

Resentment, fear, structure of the military 107 in developing a coherent and acceptable balancing schema. In Sri Lanka, the lack of a sizable pool of middle-class Sinhalese caused problems in balancing the officer corps. Furthermore, the more evenly developed Belgian society has a number of cross-cutting cleavages running through it that have not been present in the severely imbalanced Sri Lankan state. Flemish and Walloons in Belgium belong to the same unions and occupational organizations for instance; a large middle-class with similar interests is found in both Flemish and Walloon communities. The high level of development in Belgium has created a situation and atmosphere in which the composition of the military could be fairly easily balanced. For reasons addressed throughout most of the literature on civil war, higher GDP also makes war a less attractive option for political elites. War creates unacceptable opportunity costs. With a higher level of GNP, there is a bigger “pie” to be shared and the need or desire to use violence to obtain a significant share of the state’s or society’s resources is much less. In higher income states, the status of the military profession may also be lower and less a matter of contention than in lower income states. All these factors may make it easier to create affirmative action programs in the military. Clearly, even in these highly developed states, ethnic consciousness is high and political resentments develop. Obviously, since the time of this snapshot Czechoslovakia has disintegrated, Canada has witnessed a near secession and the Belgians live in a condition of semi-partition. However, the security problem was solved through negotiation. In effect, these countries moved to situation #2 – both the state and the military became multiethnic. Cluster #3 (Burma, Ghana, Iraq, Thailand, and Uganda) is both predictable and surprising. It sits between cluster #1 and cluster #5. All three clusters contain poor states. Unlike cluster #1, the countries in cluster #3 possess multiple salient ethnic groups. The higher number of groups might produce less clear competition and more possibilities for alliances among groups. Correspondingly, the state can be somewhat less intrusive in its recruiting policy and also somewhat less exclusionary. Unlike cluster #5, cluster #3 possesses countries with dominant groups and military governments. There are seven states with dominant groups, military governments, and more than three groups in the sample as a whole. Four of these states are found in this five-state cluster. These are cases where the military took over the state, but due to more groups and more ambiguous politics, the solutions for security problems could be less stark than for two group cases. Still, these cases are found close to situation #1: the military and state coordinate to maintain mutual dominance. These are states which can deter but not assure. A comparison with cluster #5 proves useful. This cluster is composed of states with non-intrusive recruitment policies and multiethnic militaries. Of six states within the sample as a whole that have civilian governments and no dominant groups, four such states are found in cluster #5. These

108 R. Petersen and P. Staniland tendencies can be illustrated by comparing Burma (cluster #3) to India (cluster #5). Both of these states have a multitude of groups (many of which are similarly caste-based), populations over 25 million, very low per capita GNPs, British colonial histories, and are located in the same geographical region. The differences are that Burma has a dominant group (Burmans) and a military government, while India has neither.8 In Burma, the military battled ethnic insurgents for years, when the former civilian government began making concessions to ethnic minorities, the Burmese army performed a successful coup in the spring of 1962. The new military government, in accordance with the assertion laid out previously, had no patience in dealing with ethnic groups and proceeded to “Burmanize” the army to an even greater degree. As one Burmese general described the ethnic minorities: “In spite of the obvious need for united action, certain dissident elements are holding fast to narrow racial outlooks with misguided, extremist policies, carrying on their disruptive work, hindering progress of development projects and imperiling national unity.”9 The solution was that the “Burmanized” military should clear away ethnic dissent and rapidly lead the country to progress. One reason the Burmesedominated military could implement their policies is that Burmese make up over 70 percent of the overall population while no other groups make up over 7 percent. As a dominant group, the Burmese-oriented military could institute its unequal policies without fear of an overwhelming backlash from the populace at large. The situation in Thailand, another state in cluster #3 had been similar while the military governments of Iraq, Ghana, and Uganda have exhibited much of the same type of impatience in dealing with the ethnic compositions of their militaries. India, on the other hand, has had a much different history in such affairs.10 This civilian-ruled state has taken a much more evolutionary approach in regard to the ethnic dimensions of its military. After independence, the Indian government inherited a highly skewed force with groups from the Punjab dominating all parts of the services. The Indian state opted to change military composition through adoption of less intrusive policies that allowed market forces and newly active informal castebased networks to create more balance throughout the military. In India, no dominant ethnic group controlled the state so the possibilities for rapid and threatening actions, such as those seen in Burma, were low. Another state in cluster #5 that has both a civilian government and no dominant group, Malaysia, exhibits a similar pattern. Malays continue to dominate the armed forces but they have gradually made concessions to other ethnic groups in Malaysian society. The state created new non-Malay regiments and recruited Chinese and Indians into the ancillary services in significant numbers. Following Figure 6.1, the cases in cluster #5 possess a more diverse demography (more groups, none dominant) that helps to decrease ethnic resentment and prevent the cycling of interactions among resentment, military structure, and fear. As opposed to the countries in clusters

Resentment, fear, structure of the military 109 #1 and #3, the cluster #5 cases were more able to move toward multiethnic states and militaries and some of these poor and diverse states were able to gain measures of both deterrence and assurance. It is difficult to draw a clear story from the cases in cluster #4 (Yugoslavia, Singapore, Kenya, Nigeria, and Syria). Here we find a mixture of properties. One guess is that level of development might be acting to distinguish this cluster. Two of its members, Yugoslavia and Singapore, had per capita GNPs of over $2,000 in 1980. Two other members, Kenya and Nigeria, possessed perhaps the most developed economies in Africa despite their relatively low per capita GNP figures. In all of these states, except Syria, the government had been active in trying to create balance in the armed forces. These processes bore resemblance to the programs in the advanced states of cluster #2 and may partially account for the upward right position of this cluster among the mass of states with more than two groups. Again, more developed states have more resources available to them in working to create a more proportionate composition. Following Figure 6.1, a higher GDP appears to have helped diminish ethnic resentments and diversify the military for cases in this cluster. Of course, the Yugoslav case is evidence that the effects of GDP have their limits. Summary of findings from the data analysis States with two groups will generally be more conscious of ethnic hierarchy. There is less information in the system and only one comparison to be made. There is less uncertainty, there cannot be multiple and shifting coalitions of groups. This argument suggests that status changes are muted as the number of ethnic groups increases – relative status differences will be smaller and less noticeable, even in ranked societies, and thus concerns about dominance and distribution will be reduced. Combined with low GNP, bipolar ethnic structures are likely to induce states to actively skew the ethnic composition of the military. These ethnically skewed militaries in lower income states are perhaps most likely to be affected by growing societal resentments. They may be most likely to feel entitled to use violence to maintain “just” or “natural” ethnic hierarchies. Especially in states with lower GNP and insecurity, the mixture of status loss and security vulnerability is deeply worrisome to these militaries. Because of the military’s, simultaneously, violence-producing and hierarchical nature, in military status reversals both fear and resentment come into play. With only two ethnic groups, elements in the military are more likely to consider coup d’etats or “spoiling” in the time period before a likely status reversal. Because status changes are much starker in two-group ranked systems, the prospect of status reversal will be more worrisome. There are no other significant groups with which to ally, and thus an imbalance in either demographics or in military strength (not necessarily the same thing)

110 R. Petersen and P. Staniland will create incentives to de-rail the future status reversal. A group’s current dominance creates a “window” through which it may be tempted to jump. The effects of the number variable may interact with other variables such as power balances and demographics. Imbalances of power in favor of rising groups, combined with an inability to commit to future peace will lead currently dominant groups to use force to halt the anticipated status reversal. Although not entirely supported by the statistical analysis, violent intervention may be more likely when the militarily dominant group is the same as the demographic majority. This is true for two reasons – first, the majority feels it should rightly hold status dominance, and second, the majority group can more reasonably believe that as a majority it can have long-run success in retaining dominance. On the other hand, the more groups in the system, the more likely the military will act as a brake on processes leading to ethnic violence. When the military is composed of many ethnic groups, it is more likely to resemble a unitary actor maximizing the organization’s interests during times of political change. Also, when the more dominant group in the military is a minority in society, the military may be more likely to accept status reversal due to both long-term potential costs and less resentment. Also, with more groups in the system, the chances are higher for a multiethnic state and military (situation #2) to exist in the first place. Although the statistical analysis only looked at militaries at one point in time, the relevance of the GNP variable suggests that development process may shape both resentment and military structure. Development can certainly generate resentments. Increasing literacy, as in Benedict Anderson’s work, provides the ability for the imagination of common identity. Urbanization brings different ethnic groups together and forces them to live with each other, and compare relative position on a daily basis. As Ernest Gellner has argued, the needs and growth of the state in education and conscription produces the need for a common language that would differ from that of many of the state’s ethnic groups. In short, development produces increased chances for various resentments to arise. Rapid development may be likely to lead to cognition of unjust status position as well as higher chances of status reversal. Colonial history, as seen in the examples in the previous section, often created the conditions for resentments. Colonial policies instrumentally created or reinforced ethnic categories to help with the day to day information tasks necessary for colonial rule. Divide and rule policies, as well as the designation of “martial races” to help with recruitment of local forces for local police and military positions, often formed a blueprint for creating hierarchical systems that would cry out for reversals in the period following decolonization. A low level of GDP, with lower opportunity costs for anti-status quo violence, also contributed to an enhanced value on controlling the state.

Resentment, fear, structure of the military 111

Seeking institutional solutions The violent cases are likely to involve a poor, ethnically bipolar state with a colonial legacy of inequality. These general variables are likely to lead to both ethnic resentments and ethnically skewed militaries. Moreover, this set of factors is most likely to create the process of situation #1 moving toward situation #4. These are the cases that must eventually change their state structures while the military remains in the hands of one ethnic group. In turn, resentments and skewed militaries are likely to produce fears and security dilemma dynamics during the process of status change. Clearly, states in cluster #1 – states with two groups, a higher general sense of ethnic hierarchy, and states that have actively created an ethnically disproportionate military – possess the conditions for the military to act as a catalyst for violence rather than a brake on it. In this section, we examine the case of South Africa. We use this case to illustrate how the conditions affecting the military actually operate. More importantly, we can use this case to address the second central question – what institutional solutions might exist to make the military a brake on, rather than a catalyst for, ethnic violence. South Africa: fragmented elites, status and power retention, and demographics In the scope of the cases in this study, South Africa can only be considered a success story. It is a case which could have gone horribly wrong with the military refusing to accept the state’s decision to dismantle apartheid. Briefly consider another case from cluster #1, Rwanda, the focus of many of the other authors in this volume. In order to make a useful comparison with the case of South Africa, consider one particular aspect of the genocide’s origins – the importance of the ethnic re-shaping of the Rwandan military that would have followed implementation of a peace agreement. In Rwanda, the Hutu-dominated army played a key role in upsetting the Arusha Accords negotiated settlement, sparking and implementing large portions of the 1994 genocide. The Rwandan Armed Forces’ (FAR) unwillingness to accept the status reversal that would accompany the peace settlement with the Tutsi Rwandan Patriotic Army (RPA) was fundamentally linked to the ethnic structure of the new military.11 The Hutu officers of the FAR would be weak within the new Rwandan army. They would be on the wrong side of a serious imbalance of power with the Tutsi elements drawn from the Rwandan Patriotic Front (RPF), and thus insecure if the RPF veterans launched a coup or purge. Most objectionably, the FAR was to be merged with the Tutsi RPA. “The rebels were to be granted 50 percent of the officer positions (and 40 percent of the enlisted ranks) in the combined army.”12 According to LeMarchand, “the critical issue hinged around the incorporation of the

112 R. Petersen and P. Staniland FPR elements into the Rwandan armed forces,”13 as the “basic restructuring of the 19,000 strong Rwandan army . . . proved totally unacceptable to the extremists.”14 Mann argues that “the worst blow was that the army would become 60 percent Hutu, 40 percent RPF, but with the officer corps split 50/50.”15 This would leave the more militarily-competent Tutsis with disproportionate power. Unsurprisingly, therefore, elite cadres within the army were among the leaders of the revolt and the genocide that erupted in 1994.16 Kuperman writes that “Virtually all of Rwanda’s elite military units were controlled by extremist Hutu” while “by contrast, moderate Hutu officers had virtually no effective troops at their disposal.”17 Moderate officers in the FAR were quickly overcome by genocidaires.18 The ethnic structure of military forces was inextricably linked with the slaughter that engulfed Rwanda in 1994. The course of events ran differently in South Africa. The South African military’s willingness to accept the 1994 transfer of power is largely explained by whites’ retention of both status and power within the military, the white elite’s lack of a clear consensus, and, in the context of these, the demographic minority status of whites. In combination, these factors allowed whites in the military to feel fairly safe in the aftermath of the power handover and reduced their interest in making a major intervention in politics. Whites retained both power and status within the military, even as broader changes led to an overall status reversal in the political system. The apartheid regime in South Africa maintained its power from 1948 until 1994. During this time the South African Defence Force (SADF) played a key role in maintaining the regime, and as part of this role, gained in power and funding.19 Howe writes that the SADF was “an integral part of minority white rule.”20 By the 1980s it was clearly the most powerful military in Africa and had engaged in a variety of operations, particularly counterinsurgency and subversion in the Southern African region (especially Angola).21 The SADF was primarily composed of Afrikaners, with English-speaking South Africans holding less power within the establishment.22 There was also a significant African contingent mainly made up of career enlisted men who served under a white officer corps.23 The Coloured and Asian groups were, and remain, minimally represented.24 The SADF’s political role was not clear-cut. It was not a praetorian military that frequently launched violent interventions into (white) civilian politics. However, during the period of greatest military combat during the 1980s, the SADF was invited by P.W. Botha into the highest councils of political decision-making in order to wage the “total strategy” against the armed cadres of the African National Congress (ANC), labor unions, and other increasingly mobilizing African groups.25 The SADF, thus, held an ambiguous role in the world of South African politics – clearly identified with, and (at the officer level) obviously benefiting from apartheid, but

Resentment, fear, structure of the military 113 also regarded as largely under civilian control. Those officers involved in the counterinsurgency campaigns both within and outside the country appear to have been primarily Afrikaner and more conservative than officers who had either served in more conventional formations within the army or who were part of the air force and navy. Thus the SADF could not be spoken of as a politically-cohesive organization.26 Thus ex ante it was unclear how the SADF would react to the slow but steady negotiations between F.W. De Klerk’s National Party government and Nelson Mandela’s ANC. Indeed, serious concerns about the SADF were raised at the time by observers.27 Given its power, the military posed a major potential threat to any peace deal. Moreover, by tacitly allowing its officers to engage in or facilitate “spoiling,” it could blunt movement toward a transition even without a conventional coup. The SADF did not launch a coup. Some of its elements did engage in significant spoiling, however. In December of 1992, De Klerk removed 23 officers from the military for covertly supporting a strange mixture of white conservatives, African nationalists, and assorted other groups uninterested in a multi-racial South African democracy.28 However, for the most part De Klerk did not purge the officer corps, nor did the officers turn on him. “Retaining politically hard-line officers and structures paradoxically may have aided, rather than lessened, De Klerk’s bargaining power,”29 writes Howe, while LeMarchand argues that “many hard-line praetorians evaded the purge, a move which, in retrospect, may have been a blessing in disguise.”30 Eventually the SADF even agreed to be integrated with the armed wing of the ANC (the uMkhonto weSizwe, or MK), the Azanian People’s Liberation Army (APLA), the former independent “homelands,” and the KwaZulu Police (KZP).31 Why did the SADF accept the transition from apartheid? This poses a puzzle for two reasons. First, its racial group would see its status reversed, from the ruling elite to a minority unlikely to ever lead a government again. Second, the organizational autonomy and essence of the military were threatened by the integration plan that would insert tens of thousands of Africans with radically different training and doctrinal backgrounds into the ranks, a group that would be supported by the new political power of the ANC.32 As Howe wrote in 1994, De Klerk “threatened the SADF’s corporate (and personal) identity by reducing its power and status.”33 Thus, SADF acquiescence stands as a puzzle, both for status- and organization-oriented theories of behavior. The puzzle can be explained by looking at the interaction of three variables. The most important is the way that the integration plan would leave the white officer corps largely intact and in control of the most powerful and best-trained/equipped formations in the new army. Also important, however, were the minority status of whites in broader society and a lack of consensus across the military about the desirability of political change, a fracturing in part due to

114 R. Petersen and P. Staniland De Klerk’s skillful management of the issue. An armed intervention would leave the military facing a hostile, mobilized majority, and since many officers were open to political change and uninterested in becoming involved in political rule, it was not even clear that the military would be able to unitarily act against the civilian agreements. We will begin with the terms of the integration of the new army, the South African National Defence Force (SANDF). The plan reached, agreed upon by the Transitional Executive Council and its Subcouncil on Defence, left SADF forces forming the significant majority of the SANDF, with 90,000 SADF personnel against only 22,000 MK guerrillas (the most militarily-significant African irregular/non-statutory force).34 Given that the SADF officer corps was overwhelmingly white and that there was no planned purge, this meant that the new SANDF officer corps would be overwhelmingly white. This plan was reached even before the transition occurred, providing “some sense of what would happen afterwards”35 and as a result of the demographic balance “would leave the officer corps largely white for the foreseeable future.”36 Moreover, “the size of the SADF element [in the new SANDF] and its institutional capacity, made it inevitable that it would numerically and technically dominate the process.”37 Indeed, Lamb contends that in 1994, “86% of positions at the senior and middle management levels were occupied by whites.”38 This simple demographic fact was bolstered in two further ways. First, De Klerk’s and Mandela’s parties agreed that placement of the nonstatutory forces within the command structure would be determined by a neutral team of British military officers.39 This de-politicized officer placement, and at least in the short run even more disproportionately represented whites in the officer corps because they had more advanced education and training (MK officers often had to spend several years going through a sequence of postings and training/educational programs before being placed permanently within the officer corps). Second, the ex-SADF component of the SANDF was much more combat-effective in conventional warfare than the integrating guerrilla forces. Ex-SADF officers had little to fear from the MK cadres – while the ANC’s military forces had waged a determined guerrilla campaign, they were conventionally much weaker than the ex-SADF forces.40 Thus the ex-SADF officer corps would retain both status and power within the military, blunting quite dramatically the effect of the broader status reversal taking place in South African society. They would be able to protect themselves by relying on the loyalty of the elite white forces, while also being placed in positions of command over the new African army elements. This mix of status and power, which often go hand-inhand in military organizations, allowed most of the SADF officers a certain degree of reassurance about their future. As the statistics below illustrate (see Tables 6.1 and 6.2), whites have retained majority status in the officer corps to the present, even at fairly junior levels. Eventually, one imagines

Resentment, fear, structure of the military 115 Table 6.1 Overall SANDF racial composition over time Race

1994 (%)

1998 (%)

2003 (%)

African White

38 45

57 30

62 25

Source: Le Roux and Boshoff (2005, 188).

Table 6.2 SANDF officer corps racial composition (October 2003) Rank

White (%)

African (%)

Brigadier-General to Lieutenant-General Major to Colonel Private

58 64 4

39 29 89

Sources: Lamb (2004). Note that Lamb (2004) does not include the ranks of Lieutenant and Captain. At these levels, in 2000 Higgs (2000) reports 55 percent white and 37 percent African at Captain-level (O3) and Nyanda (2000) agrees, saying that the Second Lieutenantto-Captain ranks are 55 percent white and 37 percent African.

that the upsurge in African enlistees will tilt the balance, but by then most of the elite officers involved in the early 1990s will be retired. The other two key variables that help explain the SADF’s acquiescence are the minority status of whites in South Africa and the lack of a unified military consensus in the run-up to transition. Being a minority made the prospects of a successful long-run intervention dubious, especially since the majority population had by the early 1990s become fairly mobilized in a variety of ways. There would be little prospect of an effective exit strategy – the intervening military and security apparatuses would have to embark on a sustained policy of intense repression for the foreseeable future aimed at the African population and possibly even those portions of the white population favoring a transition from apartheid. All things considered, ethnic military groups that are part of a majority are more likely to engage in a violent intervention because they will believe they can turn their action into a viable political program. This was not the case with the SADF, which could only hope for a return to the violent and tiring struggles of the 1970s and 1980s – LeMarchand refers to “a widely shared consensus among high-ranking officers that the costs of a military intervention would far exceed its chances of success.”41 Finally, the SADF officer corps was not all of one mind. While some elements within it engaged in some spoiling behavior, for the most part even conservative Afrikaners were co-opted by De Klerk. He reassured them by first actively fighting against an integration of the MK into the military, and then when that position proved untenable, by negotiating terms of

116 R. Petersen and P. Staniland integration favorable to the existing forces (as described above). Howe is worth quoting at length – analysts believe that the military is divided between supporters of a professional or “constitutionalist” force, and a much smaller praetorian, or interventionist force . . . experienced combat units have a more conservative reputation than do the navy, airforce, or medical services. Yet even within the army’s top ranks, a technocratic practical outlook prevails over privately-held ideology. Importantly, the overwhelming majority of SADF officers have stayed loyal to De Klerk.42 Thus an attempt to intervene in politics by SADF Afrikaner elites would likely have faced either opposition or neutrality by others within the military. The end of apartheid in South Africa was a dramatic event, but it was only possible with the acquiescence of the powerful, white-dominated military. Cochran writes that “South Africa’s senior military leadership has made a political transformation that few observers would have previously thought possible.”43 As sketched out above, it appears that the SADF stood by because its white officer corps was assured of continuing to control the military after transition and would have little chance of making a unified and long-term successful intervention into South African politics. In essence, the military elite was not going to face a status reversal or major loss of intra-organizational power, and thus it was willing to accept the political transition.

Conclusions South Africa and Rwanda were similar in military structure. Both possessed similar military characteristics in terms of composition and recruiting policy. Some structural variables undoubtedly shaped the divergent outcomes. Demographics and level of development worked their various effects along the lines mentioned in the previous section. However, the radically different paths in these two cases also suggest certain institutional solutions to handling the issue of status change in multiethnic societies. The dominant group in the military facing status reversal may be more likely to peacefully accept status reversal when it can receive some kind of commitment that it will retain both power and status within the military even after the broader political change. This military-specific status retention can include the following: • • • •

disproportionate over-representation in the officer corps; control over the civilian ministry dealing with defense/the military; control over elite units with high combative effectiveness; commitment from the to-be-dominant group that there will not be a purge or “victor’s justice” in the aftermath of the status reversal.

Resentment, fear, structure of the military 117 The key is that the group suffering reversal be confident that it will retain sufficient military power to protect itself should the other group violently turn on it after the status reversal; moreover, the formerly dominant group may wish to retain some degree of status within the organization. A final lesson here is that the international community must be more attuned to the resentments and fears not only in society, but within the military. The emotions of the people with the guns are hardly an insignificant matter during times of political and societal change. In the case of militaries, the two go hand-in-hand. Status/lack of status and power/fear are inextricably linked in military organizations.

Notes 1 See Horowitz 1985 for a general discussion. See Sidanius and Pratto 1999 for a discussion of status and hierarchy in the US context. Many of the ideas on resentment and fear in this section are based on Petersen 2002. 2 At the time of writing the authors were in the process of updating and expanding this data set. 3 In many states, the air force and navy either do not exist or are so small as to be meaningless. In these cases, a score derived from averaging the scores for ranks and the officer corps is assigned to the service category in order that the state in question will have a score that truly corresponds to its proper place on the 0–9 scale. Thus, if a country has a score of 1.0 for the ranks and 2.0 for the officer corps, then a score of 1.5 is assigned for the services. The result is a total score of 4.5 on a 9.0 scale. 4 Combinations of the above criteria are possible. For example, informal networks may come to dominate the state and then initiate a particular ethnic recruiting strategy. If the ethnic group in question had been the major source of military manpower previous to taking control of the state, the situation now becomes rather ambiguous. Is the informal network or the state now determining recruitment? It is nearly impossible to determine, so the score given here is 1.5, a compromising measure. 5 Level of development was measured with per capita GNP figures taken from the National Basic Intelligence Factbook (CIA 1979) (1979 was the base year for this study due to the nature of the resources available at the time). The distinction between military and civilian governments was derived from this same source and was coded as a simple dichotomy. Colonial history was coded into three categories: (1) non-British colonial; (2) British colonial; (3) non-colonial. Geographic region did not lend itself to multiple regression analysis and was analyzed separately according to four regions: Africa, Asia, Europe, and the Middle East. The size of the state will be measured by population figures which again are taken from the National Basic Intelligence Factbook. Four roughly equal categories are again devised: (1) below 3.5 million citizens; (2) 3.5 to 12.5 million citizens; (3) 12.5 to 25 million citizens; (4) over 25 million citizens. There was a subjective element in deciding what groups are “significant” groups in each country. The reasoning involved in the most ambiguous cases is explained in a data appendix available on request. Three categories are developed in the operationalization of this variable: (1) states containing two significant ethnic groups; (2) states with 3–4 significant ethnic groups; (3) states with five or more significant ethnic groups. A dominant group is defined in the coding as one whose

118 R. Petersen and P. Staniland

6 7 8 9 10 11 12 13 14 15 16 17 18 19

20 21 22 23 24 25 26 27 28

29 30 31 32 33 34 35 36 37 38 39

percentage of population is twice that of any other ethnic group in the state. Again, this was coded as a dichotomous variable. Thus, with Murshed’s chapter and others in this volume, we have several mechanisms by which development seems to inhibit violence. Regarding Belgium, see Young 1982, Chapter 2. On Burma’s civil–military relationship, see Lissak 1976. Silverstein 1980, 237. Keegan 1979. LeMarchand 1994, 592. Kuperman 2001, 11. Dallaire 2004, 67 refers to the Tutsi RPA’s “superior leadership, training, experience, frugality, mobility, discipline, and morale.” LeMarchand 1994, 596. Ibid., 597. Mann 2005, 441. Ibid., 452. Kuperman 2001, 80. Mann 2005, 453. In the early 1980s, 4.4 percent of South African GDP was devoted to SADF according to le Roux and Boshoff 2005, 178. By comparison, even during the current period of military challenges, the US only spends 3.9 percent of GDP. Howe 1994, 30. Nyanda 2000 refers to the armed forces being in a “state of undeclared mobilization” throughout the apartheid period. Howe writes of SADF as “an ethnic army mobilized.” Higgs 2000, 47; Kasrils 1998. Coloureds and Asians combined make up no more than 9 percent of any level in the military, from private to high command, and in most are below 5 percent. Howe 1994, 33–34; LeMarchand 1996, 598. Howe 1994, 31 argues that “generalizing about the SADF’s political beliefs and actions is risky, given its internal divisions.” Howe 1994; Southall 1992; and Williams 1993 all reflect contemporary concern over the behavior and future of the SADF. Howe 1994, 36 discusses how SADF officers supported the shadowy “third force” of anti-settlement vigilantes, in particular “conservative whites, and inter alia, the Inkatha Freedom Party, the Zionist churches, Homeland government supporters, and those Asians and Indians fearful of their minority status in a black-run country.” Ibid., 50. LeMarchand 1996, 598. Mills 1999. Higgs 2000, 47–48 outlines the very different structure, culture, and doctrine of the non-statutory forces. Howe 1994, 31. Le Roux and Boshoff 2005, 187. Higgs 2000, 48. Williams 2002. Le Roux and Boshoff 2005, 186; however, they also point out the “political leverage of the MK.” Lamb 2004. Higgs 2000, 48–49. This induction process began at a low level even before the official transition in 1994, after the 1993 interim constitution was enacted.

Resentment, fear, structure of the military 119 40 Pre-transition SADF doctrine and operations were structured around “highly mobile forces with high firepower and strategic reach.” Le Roux and Boshoff 2005. 41 LeMarchand 1996, 598. 42 Howe 1994, 41. 43 Cochran 2000–1.

7

Violence as politics ETA and Basque nationalism André Lecours

On March 22, 2006, ETA declared, for the first time in its history, a permanent cease-fire. In a videotaped statement, three hooded ETA members announced that this initiative had for its objective “the launching in the Basque Country of a democratic process for building a framework where our rights as a people can be recognized and where all political options can be explored.”1 In the three years prior to this move, the terrorist group had limited its activities to low-level detonations that sometimes caused injuries from flying debris but not death. In addition, ETA had made a habit of contacting the Basque daily newspaper Gara to warn that explosions were forthcoming in specified locations. Its intent clearly was not to kill, but simply to make a statement about its continued armed struggle and operational capacity while recalling the crux of the conflict: ETA does not accept the rule of the Spanish state over the Basque Country. Before the 2006 announcement of a permanent cease-fire, there was much speculation that ETA would, if not disband, at least cease its violent activities. After all, its last killing dated to May 2003; some of its leaders were captured in France; the leader of the outlawed radical nationalist party Batasuna Arnaldo Otegi was talking about choosing the “peaceful route”; and influential jailed members publicly advocated abandoning the armed struggle after having concluded that it ultimately played into the hands of the enemy (the Spanish state). The announcement of the permanent cease-fire has triggered a lot of hope in the Basque Country, but also much caution about what could realistically be expected. The abandonment of the cease-fire in 2007 dashed hopes in the short-term. Even if it proves to be true that ETA has definitely ceased its violent activities, the Basque Country, between 1978 and 2006, remains a puzzling case of political violence in a developed liberal democracy. The Spanish state was always strong enough to deter large-scale violence, which meant that ETA was limited to adopting guerilla tactics. More importantly, the Spanish state allows for the expression of nationalist claims through democratic channels. This is the product of the 1978 Constitution, which produced a federal system2 where the Autonomous Community of the Basque Country enjoys substantial autonomy and

Violence as politics 121 where political grievances of all types can be brought into various political arenas, most notably parliaments, parties and networks of intergovernmental relations. The Spanish case therefore suggests that political violence is not simply a function of (a lack of) security and that violence can occur even under an absence of threat. In this respect, however, Spain and ETA also suggest that violence in secure conditions is likely to be constrained and targeted. Two questions come to mind when considering such a situation. First, if violence does not stem from insecure conditions, how should it be understood? This chapter argues that ETA violence is a form of politics. In an effort to make sense of violence in the context of security, the chapter identifies two dimensions of ETA’s violence. The first dimension is instrumental. For ETA members and supporters, violence has been seen as a way to achieve the concrete political objective of Basque independence. The second dimension is symbolic. Violence expresses not only the rejection of Spain, but also the existence of a radical Basque nationalist community. The second question raised by the Basque case is how to explain the position of ETA in the democratic period as an organization enjoying modest although not insignificant support in the Basque population. A measurement for acceptance of ETA violence in the Basque Country is the electoral support for radical nationalist parties (Herri Batasuna, HB, and its successors), which have acted as the political arm of ETA. This support has ranged from 10 to 18 percent in the democratic period, but in the last two Basque elections it remained at the lower end of that range. The chapter suggests that this real yet modest support for ETA throughout the democratic period stemmed from tensions involved in Spain’s institutional framework of autonomy and the promotion of the Spanish nation by the state. The Spanish system of Autonomous Communities provides the Basque Country with considerable autonomy. At the same time, it lacks legitimacy among the Basques since the referendum on the Constitution that created this system was boycotted by a large segment of Basque society who demanded that the document recognize a Basque right to self-determination. Persisting frustration over what is perceived by some Basques as a Constitution imposed upon the Basque Country means a pool of discontents that can be mobilized by the radical nationalist movement. Radical Basque nationalism also feeds off the Jacobin vision of the Spanish nation promoted by the center-right Partido Popular (PP), which leaves little room for the type of national differences expressed by Basque governments. However, this vision is not hegemonic, and the Spanish nationalism projected by the left-wing Partido Socialista Obrero EspaZol (PSOE), less centralist and more tolerant of difference, is acceptable for a great number of Basques. The switch in government in 2004 explains in part ETA’s decision to announce a permanent cease-fire in 2006. This chapter is divided into three sections. The first section discusses the

122 A. Lecours instrumental and symbolic dimensions of ETA violence. In doing so, it contrasts ETA’s instrumentalist rationale for violence during the Franco era to its mostly symbolic use in the democratic period. The second section explains why ETA survived in a liberal-democratic environment yet at the same time was unable to attract substantial support. In providing such explanation, it emphasizes the projection by the democratic Spanish state of different visions of the Spanish nation depending on the government in place. The third section looks at the strategies of the Spanish and Basque government in dealing with ETA with an emphasis on the plan Ibarretxe and the 2006 cease-fire.

Framing ETA violence ETA (Euzkadi ta Askatasuna, Basque Country and freedom) is a creature of the Spanish authoritarian state. Its existence derived from policies of political and culture repression conducted in the Basque Country by the Franco regime in the name of Spanish unity. More specifically, the creation of ETA should be understood as a consequence of a restlessness toward the seemingly ineffective strategies of the Partido nacionalista vasco (PNV) and a disillusionment vis-à-vis the chance of the United States toppling the dictatorship. From a more substantive point of view, the founders of ETA disagreed with the PNV on both the process that could lead to the liberation of the Basque Country and its end result. The founding members of ETA were young and had little confidence that any solution to the Basque question could come through the Spanish state. In this context, they favored the creation of an independent Basque state rather than a return to the Statute of Autonomy (Clark 1984: 25–26). Violence as a political instrument: the Franco period and beyond ETA has killed more than 800 people since its creation. This violence in the Basque Country has been, for the most part, constitutive of politics rather than simply its product. The instrumental dimension of ETA’s armed struggle was at the heart of this violence-as-politics dynamic. The notion of instrumentalism refers to purposive action, that is, it involves a connection between means and end. In politics, actors make instrumental use of violence if they kill or intimidate to reach political objectives. In the case of ETA, this objective was the creation of an independent Basque state. ETA’s conception of violence as a tool to achieve the independence of the Basque Country has varied over the years. In the context of the Franco dictatorship, the instrumental use of violence involved the idea that armed struggle, through guerilla tactics, could push the forces of the Spanish state out of the Basque Country. This rationale for violence was inspired by Vasconia (1963), a book written by Basque nationalist Frederico Krutwig. In this book, Krutwig developed a

Violence as politics 123 theory of how the “occupation” of the Basque Country would end. This theory revolved around the idea of a cycle triggered by action and perpetuated by repression; the corresponding political strategy was a guerilla war. Krutwig explained that if the state’s repression were met by equally punishing acts of violence (killings, torture, etc.), the use of violence would become more indiscriminate (since the guerilleros would be difficult to find) and its level would heighten. As a consequence, the Basque population would be increasingly mobilized through this “upward spiral of resistance” (Sullivan 1988: 42–43). In the end, the Franco regime would effectively be chased from the Basque Country, and a Basque government would gradually assume the administration of the territory. It could be argued that the action–repression–action strategy contributed to bringing a struggling dictatorship to its knees in the late 1960s and early 1970s, as international media attention focused on the confrontation between ETA and the Franco regime. The democratic transition that followed Franco’s death in 1975 meant that the most straightforward understanding of instrumental violence (action–repression–action leading to the withdrawal of the Spanish state from the Basque Country) was no longer adequate. Mass and indiscriminate repression could no longer be reasonably expected now that the dictatorship had crumbled. The transition process, however, was an extremely volatile period whose ultimate outcome was very much open-ended. From a radical Basque nationalist perspective, Basque independence could not be ruled out. In this context, there was an opportunity for ETA to find another instrumental use for violence: the de-stabilization of the Spanish state. The politics of Spain’s transition toward democracy consisted primarily of enacting democratic change without alienating political forces close to the Franco regime. This balancing act was most apparent on the question of decentralization. The centralist legacy of the Franco dictatorship strongly shaped the transition period. On the one hand, Francoist elements were extremely wary of any regional autonomy, which they understood as threatening the unity, if not the survival, of Spain. On the other hand, forces that had opposed Franco saw autonomy as an essential part of, and safeguard for, any democratic system. In this context, ETA sensed an opportunity to shape the political dynamic of the transition. Until 1974, ETA had never killed more than 20 people per year, but between 1977 and 1980 there were 240 deaths attributed to ETA violence (Mees 2003: 34). Whether this was part of ETA’s strategy or not, the flurry of killings served to strengthen the PNV, which could argue that its “moderate” demands needed to be satisfied to avoid chaos and a failure of the democratic transition. ETA put forth, in early 1978, conditions for a cease-fire known as the KAS alternative that featured, among other things, amnesty for political prisoners and a Statute of Autonomy recognizing the sovereignty of the Basque Country. Although the KAS alternative was dismissed by most

124 A. Lecours Spanish politicians, many of its elements would be found in the Basque Statute of Autonomy ratified in 1980. ETA violence shaped transition politics in the Basque Country insofar as it provided incentives for Spanish politicians to satisfy moderate demands for autonomy. In the post-transition era, instrumental uses of violence have been less straightforward, but I would argue that they do exist. It is difficult to speak with certainty about this because there has not been, in recent years, an “action–repression–action” blueprint of the type put forward during the dictatorship, or a yearly number of killings comparable to the transition period. In the late 1990s, ETA sought to use violence, or more accurately the spectre of violence, to achieve its objectives by taking control of the political agenda. It was able to do that as a result of a major re-alignment of political forces in the Basque Country (Mees 2003). In 1988, the Pact of Ajuria-Enea, signed by all political parties operating within the Basque political system, except for radical nationalist HB, condemned political violence as illegitimate and contrary to the Basque popular will.3 The Pact formalized two political blocs: democratic parties and HB. Until 1998, the PNV’s strategy was to govern in a coalition with the Socialists and to isolate HB. This party alignment was broken by the PNV in 1998 when it accepted the support of representatives from the re-named Herri Batasuna, Euskal Herritarok (EH). These moves were part of a larger re-orientation of the PNV that has been dubbed the “sovereignist” or “separatist” turn (de la Granja Sainz 2002: 58). The major PNV initiative exemplifying its sovereignist turn was the Pact of Lizarra signed on September 12, 1998 with Eusko Alkartasuna (EA), the Basque branch of the left-wing Izquierda Unida (IU) and, most significantly, HB as well as a myriad of radical nationalist civil society organizations part of a loose ensemble known as the National Liberation Movement. Neither the PSOE nor the PP, the two main Spain-wide parties operating in the Basque Country, signed the document, which they criticized as being a sell-out to ETA. For the PNV, greatly inspired by the Good Friday Agreement in Northern Ireland, the ultimate objective of this initiative was political change. The PNV had come to consider the pact of Ajuria-Enea, as well as the political relationships built around it, as a stalemate: political violence was condemned but the status quo prevailed. This being said, the PNV considered that change could only occur in the absence of violence. The Declaration of Lizarra represented a way for the PNV to set up an ETA cease-fire with the intent of driving a process of political and institutional change. There was an inherent paradox to the Declaration of Lizarra.4 The Declaration needed to contain the necessary elements to be sold to ETA and radical nationalism for a cease-fire to occur. For example, the declaration framed the conflict in the Basque Country as political in nature and involving the Spanish and French states (a radical nationalist line); it specified

Violence as politics 125 that a resolution process should be open to all parties (so ETA could be an actor in this process); it also stated that every and all issues pertaining to the conflict should be dealt with (theoretically including the agenda of radical nationalism). In sum, the Declaration of Lizarra was a document written by Basque nationalists for a Basque nationalist audience (Mees 2003: 140). As a result, the document’s potential to engage the Spanish parties and, by extension the Spanish government, into a meaningful discussion over the political future of the Basque Country was limited. Why did ETA decide to declare a cease-fire in the Fall of 1998? There were many reasons (Gillespie 1999: 119–136), but the most important was that, in all likelihood, ETA’s leadership came to the conclusion that the organization could, at that point, better drive a process of change through political rather than military means. Tellingly, in its September 1998 communiqué published in Euskadi Información, ETA spoke not merely of solving the conflict, but more specifically of a unique opportunity to move forward toward sovereignty.5 In secret negotiations between ETA, the PNV and EA before the cease-fire, the three parties stipulate three political goals: cooperation of all parties favorable to the construction of Euskal Herria; non-cooperation with Spanish parties; and, most importantly, creation of new institutions featuring representatives from the seven provinces (the three provinces of the Autonomous Community of the Basque Country, Navarre and the three provinces of the French Basque Country) as a way toward terminating the existing institutional framework and proceeding toward the political unification of the Basques (Mees 2001: 813). This agreement was really the historical program of ETA. ETA only agreed to stop using political violence, which was a good strategic choice at that point in time considering the context of lowered social tolerance for violence and the aggressive policing strategy of the Spanish government. It did not, through this secret agreement and the subsequent declaration of Lizarra, relinquish the role of spearheading the Basque struggle but simply changed methods. To ensure that it would keep control of the process, ETA made sure to warn the PNV and EA that its cease-fire was conditional upon progress in the realization of these objectives (Mees 2001: 813–814). ETA used its capacity and willingness to kill to capture the political agenda by promising an absence of violence for an indeterminate length of time. Of course, the cease-fire lasted only 14 months and did not lead to any tangible results for ETA. In hindsight, there was never a real chance that the cease-fire could yield a long-term solution to the conflict in the Basque Country. The main problem was that it hinged on the full realization of the maximalist program of radical nationalism. Tellingly, ETA revoked its cease-fire after the PNV and EA rejected a new proposal for the election of a Basque “national” parliament representing inhabitants of the seven provinces as a way to effectively break out of the existing institutional order (Mees 2003: 111–112).

126 A. Lecours Violence as a symbol: the democratic period ETA violence did not only have an instrumental dimension. For the last 20 years, there was never a possibility that ETA could “militarily” defeat the Spanish state, nor was there a clear chance that it could de-stabilize the state to the point of achieving independence. In the late 1990s, ETA sensed an opportunity to take control of the political agenda and to make things happen along the lines of its program. However, this window of opportunity was not present through the 1980s and for most of the 1990s, and it effectively closed after the end of the 1998 cease-fire. Violence, however, has been a permanent feature of ETA in the democratic period, which means that it must have some other dimensions than pure instrumentality. A good case can be made that ETA violence had a symbolic dimension insofar as it represented and crystallized a political community. Daniele Conversi has argued that the lack of a functional common language in the Basque Country meant that violence became a defining element of the Basque nation (Conversi 1997). From this perspective, the presence of such a language in Catalonia explains the absence of violence there, despite a similar history of political and cultural repression as in the Basque Country. This is a theory that had some merit in the 1980s for two reasons. First, the Basque institutions were still recent, and they might not have completely taken over as the main reference point for the Basque identity. In this context, ETA violence, which had served as a rallying force for the Basques during the dictatorship, represented an element of continuity. Second, ETA members tended to be then viewed as the ‘lost sons’ of the nationalist family whose methods might have been questionable but whose objectives were honorable. The corollary of this widespread view was a substantial tolerance for violence in the Basque Country. In the last decade, the view of ETA members as patriots and dogooders has lost much support, and tolerance for violence in Basque society is now low.6 The idea that violence served a function of symbolic representation should not be lost. However, the community represented and re-produced by ETA killings was not so much the Basque community at-large but rather the radical Basque nationalist community. ETA was never a band of isolated fanatics. If we use the performance of radical nationalist parties as an indication of sympathy, we have to recognize that approximately 10 percent of the Basque population is, at least, not critical of ETA, or offer it tacit support.7 This figure corresponds to the radical Basque nationalist community, which represents a social network featuring friends and relatives of the hundreds of “political” prisoners (Hamilton 2000: 153–171), as well as radicalized youths involved in “street fighting” (Kale Borroka). This network serves to radicalize youngsters through political socialization in various bars and organizations (Kasmir 2003: 39–68) and expresses the existence of radical nationalism via carefully-organized demonstrations

Violence as politics 127 against the Spanish state (Raento 1997: 191–204). From the social network of radical Basque nationalism emanates a myriad of civil society organizations (trade-unions, cultural associations, etc.) that form an eclectic social movement, the National Liberation Movement, whose political expression is the (now outlawed) party Batasuna.8 The political party of radical Basque nationalism as well as the segment of civil society upon which it rests are eclectic coalitions that were held together by violence. The announcement of the permanent cease-fire in 2006 raised serious questions about the relationship between this radical segment of civil society and the political scene. ETA has been the guardian of this radical nationalist community because without this network of social support, the organization could not exist (Mata López 1993). ETA was, for the longest time, successful against Spain’s policing strategy, in large part because of its potential for re-generation, which it owed to a structured radical nationalist sub-culture. Violence is what has kept this sub-culture alive, which is why ETA’s latest cease-fire could have radically transformed Basque politics and society. ETA is an image (Watson 1999), the image of violence and of the radical Basque nationalist community. It was also a concept (Douglass and Zulaika 1990), a simplified representation of something more complex. It is telling, for example, that despite ETA’s many schisms, which have produced splinter groups, the organization always remained “ETA” (Douglass and Zulaika 1990: 252). This understanding of ETA is supported by the fact that it has never looked to kill a substantial number of people at once. Most probably, its leaders recognized that such action was likely to lead to a fatal backlash from the larger Basque society, but there was also the more fundamental issue that a single killing was enough to express and structure the continued existence of a radical Basque nationalist community. From this perspective, ETA murders are ritualistic; victims are “offered” to the radical nationalist community (Zulaika 1988). Of course, one could see an instrumental aspect to symbolic murders insofar as the reproduction of a community whose survival was necessary for ETA’s own continued existence. For example, in 1983, ETA murdered a former member, María Dolores Gonzáles Catarain (a.k.a. “Yoyes”), who had spent time in Latin America (Douglass and Zulaika 1990: 251). The killing of Yoyes was a message to ETA members who might have been considering “re-insertion” into society, that is, leaving the organization and obtaining a formal pardon from the Spanish government. This particular case is one of the rare instances where ETA miscalculated the public reaction toward its violence (Yoyes was murdered walking her baby in a public plaza). After the end of the 1998 cease-fire, ETA’s violence was primarily symbolic. The organization began killing shortly thereafter, to make a statement about its continued existence and the existence of radical Basque nationalism. Then, sensing a sharp decrease in social tolerance for violence (see Figure 7.1), ETA adopted the strategy of detonating small bombs but often “calling them in” beforehand. There was obviously no pure

128 A. Lecours

Ends yes, means no 12%

Does Critical not know justification 2% 3%

Before yes, now no 17%

Total rejection 60%

Indifference 2% Fear 4%

Figure 7.1 Attitudes of the Basques towards ETA (source: Euskobarómetro November 2005).

instrumental value to these actions, but they were symbolically important. Political and societal actors in the Basque Country did not consider the situation “normalized” as long as ETA made threats and showed some capacity to detonate bombs or deploy shooters.

Political institutions, Spanish nationalism, and ETA ETA’s use of violence has never been widely supported in the Basque Country. As indicated earlier, votes garnered by radical Basque nationalist parties at any one recent election suggests that approximately 10 to 15 percent of the Basque population is at least willing to accept violence. What explains that a small but certainly not insignificant portion of the Basque population considered violence as a legitimate form of politics? ETA’s position in the Basque Country is the product of tensions inherent to both the Spanish framework of political territorial autonomy and the conceptualization of the Spanish nation by the main Spanish-wide parties. The simplest answer to the question of why ETA has never enjoyed support from a majority of Basques in the last two decades (the way it most probably did during the dictatorship) is that the 1978 Constitution transformed Spain into a liberal-democratic state where the Basque

Violence as politics 129 Country enjoys significant autonomy. The Constitution established a system of 17 Autonomous Communities that could assume power over, among other things, social assistance, transportation, culture, tourism, and some public works.9 With the Basque Country, Catalonia and Galicia in mind, it also created the category of “historical nationalities,” units that could acquire autonomy faster than the “regions.” The Constitution, therefore, allowed for the negotiation of a Basque Statute of Autonomy that defined the jurisdictions of the Basque government and its relation with the state with respect to policy-making (Title 1). The Statute makes some statements of symbolic importance such as proclaiming Euskera the language of the Basque people (and one of the official languages of the Basque Country along with Castilian), and specifying the colors and outlook of the Basque flag. It also stipulates that fiscal relations between the Autonomous Community of the Basque Country and the Spanish state are to be governed by the conciertos económicos. These conciertos (agreements) allow the Basque Country or, more specifically, its three constituent provinces, to levy almost all taxes on its territory. A portion of these revenues (the Cupo) is then transferred to the state for the financing of services it assures in the Basque Country. Why, considering this framework, did support for the violent separatist claims of ETA hold through the democratic period? One important element is that the constitutional and institutional arrangement suffers from a legitimacy deficit. In the Basque Country, the referendum on the constitution was a volatile and divisive affair. ETA was not created for, and never had experience in, contesting elections, but radical Basque nationalism found expression through the political party HB. HB advocated a “no” vote in the referendum because the Constitution did not acknowledge the “original sovereignty” of the Basques, that is, it did not recognize them a right to selfdetermination. The moderate nationalists of the PNV called for abstention on similar grounds. As a result, there was a very high abstention rate in the Basque Country for the 1978 referendum and support for the Constitution was lower in the Basque Country, where it ranged from 63.9 to 71.4 percent depending on the province, than in Spain as a whole where it was 87.8 percent (Penniman and Mujal-Léon 1985: 241). The consequences of this process of constitutional ratification are still felt today since the Constitution lacks legitimacy in the Basque Country (Moreno 2004: 29–51) whereas it tends to be viewed almost as a sacrosanct document by politicians in most other parts of Spain. This legitimacy gap served ETA well since it contributes to the negative attitudes held by a substantial number of Basques toward the Spanish state. Therefore, the particular nature of the state might have changed from authoritarian to democratic in the 1970s, but its alien status vis-à-vis an important segment of Basque society remained. The modest but non-negligible support for ETA in the democratic period also owed much to the existence of different conceptualizations of

130 A. Lecours the Spanish nation. Spanish nationalism is a differentiated phenomenon, which ranges from Jacobin visions of the country, where the emphasis is on unity, cohesion and centralism, to more multi-composite views highlighting plurality, diversity and decentralism. The state is, therefore, crucial in explaining attitudes toward ETA, not only for its management policies and strategies but also, and more fundamentally, for the type of nationalism it projects. A very influential vision of Spain, especially popular on the Right, carries the implicit assumption of a nation characterized and united by its Castilian roots and language. For example, PP politicians can still be heard recounting the development of the Spanish nation using references to the Reconquista against the Moors.10 From this perspective, Spain’s history, thereafter, is about the expansion of Castile and the making of the Spanish state, empire, and nation. Spain’s national holiday, el Día de la Hispanidad, commemorates the “discovery” of the Americas. Tellingly, the Aznar governments, together with the Real Academia de la Historia, were very much concerned with the way Autonomous Communities taught history in school, fearing a “distortion” of the Spanish historical developmental process (Resina 2002: 382). For this brand of Spanish nationalism, multilingualism is viewed with suspicion since it is considered to threaten one of the bonds holding Spain together. This vision is strongly present in Spain’s language regime. Different languages are spoken in Spain, but only Castilian is an official language of the Spanish state, which all Spaniards have the duty to know. Catalan, Basque, and Galician are official only within the corresponding Autonomous Community. Despite the multilingual nature of society, Castilian (which, outside Spain, has come to be referred to simply as “Spanish”) seems to assume a status as the only legitimate language when it comes to expressing the Spanish national identity.11 For example, there is no formal status for minority languages in the Spanish Parliament while in the Senate the use of these languages is allowed only one day per year. The Spanish national identity card is written only in Castilian. The Aznar government, which vigorously promoted the Spanish identity through linguistic references, pushed Autonomous Communities to emphasize a strong education curriculum in the Castilian language and literature.12 For this Spanish nationalism suspicious of diversity and decentralization, nationalist movements such as the Basque are reactionary and even totalitarian since, somewhat ironically, they are viewed as imposing a single culture upon residents of their Communities (Núñez 2002: 730). From this perspective, there is a “demonization” of sub-state nationalism that features prominently in the construction process of the “new” Spanish identity. The Basque Country has been a prime reference in this discourse of Spanish nationalism. Moderate Basque nationalism is viewed as endangering the unity of the country with its insistence on self-determination and the demand to modify the constitutional framework and the Basque

Violence as politics 131 Statute of Autonomy. The most intense denunciation of sub-state nationalism, however, came in the form of the discourse about terrorism and ETA. ETA became a central reference point against which Spain is being defined. Through this reference, the Spanish state could be put in a positive light since it was depicted as a modern, liberal and democratic entity struggling against violent illiberal terrorists seeking to destroy Spain and all the positive values for which it stood. It is no coincidence that the PP’s hard-line policy toward ETA was so successful among the Spanish population; it was presented as a basic good-versus-evil struggle featuring Spain and the enemies of Spain.13 This brand of Spanish nationalism was, somewhat paradoxically, favorable to ETA, not only because it tended to give credibility to the organization’s claims that the Spanish state was fundamentally intolerant of the Basque identity but because it polarized Basque society by associating all Basque nationalists with terrorism. The result was a crude dichotomy between Basque nationalists and non-Basque nationalists, which juxtaposed ETA alongside the moderate Basque nationalists rather than isolating it. The PP’s conflation of nationalism and terrorism led to policies that proved very controversial in the Basque Country, for example, shutting down, through requests to Spanish courts, Basque political and cultural organizations for supposed collaboration with ETA. The prime examples of this strategy was the closing of the Basque language newspaper Egunkaria in 2003 and the outlawing of Batasuna (the re-named Euskal Herritarok) for a three-year period in 2002 and then permanently in 2003. Both moves triggered sharp criticism from all sectors of Basque nationalism. This brought together, albeit temporarily, moderates and radicals and worked to further deteriorate the relationship between the Basque government of the PNV and the Spanish government. In fact, Spanish Prime Minister Aznar simply refused to speak to its Basque counterpart Juan José Ibarretxe. For example, asked about the possibility of establishing dialogue with the Basque leader, Aznar said: “Dialogue to achieve what? I have nothing to say on the question of self-determination.”14 In this context, the political climate in the Basque Country from the end of the 1998 cease-fire to 2004 was one of intense confrontation and polarization, fertile ground for ETA. If Spanish nationalism was consistently expressed in such a way, it could be expected that ETA’s support would be stronger. Sometimes, there are political barriers to this expression. For example, the first Aznar government (1996–2000) did not have a majority of seats in parliament and relied on the support of the moderate Basque and Catalan nationalist parties to govern. Not surprisingly, this government had a less critical discourse toward nationalist claims than its successor. There is, however, a different brand of Spanish nationalism than what has been articulated by the PP. This alternative view of Spain, which corresponds to the PSOE’s position, incorporates more substantively the country’s cultural and

132 A. Lecours linguistic diversity. It is more comfortable with the political recognition of these types of differences through, for example, asymmetry in the Autonomous Community framework, and more accepting of decentralization as a principle of governance. This brand of Spanish nationalism is less rigid than the first version when it comes to discussing institutional and/or constitutional change with the Basque and Catalan governments insofar as it is more willing to accept a dynamic view of the existing frameworks. In this context, the election of a PSOE government in 2004 softened the support for ETA, an organization that thrived in a polarized political and social context. Significantly, one of the first political moves made by the new Spanish Prime Minister Zapatero was to meet with Basque president Ibarretxe to re-establish dialogue. The likelihood that ETA would have announced a permanent cease-fire with a PP government in Madrid was most certainly lower than it was with the Socialists in power. The existence of competing visions of Spain and of institutions accommodating diversity but suffering from a legitimacy deficit explains the paradox of ETA as a violent organization that enjoyed modest but very real social support in a liberal democratic state. What efforts did key political actors, namely the Spanish and Basque governments, make to combat ETA? We now turn to this issue.

The plan Ibarretxe: undermining radical Basque nationalism? ETA violence has prompted responses from both the Spanish and Basque governments. For the Spanish government, the objective has been to eliminate the organization representing the only asterisk in the progression of Spain to the status of a “normal” liberal democracy. The Spanish government periodically employed a political strategy for achieving this objective, that is, it negotiated with ETA leaders, often despite claims to the contrary. After the collapse of the 1998 cease-fire, however, such a political approach was rejected in favor of a straight policing strategy. Spanish police forces, benefitting from the collaboration of their French counterparts, captured many influential ETA members, and the Spanish government sought to mortally wound the organization by hunting down and jailing its leaders. Under the PP government, the Spanish state alienated many moderate Basque nationalists as it widened the net of its policing operations by coming down on radical nationalist networks whose members became automatically suspected of connections to ETA. Given political polarization in the Basque Country, this same strategy was, at the same time, welcomed by a significant portion of the Basque population, that is, by non-Basque nationalists. The current PSOE government took a different approach, choosing to separate more clearly

Violence as politics 133 the struggle against ETA from the political issue of Basque nationalism. In this context, policing ETA became a strategy that could find even greater legitimacy in the Basque Country. In the three years or so preceding the 2006 declaration of a permanent cease-fire, ETA was not a huge problem for the Spanish government insofar as it had not killed since 2003. The same could not be said for the PNV-led Basque government. The PNV has long been seeking greater autonomy for the Basque Country, but these claims were typically rebuffed by a Spanish government invoking the connection between Basque nationalism and violence. For the Basque government, the elimination of political violence was a more pressing political problem than for the Spanish government, which meant that waiting for Spanish police forces to completely destroy ETA is a risky and costly option. Between 2001 and 2003, the Basque president developed a proposal for re-structuring the relationship between the Autonomous Community of the Basque Country and the Spanish state, the so-called plan Ibarretxe. The plan Ibarretxe proposes a type of confederal arrangement, or “statute of free association” between the Basque Country and Spain. It involves the creation of a Basque judicial system; exclusive powers for the Basque government over virtually all major policy sectors such as culture, education, social policy, and social security, transportation, environment, and natural resources (articles 46 to 54); direct representation in European Union institutions (article 65) and the right to have representation abroad and to sign international agreements in fields that are domestically in the jurisdiction of the Basque government (article 67); the establishment of a Basque Country–Spain Bilateral Commission, as a forum of coordination and information exchange (article 15); a special tribunal to arbitrate conflicts (article 16); and mechanisms for amendments preventing unilateral change from Madrid (article 17). The rationale behind the proposal is to engage the Spanish government into negotiations while at the same time using an ambitious project to draw support from radical nationalism and therefore undermine ETA. Radical nationalists initially criticized the plan for falling short of independence and not corresponding to their (maximalist) conceptualization of the Basque nation, which includes Navarre and the French Basque Country. They argued for the need of an independent state before forging any type of association with Spain. In the Fall of 2004, however, some radical nationalist representatives in the Basque parliament unexpectedly opted to support the plan Ibarretxe, judging that political violence was unlikely to bring about change. At that point, the initiative seemed to be working. The results of the 2005 Basque elections, which were fought in large part around the plan Ibarretxe, cast at the time serious doubts on the notion that support for radical nationalism could move toward the PNV and, therefore, seriously undermine ETA’s structure of social support since the radical vote went up two percentage points.15 This being said, the strongly nationalist position of the plan most likely

134 A. Lecours helped convince ETA that advances were more likely in the absence of violence. The most striking display of potential for change with a PSOE government in Madrid came from the negotiations, in late 2005 and early 2006, between the Spanish government and the Catalan Generalitat for a reform of the Catalan Statute of Autonomy. As a result of these reforms, approved in referendum in June 2006, Catalonia will have increased fiscal autonomy and be recognized as a nation. The future of the plan Ibarretxe and other proposals remains very much in doubt. The PNV’s initial strategy was to hold a referendum on the proposal even if the Spanish government refused to enter negotiations on it. A referendum, however, would probably be a recipe for violence rather than a solution to it. In the case of a loss, ETA might not accept the result and would pursue its violent activities. Even if the vote produced a strong majority in favour of the Ibarretxe plan, violence would still be a concern. When the plan was first put forward, the Spanish government, then formed by the PP, reacted swiftly by pushing legislation, supported by the PSOE, specifying that anybody holding a referendum on an anti-constitutional proposal could be jailed. This suggests that the Spanish government would be unlikely to accept a winning result for the Basque nationalists. The type of tense stalemate that would undoubtedly emerge could bring a surge of ETA violence.

Conclusion ETA violence should be understood primarily as a form of politics rather than purely as the consequence of a conflict. The organization has used violence in the hope of achieving its objective of independence for the Basque Country and for the re-production of the radical Basque nationalist community. In the democratic period, the instrumental dimension of violence was more complex than during the dictatorship since the notion of defeating the state militarily became absurd. In the late 1990s, ETA saw an opportunity to drive the political agenda toward its program through a conditional cease-fire, but this tactic failed. In the post 1998 cease-fire period, there was little instrumentalism to ETA’s (much diminished) violence. The tolerance for ETA and political killings became lower than ever in the Basque Country, which partly explains the 2006 announcement of a permanent cease-fire ETA’s resumption of violence in 2007 is linked to symbolic rather than instrumental politics. ETA survived into Spain’s democratic period because the institutional and political accommodation of Basque nationalism by the Spanish state was challenged by the moderate nationalists of the PNV. Institutionally, the system of Autonomous Communities is contested in the Basque Country because the Constitution that created it suffers from a legitimacy deficit resulting from nationalist protest over the absence of recognition of a Basque

Violence as politics 135 right to self-determination. Politically, the more rigid brand of Spanish nationalism associated with the PP leaves little room for expressions of Basque national distinctiveness. Discontent stemming from these institutional and political forms was the backbone of support, or at least tolerance, for ETA. At the same time, the substantial degree of autonomy afforded to the Basque Country within the current institutional framework, combined with a softer type of Spanish nationalism articulated by the PSOE, was enough for most Basques to stay away from supporting the violent tactics of ETA. Theoretically, the Basque case suggests that understanding the ebbs and flows of nationalist violence requires a focus on the state. This chapter makes the argument that it is not only public policies and political strategies per se that shape such violence, but also the way the state defines the nation. Indeed, the conceptualization of a national identity by the state defines the place of minority communities within the (state) national framework through, for example, recognition or non-recognition. If a substantial segment of the minority community feels excluded from the state’s definition of the nation, the feelings of alienation that can lead to violence, even in the absence of security threats, may very well emerge. From a policy point of view, the case of the Basque Country suggests that liberal democracy, coupled with political decentralization and guarantees of minority rights, reduces popular support for political violence. In this context, the democratic Spanish state was able to pursue a vigorous policing strategy toward ETA without bolstering support for the organization. However, the policy of targeting the extended network of radical Basque nationalism, employed primarily by PP governments, likely contributed to keeping ETA support alive. The Basque case, therefore, suggests that employing coercive methods toward segments of civil society and political parties that support, or does not condemn, political violence does not contribute to eliminating that violence. Spain is not a case of optimal assurance provided to nationalist movements insofar as, contrary to Canada, for example, its leaders reject the idea that Basque independence could be achieved, even through peaceful democratic means. Rather, the Spanish political class insists on the indivisibility of Spain and the unconstitutionality of potential referendum on independence.16 Perhaps this type of stand served to perpetuate violence during the democratic period since hardline Basque nationalists did not have a reasonable expectation that they could fulfill their objective peacefully and within the existing political and constitutional framework. Moreover, the powerful projection of Spanish nationalism as embodied by the notion of the indivisibility of Spain is viewed by some Basques as a threat to their own identity. A different avenue for Spain would be to adopt the position of the Canadian government, which has acknowledged the possibility of Québec secession. The value of such a position is to make a peaceful secession more likely. In the current Spanish context, if secession ever becomes a distinct possibility following a positive referendum vote, Spain could

136 A. Lecours experience conditions of serious tensions and uncertainty, and possibly a return of the violence. The Spanish government is at this point unlikely to accept the possibility of secession and the negotiation of a new Statute of Autonomy for Catalonia seemed to open a new era for nationalist management in Spain.

Notes 1 “Mensaje de Euskadi Ta Askatasuna al Pueblo Vasco.” My translation. 2 Spain is not formally a federation but can certainly be considered a federal country since regional autonomy is constitutionally-protected (Watts 1999). 3 The Pact can be viewed online in English at www.euskadi.net/pakea/ pacto_i.htm. 4 The Declaration of Lizarra can be viewed online at www.euskadi.net/ pakea/indicel_c.htm. 5 “Basque Separatists Announce Ceasefire,” September 17, 1998, www.ict.org.il/ spotlight/det.cfm?id=162. 6 By 1996, over 75 percent of Basques said either that they rejected or were afraid of ETA, up from 57 percent in 1987 and 36 percent in 1981. The 75 percent of 1996 represents the categories “total rejection,” “before yes, now no” and “fear.” Other possible choices were: “total support” (1 percent), “critical support” (5 percent), “aims yes, means no” (11 percent), “indifferent” (2 percent), “no response” (5 percent). Interestingly, it is the “no response” category that shrank most from 1986 and 1981 where the figures were 34 percent and 23 percent, respectively. This is reflective, in all likelihood, of the greater social acceptance for being publicly critical toward ETA (Mees 2003: 99). 7 The radical Basque nationalist parties have always refused to condemn ETA, preferring instead to “explain” its actions. In this context, and since the ascendancy of ETA over these parties is well-known, electoral support for the political voice of radical Basque nationalism gives a sense of the popular acceptance of ETA. 8 Batasuna was the successor to HB and EH. 9 Autonomous Communities have also come to be active in other spheres, such as education, alongside the Spanish government. 10 This was the case of former Prime Minister Aznar in his inaugural address at Georgetown University in September 2004. Aznar argued that Spain’s problem with Islamic terrorism did not begin with the PP government’s support for the American military intervention in Iraq, but rather with the Reconquista described as “a long battle to recover its [Spain’s] identity.” See “Seven Theses on Today’s Terrorism,” September 21, 2004, p. 3. http://data.georgetown.edu/ president/aznar/inauguraladdress.html. 11 This is slowly changing. For example, the Spanish government produced the proposed European constitution in Catalan, Basque and Galician. This has sparked protest from the Valencian PP, which argues that Valencian is different from Catalan. 12 This was the Ley de Calidad de Educación. 13 The PP’s intense focus on ETA has led to some bizarre statements. For example, several months after the Madrid attacks and with all evidence pointing toward Islamic fundamentalists, former Interior Minister Ángel Acebes still maintained the possibility of some ETA involvement. See “Acebes insiste en que aún planea ‘la sombra de ETA’ sobre el 11-M,” El Correo, July 29, 2004, 1.

Violence as politics 137 14 www.transnational.org/forum/power/2004/03.01_Madrid.html. 15 Batasuna, the formal party of radical Basque nationalism, was outlawed for these elections but the party’s leadership urged supporters to vote instead for the little known Partido comunista de la Tierras Vascas (PCTV). 16 See Murshed’s discussion of indivisibility.

8

Africa’s power sharing institutions as a response to insecurity Assurance without deterrence Donald Rothchild*

Power sharing arrangements are currently the preferred strategy for assuring weaker parties about their security and well-being after civil wars. Such institutions are reassuring to weaker parties that the political majority will not capture the state and threaten their basic interests. Because citizens are assured that their intermediaries will represent them within the ruling coalition, they have an incentive to commit to agreements. However, a dilemma emerges during implementation: if power sharing systems deter violence in the short term, they may exacerbate it in the long term. The effect can be to increase insecurity and instability over time. In this chapter, I focus upon “inclusive” decisionmaking (Rothchild and Roeder 2005). By establishing formal rules on the inclusion of all major groups in key governmental positions according to the principle of proportionality, executive power sharing institutions ensure the main actors access to decisionmaking at the highest levels. The effect of this is to promote confidence among the bargaining parties about their future roles. They view inclusion in the inner circles of state power as representing a guarantee of participation in the matters that affect them most critically. Furthermore, external mediators and observers, who often prove critically important in the adoption of a power sharing strategy, are often inclined to favor institutions that hold out the promise of minimally satisfying the expectations of all negotiators. But, does this widespread support for power sharing take full account of the dilemmas of implementing such arrangements after civil wars? Is the likely outcome of such systems a transition to the joint exercise of political power over time, or a step on the way to the dominant party’s consolidation of power? To be sure, power-sharing arrangements have proven extremely useful in facilitating a transition to majority rule, as was seen with South Africa’s move to a majority rule constitution in 1996. But can these institutions be constructed to provide for patterns of governance that will prove durable and efficient? In this chapter, I begin by discussing power sharing institutions in terms of their short- and long-term implications, examining the possible lack of fit between power sharing as an incentive to reach agreements during the

Africa’s power sharing institutions 139 negotiation phase while proving a source of conflict during the longer-term consolidation phase. In the next section, I plan to analyze the different African experiments with power sharing, looking at the details on the experiences in the 1990s after civil wars. With this information in hand, I will discuss the strategy’s limited capacity to reassure weaker parties, linking the search for increased political, economic and military security during the negotiation phase with the changed circumstances that prevailed during the consolidation phase. In particular, I note the inadequate presence of three necessary conditions – reliable information, credible commitment, and appropriate external protection – for the effective implementation of a power sharing strategy. Then, after discussing the anticipated and unanticipated limits of power sharing systems in Africa, I will point to a number of strategies such as power-dividing, integrative electoral systems, and non-ethnic federalism that hold out greater hope of protecting weak parties by checking power with power through a multiple majorities framework.

The short- and long-term implications of power sharing Reassuring weaker parties about their future participation in governance is a problem during negotiations and implementation, because the short-term motives for adopting power sharing arrangements may conflict with the long-term incentives to consolidate political power (Rothchild and Roeder 2005). During negotiations on a peace agreement, power sharing institutions are attractive to weaker parties, for they hold out the prospect of inclusion in decisionmaking activities and, therefore, the ability to secure information and protect the interests of their communal membership. “In times of crisis,” writes Ahmedou Ould-Abdallah (2000, 74), the UN secretary-general’s special envoy to Burundi in the mid-1990s, “the presence of a community’s representatives within a government acts as some reassurance to that community that its vital interests will not be ignored.” In most African countries, with the state a critical actor in terms of allocating resources and providing security and the private sector small in size and offering limited opportunities, weaker groups feel that it is critical for them to be a part of governmental deliberations at the highest level. Group leaders reason that to be shut out of the cabinet, the legislature or other decisionmaking bodies are to be excluded from the spoils of office and to be unable to protect their group against exploitation, even victimization. This urge for inclusion has led numerous political oppositions to cross the aisle and join the dominant party and, after civil war, to negotiate for a proportional role at the country’s political center. Thus power sharing arrangements respond to a weaker party’s felt need for participation in affairs of state. In situations where weaker parties have not been defeated on the battlefield and a continuance of the war holds out no prospect of military victory, their spokespersons are likely to consider

140 D. Rothchild some form of power sharing to be less costly than prolonged fighting. When dominant parties or ruling coalitions are prepared to act in an accommodative way on this issue, it becomes an incentive to weaker parties to reach agreement, because the state will uphold the status quo and dominant groups will not capture the state (see Saideman and Zahar, Chapter 1). Barbara Walter’s data shows that if a peace treaty includes power sharing guarantees, 38 percent of the combatants are more likely to sign the agreement (Walter 2002, 80). Walter adds, “rival factions appear concerned with the postwar distribution of power and do seem to demand guaranteed representation as the price for peace” (p. 80). Being unable to achieve a military victory and at the same time unprepared to accept partition or separate independence, the negotiating teams may compromise on a transitional arrangement to share power in major state institutions. They seek to allay minority uncertainty about co-existence in a common state through a cooperative arrangement, designing institutions for joint decisionmaking on the basis of some predetermined formula of group representation. Power sharing is exemplified by the twoyear transitional constitution of the Democratic Republic of the Congo (DRC), agreed upon in April 2, 2003 by government and rebel representatives of the DRC at Sun City, South Africa. At Sun City, provision was agreed upon that DRC President Joseph Kabila would retain his post and would be assisted by four vice-Presidents. These vice-Presidents would take charge of government commissions, each comprising Ministers and Deputy-Ministers. The government ministries were distributed among the parties according to a predetermined formula – seven each for the former government component, the former rebel Rassemblement Congolais pour la Démocratie (RCD-Goma), the former rebel Mouvement pour la Liberation du Congo (MLC), and the unarmed political opposition, and two each for four smaller parties (IRIN 2004a). Thus, the All-Inclusive Agreement sought to end the fighting by bringing the government representatives together with the rebel leadership in the same institutions of governance. The consequences of such arrangements remain uncertain. Although this diplomatic outcome did reduce the fighting, the arrangement nonetheless seemed very shaky. In August 2004, for example, RCD-Goma leader Azarias Ruberwa, warning of war over the murder of Congolese refugees in Burundi, suspended his party’s participation in the power sharing government (BBC News 2004). Power sharing under post-civil war circumstances essentially represents a concession by a more powerful actor to less powerful ones in an effort to gain the latter’s assent to the peace accord. In a context of prevailing distrust and uncertainty, concessions on inclusion become essential to assure weaker parties about their future; nevertheless, the effects are often to create a weak state with limited reach and capacity for effective governance. When parties sign on to a power sharing agreement, what does this compromise entail? In contrast to the Westminster model which

Africa’s power sharing institutions 141 is based on “competitive,” even adversarial relations, a power sharing regime is “coalescent” and involves rules to ensure the inclusion of the main parties in a government of national reconciliation (Lijphart 1977, 25). Third parties can encourage local actors to commit to a grand coalition of elites during the transition, but the survival of this fragile institution over the long term depends upon the negotiators’ acceptance of the rules of the game and the state’s preparedness to impose significant costs on those who would break the terms of the bargain. “[P]ower-sharing practices are likely to have conflict-mitigating effects,” warns Timothy Sisk, “only if the disputants arrive at them through a process of negotiation and reciprocity that all significant parties perceive as fair and just, given their own changing interests and needs” (Sisk 1996, 9). Current data suggests that the presence of an international mediator facilitates both state commitment to uphold the status quo and insurgent perceptions of the credibility of the agreement. Walter finds that once peace agreements have been signed, the parties are 20 percent more likely to implement the arrangement if a third party acts as a protector (Walter 2002, 83; Hartzell et al. 2001, 199). Yet, in the later phases of implementation, when the external actors disengage, an early exit can contribute to disrupting the equilibrium between deterrence and assurance. At this point, it is the local parties who must take up the slack and make the arrangement a credible one. Hence the likely durability of the rules and their ability to meet the essential needs of the majority and minority parties for security, participation, and effective governance are critical to continued cooperation. In sum, power sharing measures are a logical response to the configurations of power in contexts where the forces are deadlocked militarily and view the costs of compromise on peace as lower than the continuance of war. Power sharing is a face-saving mechanism that enables the adversaries to avoid a worse outcome. However, the short-term benefits in terms of bringing organized violence to a halt may come at a long-term cost in terms of effective governance and political uncertainty about future relations. This results in a dilemma. Power sharing arrangements may enhance the prospects of peace in the short term, providing incentives to weaker parties to sign on to agreements, while becoming a potential source of instability, ineffective governance, and intergroup conflict in the long term. And, as S. Mansoob Murshed (see Chapter 4) stresses in this volume, breaking an agreement harms “future reputation,” gravely complicating the process of further negotiations at a later date. The problematic remains unresolved: How can an energetic majority and an insecure minority coexist simultaneously within the same state during the consolidation phase after civil war?

142 D. Rothchild

Patterns of experiences with power sharing in Africa Not surprisingly, power sharing institutions have attracted considerable support in contemporary Africa. Following civil wars, African governments have found that signaling a willingness to collaborate with insurgents in national reconciliation governments is costly but acceptable in order to assure weaker parties that the state will not victimize them. In signaling a preparedness to concede the sharing of power, the dominant coalition is indicating an understanding of the security fears that weaker parties have about their future and a willingness to establish institutions to protect themselves against the possibility of a re-emergence of majority tyranny. It is implicitly recognizing the validity of Paul Collier’s (2000) finding that ethnic dominance doubles the likelihood that ethnic conflict will materialize after civil war. Provided the fears and antagonisms of civil war are superseded by the emergence of shared norms and practices on constitutionalism and moderation, power sharing agreements can set the foundation for democracy and for governmental respect for human rights. Where the best-case scenario prevails and informal practices of democracy across groups develop, as in South Africa, problems of credible commitment may ease and civility and respect for difference may become expected practice. However, moderation and civility are normally in short supply after the brutality of civil war. Groups remain polarized and deeply resentful of the treatment their soldiers and civilians have received during the war (Paris 2004). The resulting mistrust withers slowly and only after members of the dominant groups are perceived to be displaying a genuine concern for the well-being of weaker entities. Moreover, the economic scarcities and lack of opportunity that mark relations in a post-civil war context heighten conflictive relations. As Victor Azarya (1994, 91) observes, “civility in social conduct may be hard to expect in countries with acute shortages and extreme gaps between levels of aspirations and accomplishments.” Not surprisingly, therefore, contemporary African experiments with power sharing institutions display mixed outcomes. Following the signing of peace accords in my sample of recent cases – Burundi, Côte d’Ivoire, DR Congo, Liberia, Rwanda, and Sudan – several patterns are in evidence. First, a shaky co-existence is present in Burundi, DR Congo, Liberia, and Côte d’Ivoire, where, following intense mediation efforts by external parties, power sharing institutions have been agreed to (in principle, at least) by local patrons, but the implementation process has proven difficult and incomplete. The incomplete nature of these consolidations is indicated by the continuance of lawlessness and violence and the slow emergence of trust that has developed among the cartel of elites making up the ruling coalition. Burundi’s National Forces for Liberation (FNL), which continues to engage government forces in the field, still has not agreed to talks to bring the decades-long war to an end (Africa Research Bulletin 2004b,

Africa’s power sharing institutions 143 15720). And UPRONA (the Tutsi-led National Union for Progress) resisted implementing the Pretoria Agreement provisions on power sharing in July 2004, urging that the percentage of seats reserved in the National Assembly for the Tutsi be designated for the Tutsi parties only and not include Tutsi who are members of the Hutu-led parties. Unless UPRONA is assigned 40 percent of the seats, warned its spokesman, “the elections will signify the elimination of the Tutsis from power” (Ndikumana 2004; see also Agence France Presse 2004).1 In Liberia, serious divisions surfaced in 2004, not only among the factional patrons who made up the transitional cabinet but also within the main rebel group, Liberians United for Reconciliation and Democracy (LURD) (Africa Research Bulletin 2004a, 15608). Order was restored at each of these levels, but only after the UN Secretary General’s Special Advisor, Jacques Paul Klein, and the US Ambassador, John Blaney, helped shore interim President, Gyude Bryant, against opposition leaders from within his own cabinet and Bryant himself took measures to halt the crisis within the LURD leadership (Africa Research Bulletin 2004a, 15608; Africa Confidential 2004, 2). In Côte d’Ivoire, the peace agreement, negotiated by the government of Laurent Gbagbo and the northern rebel groups in Marcoussis, France, was a shaky political and military deal from its outset. This compromise arrangement split power at the top between Laurent Gbagbo, who stayed on as president, and Seydou Diarra, a northern Muslim who became Prime Minister charged with heading the government of national reconciliation. Political balance was evident in the selection of government ministers, for two ministers were appointed from Gbagbo’s Ivorian Popular Front (FPI), two from the rebel forces, two from ex-President Henri Konan Bedie’s Democratic Party (PDCI), and two from former Prime Minister Alassane Ouattara’s Rally of the Republicans (RDR). Instead of assuring various interests in the country, the peace agreement was born amidst rioting against French interests by Gbagbo’s supporters, who viewed the French diplomats as biased in favor of the northern Muslims. This sense of southern uncertainty was heightened, particularly in the army, by rebel demands that northern leaders be appointed to the important ministries of defense and the interior (BBC News 2003a; 2003b). Given the prevailing hostility and the ambitions of the Ivorian elite, it is not surprising that implementation of the Marcoussis power sharing agreement has proven to be difficult. Southern political leaders frequently criticize the Marcoussis agreement, describing it as a French-mediated sellout to terrorism, while northern leaders claim that Gbagbo reneged on private and public promises he made regarding the powers to be exercised by the Prime Minister and appointments to the cabinet. Northern resentment threatened the continuing operation of the power sharing system in March 2003, as rebel representatives failed to appear for meetings, and in July, rebel military commanders ordered their ministers to suspend further

144 D. Rothchild participation in the government. Matters reached a new level of intensity on September 23, 2003 when the rebel group, now renamed New Forces, again suspended their participation in the government to protest Gbagbo’s failure to give effective powers to the Prime Minister. In the end, it was only the key New Forces Ministers who temporarily terminated their participation in the power sharing cabinet, for the ministers of such lowerranked ministries as Handicrafts and Organization of the Informal Sector, Technical Education and Professional Training, Territorial Administration, and Youth and Civil Service stayed on in Abidjan to meet with their colleagues. Northern representatives did return to the cabinet in late December, a decision that not only reflects intense international pressure but also a determination to overcome rifts developing in New Forces ranks. Subsequently, they did leave the cabinet again, in March and October 2004. The deeper meaning of these departures lay not in differences over policies on appointments but in the polarization of perceptions and the evident lack of trust. In the second pattern, negotiations progressed to the point of a North–South agreement in Sudan but remained to be applied to the country in its entirety. There, the bargaining parties did achieve a milestone with the signing of the Nairobi Declaration on the Final Phase of Peace in the Sudan on June 5, 2004, reaffirming, among other things, the Protocol between the Government of Sudan and the Sudan People’s Liberation Movement (SPLM) on power sharing. These arrangements were reaffirmed in the Comprehensive Peace Agreement of December 31, 2004. The agreement on power sharing carefully balanced power in the national cabinet between various northern and southern interests, distributing cabinet seats to the various parties according to the following proportions: the National Congress Party (NCP), 52 percent; the Sudan People’s Liberation Movement (SPLM), 28 percent; other northern political forces, 14 percent; and other southern political forces, 6 percent (see Table 8.1). However, Sudan’s peace seems incomplete, with intense fighting still taking place in the northern enclave of Darfur (IRIN 2004b). In this case, racial antagonisms combined with misgivings on the part of Darfur’s farmers over what they perceive as a discriminatory distribution policy, led to claims for autonomy and self-determination. As Jan Pronk, the UN envoy to Sudan, observed: “Without a solution in Darfur, the north–south will not remain a sustainable peace agreement” (BBC News 2005; Kasfir 2005, 201). The third pattern involves the full breakdown of power sharing institutions. In the Rwanda experiment, the institutions were designed to promote a sharing of power among President Juvénal Habyarimana’s largely Hutu Mouvement Révolutionnaire National pour le Développement (MRND) and the Tutsi-led Rwanda Patriotic Front (RPF) as well as various moderate Opposition parties that sought to avoid extremist ethnic politics (see Table 8.1). The externally-mediated Arusha Accords distrib-

Africa’s power sharing institutions 145 uted power equally in the cabinet between Habyarimana’s MRND and the insurgent RPF, with the main Opposition parties being allocated a significant bloc of positions. The effect of this compromise was to shift the perceived balance of forces in the cabinet to the advantage of the RPF and the moderate Hutu Opposition parties. This proved highly destabilizing, for the hard-line Hutu leadership “perceived [this] negotiated outcome to be inimical to their power” (Khadiagala 2002, 469), causing Habyarimana to lash out in late 1992 against the Protocols and to call upon the militia for continued backing. Habyarimana pursued what Alan Kuperman (2000, 96) describes as a two-track strategy, attempting to co-opt the Hutu moderates while simultaneously working with the extremists to develop “a forceful option.” Perception of a changing balance of forces resulted in a sense of imminent exclusion (especially on the part of the hard-line Coalition pour la Défense de la République (CDR) from the power sharing coalition created under the 1993 Arusha Protocol). The effect heightened tensions among the Hutu extremists, contributing to their fateful decision to launch a concerted program of genocide. The power sharing formulas in the Broad-Based Transitional Government featured significantly in this terrible outcome, because “it pushed well beyond what was acceptable in key sectors in Kigali” (Jones 2001a, 95). The data in Table 8.1 on recent African experiences with executive power sharing institutions indicates that a preparedness on the part of both stronger and weaker parties to commit to agreements is more likely when power sharing institutions are included than when they are not. Power sharing is certainly not everyone’s preferred solution, but it is a readily available one that may be mutually acceptable to all the main negotiating parties. Matthew Hoddie and Caroline Hartzell (2005, 103) appear to be right that, “Power-sharing provisions in peace settlements have a demonstrated ability to provide a sense of security to former combatants facing the immediate prospect of working together peacefully after a severe conflict such as a civil war.” When the terms of power sharing were carefully negotiated by the local parties over time, as in the Pattern 1 case of Burundi 2003 and the Pattern 2 case of Sudan 2004, the possibility of a stable transition at the national level seems more likely. By contrast, where power sharing institutions were not implemented, as in Rwanda, the prospects for instability and civil war – and in this case, genocide – were greater. All of the cases show that it is important not to obscure the messiness of the situation on the ground or the continuance of feelings of suspicion or insecurity. Hence, the strategy of returning to peace by means of powersharing arrangements often comes into conflict with the uncertainty of implementation, and reflects the fragility of power sharing initiatives over the long term. The implementation of peace accords after civil wars encounters multiple and complex challenges. Power sharing institutions

Executive/cabinet

Twenty-five member Power Sharing Cabinet: Presidential majority Frodebu (predominantly Hutu) – 55 percent Opposition minority – 45 percent. After 18 months in office, President Pierre Buyoya (Tutsi) will be replaced by a Hutu president for 18 months until elections. Transitional cabinet: G-7 (Hutu parties in the government) –60 percent for the G-10 (Tutsi parties) – 40 percent

Per Arusha, on April 30, 2003, Hutu Domintien Ndayizeye succeeded Pierre Buyoya as President.

Country/accord

Burundi Convention of Government 10 September 94

Arusha Process and Reconciliation Agreement 28 August 2000

Pretoria Protocol on the Arusha Peace Process and the Global Cease-fire Accord 8 October 2003, 16 November

Senate: New body includes former presidents, dignitaries, Twa, and equal numbers of Hutus and Tutsis. Transitional National Assembly: 60 percent Hutu and 40 percent Tutsi. CNDD–FDD – 15 deputies (includes two 2nd Vice-

Assembly: 60 percent Hutu and 40 percent Tutsi regardless of party affiliation

Composition of National Assembly is not affected by power sharing but powers are reduced. Frodebu retained its majority.

Transitional legislature

Table 8.1 Recent post-civil war power sharing peace agreements in Africa

Regular troops: 50 percent Hutu – 50 percent Tutsi, regardless of their political affiliation.

Security forces

CNDD–FDD – three provincial posts, five counselor posts, two ambassador posts, and 30 municipal mayorships.

Other

Rwanda Arusha Accords 25-July-93

Liberia Accra Comprehensive Peace Agreement 18 August 2003

2003, and August 6, 2004

Ministers: Government of Liberia, LURD (Liberians United for Reconciliation and Democracy), MODEL (Movement for Democracy in Liberia) – five each. Political Parties and Civil Society – six. MRND (National Revolutionary Movement for Development) – five ministries plus presidency. Opposition Parties: MDR (Democratic Republican Movement), PL (Liberal Party), PSD (Democratic and

Twenty-seven-member transitional executive cabinet: 60 percent–40 percent ethnic split (Hutu-Tutsi), regardless of political affiliation. CNDD–FDD [Rebels] (Conseil National pour la Défence de la Démocratie-Forces pour la Défense de la Démocratie) – four ministries.

Power of the legislature and the presidency was greatly reduced in favor of the power sharing cabinet.

Seventy-six-Seat Transitional Legislature: Government of Liberia – 12. LURD – 12. MODEL – 12. Political Parties – 18. Civil Society – seven. County Representatives – 15

Presidencies and Deputy Secretary) and the assistant General Secretariat. All other parties maintain 1993 election seats: Frodebu (Front pour la Démocratie au Burundi) – 65 seats, while Uprona (Unité pour le Progrés National) – 16 seats. CNDD–FDD will hold the direction of 20 percent of the public enterprises.

New armed forces: Rwandan Armed Forces – 60 percent. Rwandan Patriotic Front – 40 percent. Officer level (down to battalion commanders): 50 percent (FAR)/50 percent (RPF) split was proposed.

Police force and intelligence: Transitional Government-65 percent CNDD–FDD – 35 percent, based on 50 percent– 50 percent Hutu–Tutsi ethnic balance. Forces to disengage. New forces created with respect to “national balance,” educational, professional, medical, and fitness qualifications as well as human rights abuse history.

Military staff/senior officers: CNDD–FDD – 40 percent Transitional Government (FAB) – 60 percent Implementation of an integrated Headquarter.

DR Congo Pretoria Agreement – Global and Inclusive Agreement on Transition 17 December 2002

Country/accord

Table 8.1 continued

Cabinet: Government of DR Congo, RCD, OP, MLC – seven ministries and four deputy ministries each, Civil Society – two ministries and three deputy ministries, RCD– Mouvement de libération, RCD–National, and Mayi-Mayi militias – two ministries and two deputy ministries each.

Socialist Party), and PDCCentrist Democratic Party – nine ministries plus Prime Minister. Rwanda Patriotic Front (RPF) – five ministries plus Vice Prime Minister. Four Vice-Presidencies: PPRD (Parti pour la réconciliation et le développement), OP, MLC (Mouvement pour la libération du Congo), RCD (Rassemblement congolais pour la démocratie).

Executive/cabinet

Transitional National Assembly/Senate: Government, OP, MLC, RCD, Civil Society – 94 Deputies/22 Senators each, RCD–ML – 15 Deputies/ four Senators, RCD–N – five Deputies/two Senators, Mayi-Mayi militias – ten Deputies/four Senators.

Olivier Kamitatu (MLC) heads the National Assembly and Marini Bodho (civil society) leads the Senate.

Transitional legislature

Incorporating elements from former rebel groups and Mayi-Mayi militia, in 2003 Kabila appointed the Chiefs of Staff of the Armed Forces, as well as the commanders and deputy commanders of each of the 11 military regions in the DRC. At least three RCD commanders and deputy commanders are among those named.

Security forces

Governor post in the Katanga province is appointed to the PPRD (a post that was to be held by the Mayi-Mayi).

Other

Côte d’Ivoire Linas-Marcoussis Agreement 24 January 2003

Sudan Machakos Protocol and the Naivasha Protocol on Power Sharing 20 July 2002 and 26 May 2004 National Legislature: NCP (National Congress Party) – 52 percent, SPLM – 28 percent, Other Northern political forces – 14 percent, Other Southern political forces – 6 percent

Cabinet: National Government of Southern Congress Party (NCP) – Sudan (GOSS): SPLM – 52 percent, Sudan People’s 70 percent of GOSS posts Liberation Movement and the interim Southern (SPLM) – 28 percent, legislative assembly. Other Northern political forces – 14 percent, Other Southern political forces – 6 percent. Government-FPI (Front Populaire Ivoirien) – nine ministries, Opposition – PDCI (Parti Democratique en Cote d’Ivoire) – seven ministries. Opposition – RDR (Rassamblement des Républicains) – seven ministries. Rebels – MPCI (Mouvement Patriotique de Cote d’Ivoire) – seven seats (including defense and interior ministries, hotly disputed by Ivorian government and armed forces).

Vice President/President of GOSS/Commander-inChief of SPLA: SPLM Chairman.

President/Commander in Chief: current incumbent

National Civil Service: 20 to 30 percent of positions reserved for Southerners.

Joint/Integrated Units: equal number of Sudanese Armed Forces (SAF) and of Sudan People’s Liberation Army (SPLA) during the Interim Period.

The transitional government was to propose a naturalization law, within six months of the agreement with a straightforward application process. As this did not occur, it remains a crucial point of contention between the rebels and the government of the Côte d’Ivoire.

Oil Revenue: 50 percent of oil revenue derived from oil produced in Southern Sudan shall be allocated to the GOSS and 50 percent to the National Government and the North.

Federalism: Autonomy for the south with a promised referendum on self-determination in six years.

150 D. Rothchild appear to be logical responses to the need for ethnic self-determination and representativeness; however, as I indicate in the next section, these institutions can involve a possible price in terms of political stability, effective governance, and economic development.

The limits of reassurance As Table 8.1 indicates, power sharing institutions are up and running in contemporary Africa. Few people, scholars or practitioners would contend that these are ideal institutions but rather represent pragmatic adjustments to difficult circumstances in deeply-divided societies (Lijphart 1985, 133; O’Leary 2001, 81–82).2 Certainly, in the best of circumstances, powersharing institutions offer several positive potential benefits. First, they can assist the parties to end the fighting, or, as in Burundi and the DRC, to significantly reduce the violence. By bringing the leaders of the various warring factions into the ruling cartel, they help to make them stakeholders in the political process. This enables leaders of weaker parties to gain an advantage by working through the system to achieve their objectives instead of rebelling against it. Second, the proximity to other patrons opens up new possibilities for interaction among political parties, as well as among former adversaries. As the government of national reconciliation meets to engage in common problem solving activities, it fosters reciprocities and the development of bargaining norms which can have a stabilizing effect. By promoting an iterated exchange process, power sharing institutions can prove reassuring, laying the basis for an ongoing relationship. Over time, these iterated exchanges can create networks of trust among separate political actors, which are the basis for cross-cutting perspectives on specific issues. The effect of this can be to encourage self-sustaining encounters. Power sharing, like democracy, does not end uncertainty, but it can create institutions and facilitate relationships. These institutions may hold up during the transition because the alternatives are so frightening (Przeworski 1991, 26, 32). Despite these possible benefits, power sharing institutions awaken skepticism regarding their ability to survive in the face of intense conflicts or to protect the interests of weaker groups. In the uncertain conditions after a civil war, groups have negative political memories, whether real or unreal, of their opponents, and they project these images into the future. Such memories and projections lead to fears of imminent exploitation or to uncertainty over the inability of the state to deter violent actions on the part of tenacious foes searching for the first opportunity to take advantage of their vulnerability. Hence the task of transforming the interactions among groups toward closer, more predictable relations is a sensitive and difficult process. In many instances, the uncertainty of a post-civil war context seems likely to persist into the peace that follows. Because power sharing rests on

Africa’s power sharing institutions 151 the maintenance of a balance of forces, it may contribute little to political consolidation. In attempting to explain the limited ability of power sharing arrangements to reassure weaker parties, I will begin by analyzing the insufficiency of three necessary conditions – reliable information, credible commitment, and appropriate external protection – for the effective implementation of a power sharing strategy. After that, I will turn to some specific aspects drawn from the African experience. Unreliable information Following a civil war, communal polarization remains extreme and crosscutting linkages are circumscribed. Although political elites meet formally for prescribed purposes, groups live for the most part within recognized and separate spheres. With contact essentially confined to the group, reliable public information is scant, contributing to widespread uncertainty over the intentions of former combatants. Suspicion over the good will of rivals who otherwise dwell in separate, watertight compartments often emerges and complicates negotiations within power sharing institutions, as leaders fear that their opponents are not bargaining sincerely and will renege on an agreement when it becomes advantageous to do so. This problem is further complicated when parties deliberately misrepresent the facts in an effort to minimize these weaknesses or to maximize their strengths. “Leaders know things about their military capabilities and willingness to fight that other states do not know,” writes James Fearon (1995, 381) “and in bargaining situations they can have incentives to misrepresent such private information in order to gain a better deal.” Either way, lack of reliable information or the presence of misinformation contributes significantly to bargaining failure (Lake and Rothchild 1998, 12). Such failure undermines political stability and undercuts the assurances that emerged from the peace agreement, possibly contributing to new spirals of violence. The lack of credible commitment If peace agreements are to be viewed as the equivalent of binding contracts between former belligerents, they require trust on all sides that their terms will be upheld. This means that each party must believe that the other party will honor the bargain. Rival leaders will be expected to persuade their ethnic or religious supporters to maintain their commitments, even in the face of militant outbidding politicians who seek to rally support for a less accommodating course. In theory, all parties will gain by maintaining the peace and abiding by the agreement; in practice, however, majority party leaders cannot credibly commit that they will abide by the bargain. Despite reassuring gestures regarding future intentions, Sudanese government leaders cannot guarantee that upcoming leaders will actually hold the

152 D. Rothchild promised referendum on southern self-determination six years after the 2005 peace agreement was reached. Later, after the confidence-building period has elapsed and the agreement has begun to structure stable relations, the urgency of maintaining the compromises agreed to during the negotiations may fade and majority elites may no longer feel the need to reassure weaker parties about their security. They, therefore, come to place a higher priority on achieving their programs and fulfilling the interests of their constituents, and in the process pay less heed to implementing the peace bargain. With little in the way of accepted norms and values, or an established high court system to prevent them from reneging on the agreement, the ruling coalition may decide to alter the accommodations originally put in place. This inability to commit in a credible way underlines the fragility of compromises on power sharing institutions. Power sharing arrangements, which carefully balance competing group interests in order to reassure the weaker parties with respect to their inclusion in the decisionmaking process, may be seriously impaired by the decision of the dominant majority to alter the rules of the game in favor of a more integrated approach. Because the power imbalance facilitates majority party leaders in their determination to capture the state, there is little the opposition can do on its own to defend the agreement. The rules have failed to protect, contributing to the opposition’s frustration, even to its sense of betrayal. Paradoxically, then, the very provisions in the agreement intended to promote the security of weaker parties may become a cause of insecurity and lead to intense intra-state conflict. Inappropriate external protection Because it is difficult for ex-belligerents to overcome the information and credible commitment dilemmas on their own, they often rely on external actors to play an important role in maintaining their peace agreements. External actors can help to keep open the lines of communication between the parties and mediate issues left unresolved in the agreement. They can also reassure the parties that the other side is delivering on its bargains, monitor the implementation of the agreement, and provide desperately needed relief and development assistance. That the third party interveners are critically important in building support for the implementation process is suggested by recently compiled aggregate data that indicate that the presence of a third-party state increases the chances of maintaining the peace during the consolidation phase by an estimated 98 percent (Hartzell et al. 2001, 199). When the external actor intervenes and is effective in protecting the peace agreement, the parties may be reassured by its involvement. However, as the third party fails to assume its role fully and as it seeks to disengage from the risks and responsibilities of protecting the agreement, the weaker parties are likely to lose confidence in the dominant

Africa’s power sharing institutions 153 majority’s willingness to remain committed to its earlier accommodations on power sharing. The UN, which had been asked by both sides to assist in implementing Rwanda’s 1993 peace agreement and to support the transitional government, failed to deploy sufficient forces to the area; as a consequence, the transitional government was never inaugurated, contributing to a loss of confidence in the outside protectors’ ability to deter spoilers that still lingers (UNDPKO 2001). Thus far, I have been assuming that the third party plays the role of a reasonably impartial protector of the agreement. But what happens when the third party is biased in favor of one participant in the power sharing coalition as against the others? In different periods in Lebanon, French authority was partial toward the Maronites during the Mandate period and the Syrian administration favored the Muslims after the Ta’if agreement was negotiated. In these cases, the outsider propped up the power sharing institutions for an extended period of time, but this external influence merely masked the deep intergroup resentments that divided these interests from within. Clearly, where the survival of power sharing institutions is dependent on a foreign power as protector, the arrangement is likely to be lacking in legitimacy and possibly conflict creating. As MarieJoëlle Zahar explains, the presence of the third-party protector creates winners and losers (Zahar 2005c, 226–227). Once the third party begins to disengage, it weakens its hold, opening the way to a restructuring of relations. As internal relations shift, the dominant majority may seek to centralize power, or the weaker parties may attempt to decentralize it. Either way, as Yeats observed, the political center cannot hold. In addition to these general difficulties with implementing powersharing institutions after civil war, several significant aspects drawn from the African experience are worth noting. First, groups may fear a weakening of power as a consequence of participation in a national grand coalition. It is assumed that participation in a grand coalition has a constructive spillover effect, pushing the parties in the direction of cooperation. The political moderation that follows from this, however, may actually weaken their autonomous power base. In Liberia, for example, the leaders of two former rebel movements – Liberians United for Reconciliation and Democracy (LURD) and the Movement for Democracy and Elections in Liberia (Model) – who were members of the transitional cabinet reportedly tried for a time to undermine it. “This is partly because LURD leader Sekou Damateh Conneh and Model leader Thomas Ninely Yaya want[ed] to keep their militias intact should their services be required again by their respective sponsors in Guinea and Côte d’Ivoire” (Africa Confidential 2004, 2). They feared that if the militias were dismantled their power bases would be gravely weakened. The effect was to place a strain on the institutions of power sharing during the initial confidence-building period.3 Second, because power sharing results from a bargain among the main parties, it can make no claim to being a fully inclusive arrangement. This

154 D. Rothchild incompleteness creates potential problems in its wake. Those parties left out of the decisionmaking process may feel disadvantaged and possibly vulnerable. At the 2002 Inter-Congolese Dialogue in Sun City, South Africa, the government of Joseph Kabila and Jean-Pierre Bemba’s MLC negotiated an agreement, but they excluded Etienne Tshisekedi’s Union pour la démocratie et le progrés social and the important rebel group, the RCD-Goma, led by Azarias Ruberwa (ICG 2002, i). Moreover, as one report on the Liberian peace process noted, exclusion from the powersharing coalition can give “incentives for warlords to ‘spoil’ the [peace] process . . . and for current groups to further factionalize” (United States Institute of Peace 1996, Section IV). Third, African experiments with power sharing have been handicapped by a lack of shared norms and aspirations. The polarized and hostile perceptions that contributed to civil war are not transformed by powersharing arrangements after the peace. A local observer, noting the lack of confidence prevailing in the DRC, stated pointedly that “leaders themselves do not have trust in each other; as the power sharing is conceived in [a] zero-sum game perspective: one wins, not with the others, but at the expense of the others. The necessary atmosphere of trust is lacking” (Dougherty 2004). These arrangements preserve the separate identities of the former belligerents and include them in the grand coalition at the political center, which may bring limited compromises but not decisive moves to solve common problems. Without common norms and aspirations, it becomes difficult to maintain a balance of forces, especially where uncompromising leaders, including warlords, enter the cabinet. If the Sudanese negotiators had common aspirations, an IRIN report concludes, the negotiating parties “would have [achieved] a done deal months ago. It is their lack of common aspirations and conflicting political agendas that have meant they have to fight tooth and nail for every gain they make or loss they concede”(IRIN 2004c). Not surprisingly, these political leaders carry conflicting political agendas over into the grand coalitions that are cobbled together to consolidate the peace. Fourth, experiments with African power sharing are constrained by the painfully difficult economic situation that follows a civil war. The immediate costs of financing demobilization, disarmament, the return of former combatants to their villages and to socially useful occupations, the reintegration of armies, and the reform of the police are all very expensive undertakings. Moreover, the task of getting the economy moving again is an urgent matter that requires substantial investments. As Collier (2000, 3) warns, “the more severe the poverty, the greater are the risks of conflict.” Third party promises of economic assistance if the belligerents will sign a peace accord hold out an attractive prospect of a peace dividend. However, recent work indicates that in a majority of African cases there was evidence that the end of civil wars was followed by significant rises in relief but not in US development assistance (Rothchild and Emmanuel

Africa’s power sharing institutions 155 2004). The impact of this disappointing performance is to place limits on the possibilities for economic expansion and, with it, the ability of members of the ruling coalition to channel resources to their clients. As a consequence, the constrained economic circumstances that follow a civil war may give the members of the power sharing coalition little incentive to act with civility toward members of marginal groups. The pie remains small. This results, all too often, both in a wasteful collusion among members of the power sharing cartel over the distribution of resources and in an intense competition between them and others for control of the meager resources available to the state.

Conclusion Can one reasonably expect power sharing arrangements to reassure weaker parties in Africa about their futures after civil wars? In theory, such institutions are logical responses to the need for ethnic self-determination and fair representation. They can, as the experiences of Burundi, Côte d’Ivoire, Rwanda, Liberia, Sudan, and the DRC indicate, help to provide a basis for ending the fighting – at least in the short term. This occurs because they propose a mixed package of incentives to the combatants, offering the stronger parties a means to bring an end to the war while holding out the possibility of inclusion in the government to the weaker actors. Moreover, agreement during the negotiations on power sharing creates the possibility that the parties may be able to use this arrangement as an initiating point for an iterative bargaining process that can lead to new institutions during the political consolidation process. The ruling coalition’s concessions on power sharing were acceptable to the various weaker party elements in at least half the cases analyzed in this chapter. In particular, the concession on inclusion of group leaders in the governing institutions at the political center goes far toward satisfying this elite’s desire for political power and its members’ need for security and access to scarce resources. However, the political center cannot be expected to remain stable. If power sharing arrangements were reassuring to weaker parties during the confidence-building phase of implementation, they all-too-often become the source of suspicion and intense rivalry in the later political consolidation phase as the dominant majority seeks to maximize its interests and feels less and less inhibited by the need to allay the uncertainties of weaker actors. In the later period, new elements assume positions of authority who may not feel bound by the promises made by their predecessors at the bargaining table. In addition, the presence of a third-party protector during the initial phase becomes less and less of a restraint after the general elections and the passage of time. At this point the general constraints of credible commitment, unreliable information, and inappropriate external protection come into play. Also, as seen with the African

156 D. Rothchild experiences, such factors as weakening group power as a consequence of participation in the grand coalition, lack of full party inclusion, an insufficiency of shared norms and aspirations, and the difficult post-conflict economic environment all threaten the precarious balance necessary to maintain a power sharing approach. If power sharing systems seem rigid and carry with them a high potential for disruptive behavior over the long term, it is important to recognize this and to think about possible alternatives. Power sharing among a number of contentious groups can weaken the ability to develop a successful assurance strategy in Africa. Its efforts to balance ethnic and religious elite interests are inevitably risky and can be a source of systemic breakdown. Because such a balance seems precarious, it becomes essential to ask: What institutions can be put in place that will respond in a more flexible manner to the need to safeguard minority interests while providing for majoritarian participation? “[E]very government,” commented Alexander Hamilton (N.d., 384) in 1788, “ought to contain in itself the means of its own preservation” (italics deleted). In certain circumstances, external actors can pressure and facilitate, but ultimately a country’s institutions must build a consensus that supports an enduring legitimacy. If elite power sharing regimes prove risky because they build upon and maintain separate identity loyalties, long-term political stability would seem to require communal integration, not division. Certainly it seems unwise to assume that alternative majoritarian institutions will necessarily provide effective assurance. The dangers that the majority will capture the state and act in a tyrannical manner are all too apparent. Yet the possibility of political legitimacy that these majoritarian institutions hold out makes it imperative to search for options that will combine majority participation and minority protection. How, in the words of Aysegul Aydin and Scott Gates (see Chapter 5) can one create multiple agents in the same system that have divergent preferences and act as a check on each others’ ambitions? Assuming that the diverse communal groups that make up the country want to stay together and that partition is neither necessary nor desirable, the following alternatives are put forward briefly to illustrate the broad scope of integrative institutions available to statesman designing constitutions after intra-state wars (Rothchild and Roeder 2005, 1–82). Separation of powers By dividing political power among majority-elected branches of government, a separation of powers system encourages multiple majorities and built-in checks and balances on each of the branches. These systems combine individual liberty and participation in the electoral process with limits on the concentration of power in executive or legislative hands. As Philip Roeder puts it, “rather than empowering a single majority in a supreme organ such as the Westminster Parliament, power-dividing insti-

Africa’s power sharing institutions 157 tutions empower multiple majorities each construing the public interest somewhat differently, in separate, independent organs of government” (2005, 52). With power checking power, multiple majority systems encourage the protection of minorities by opening up new possibilities for factional competition within majority groups as well as creating possibilities for alliances between majority and minority factions. By linking majority and minority factions, such alliances may reduce, but not eliminate, the information and credible commitment dynamics. Even so, the overall effects can be to focus on cross-cutting issues rather than the protection of separate ethnic interests. The multiple majority system thus reassures minorities by opening up possibilities for cross-cutting factional interests while promoting effective governance by avoiding the implementation of rigid formulas on the recruitment of leaders into high government positions. Bicameral legislature For similar reasons, a constitutional design providing for a two-house legislature in which the two houses are selected on different criteria and represent different sets of interests further multiplies the points of power. Consequently, it is potentially reassuring to minorities. Liberty, the authors of The Federalist wrote, is threatened by the accumulation of power in any one branch of government. The answer to these centralizing tendencies is to divide power among them, and, then, if possible, to further divide it by establishing competing agencies within the branches. “[A] senate, as a second branch of the legislative assembly, distinct from, and dividing the power with, a first,” wrote Hamilton or Madison (N.d., 403), “must be in all cases a salutary check on the government.” Provided these different branches are elected by different majoritarian formulas and have well developed party systems, a structure is in place which encourages the legislators to adopt a different and competing view of their interests (Moser and Scheiner 2004, 576). In South Africa, for example, the 400 members of the National Assembly are elected under a proportional representation system to serve five-year terms; meanwhile, the National Council of Provinces, consisting of 90 members (ten elected by each of the nine provincial legislatures for five-year terms) is given special powers to protect subregional interests, including the safeguarding of the cultural and linguistic traditions of ethnic minorities. Such a structure can prove an important protection of minority security and well-being. An independent judiciary Similarly, an independent judicial branch with autonomous powers of interpretation, and possibly judicial review, can add another point of

158 D. Rothchild power, protecting the general public against arbitrary acts by the legislative or executive branches. The very presence of an autonomous judicial branch acts as a potential restraint on the passage of measures that will repress minority liberty and the equal protection of the laws. In March 2002, for example, Nigeria’s Supreme Court found that the recently enacted Electoral Act was unconstitutional on the grounds that the National Assembly had exceeded its constitutional powers by altering the tenure of elected officers of the local government councils (Okorie 2006). A multiple-majorities system depends, in the final analysis, upon active public participation in selecting its representatives. This is safeguarded by an independent judiciary, which protects the public’s right to engage on an equal basis in these different political processes, as well as to retain a capacity to form associations for the advancement of their legitimate interests. An integrative electoral system As indicated above, a majoritarian electoral process is critical if a multiple majorities system is to prove an effective check upon arbitrariness. What is needed is to design an electoral system that produces broad ethnic and religious representation in public offices without specifying special protections for any group. Such special protections can become a point of future contention, as electoral outcomes come to reflect the perceived advantages or disadvantages that particular groups have received. Such potential conflict weakens the norms of reciprocity and political exchanges among possible winners and losers. Ultimately, should outbidding behavior occur, this weakening of integrating norms can undermine the legitimacy of the electoral system. Hence, the question becomes: What electoral institutions can be expected to promote broadly representative and integrative outcomes simultaneously? The classic debate in political science on the relative advantages of proportional representation (PR) and plurality systems throws only a limited light on the issue of inclusiveness. Depending on the way that single member districts are defined, they can be structured to be broadly inclusive or narrowly representative of identity interests; depending on whether the least or most proportional of PR formulas is adopted, the outcomes are likely to vary in terms of effective communal representation and party dominance (Mozaffar 1998, 90). PR appears to offer certain advantages in achieving integrative outcomes: where the communities are interspersed, the parties have an incentive to set up balanced lists from all major communities to further their chances of winning at the polls in multimember districts. In terms of the goal of inclusiveness, however, Ben Reilly’s (2005) findings give us reason to be cautious about assuming that greater linkage exists between PR and ethnically and religiously inclusive executive coalitions when compared to plurality system.

Africa’s power sharing institutions 159 Rather, on the basis of a comparison of nine divided societies in the total sample of 36 established democracies, he finds, somewhat counter-intuitively, that five use majoritarian, not proportional, electoral systems (Reilly 2005, 166). “[A]mong the world’s ethnically divided countries that have a long duration of unbroken democracy,” he concludes, “there is a compelling relationship with patterns of executive inclusiveness, but not with the adoption of PR electoral systems” (p. 167). PR might be preferred on other grounds (representativeness, restraints on outbidding, and lower levels of ethnic conflict – Saideman et al. 2002, 122), but the case that it is most likely to facilitate ethnic inclusiveness remains to be proven. Perhaps more convincing in terms of integrative objectives is the structuring of presidential elections to ensure that the winner will have the backing of a broad-based national constituency. It is hoped that requiring wide communal and sub-regional participation in the selection of a president will encourage potential candidates to make moderate appeals in order to attract majority support. The possibilities for using electoral design to make moderate behavior rewarding are illustrated by Nigeria’s experience since 1979 (Horowitz 1991). For example, Nigeria’s 1999 Constitution provided that a candidate for the presidency would be deemed elected when there are only two candidates and one of these has secured a majority of the votes cast and has not less than one-quarter of the votes cast in two-thirds of the states. If there are more than two candidates, the winning candidate must receive a plurality of the votes cast, as well as one-quarter of the votes cast in each of two-thirds of the states. If this proves insufficient, one (and possibly two) run-off elections will be held between the two highest candidates, and, if necessary, the outcome will be determined by a majoritarian decision-rule (Federal Republic of Nigeria 1999, 134 (1)–(4)). In the past, experience with this broad-based electoral design indicates that this provision has provided an incentive for moderation. Non-ethnic federalism Non-ethnic federalism exists when a federal system is in place in which there is significant sub-regional jurisdiction along economic and social lines, not based on the ethnic principle but in terms of functional interests. Rather than protecting ethnic groups by obstructing central government initiative (i.e. tensional federalism), non-ethnic federalism emphasizes the decentralization of limited political power to ethnically heterogeneous subregions, local authorities, and civil society. In doing so, it seeks to achieve efficiency, diversity, and scope for experimentation (i.e. cooperative federalism). South Africa’s policy of “cooperative government,” which decentralizes limited responsibilities on local matters to provincial and local authorities, is one example of this approach. By allowing these elites some limited discretion to adapt local rules to the particular social and cultural

160 D. Rothchild needs of their constituency, non-ethnic federalism can ease local fears of uniform treatment. In sum, if power sharing arrangements prove rigid and potentially conflict creating, alternative multiple majorities measures can be utilized which offer various possible means of assuring groups that they have access and that the state will not be used against them. Rather than attempting to thwart majority power by legal and political restrictions on effective action, something that often seems a futile undertaking, multiplemajority measures seek to protect minorities by spreading power and enabling strong and weak ethnic groups to form political coalitions with like-minded interests across the ethnic divide. Currently, state and external guarantees have shown only a limited capacity to resolve the dilemma of assurance and deterrence over the long term. However, by adopting a multiple-majorities approach at the state level and a more assertive stance at the international level, progress power can check power, raising hopes that the challenge of protecting ethnic minorities can be resolved.

Notes * I wish to thank Nikolas Emmanuel, Camille Sumner, and Eileen Ortiga for their help on the research and Stefan Wolff, Daniel Posner, Steven Saideman, and Marie-Jöelle Zahar for their useful suggestions on revising the manuscript. 1 Statement by M. Manwangari. 2 Arend Lijphart, a leading advocate of the consociational democracy approach, states, for example, that “it is nobody’s preferred ideal.” 3 In November 2004, it was announced that both LURD and Model were disbanded (see Paye-Layleh 2004).

9

The “chicken or the egg”? External support and rebellion in ethnopolitics Yasemin Akbaba, Patrick James, and Zeynep Taydas

Introduction Why do some ethnic conflicts attract external support from states for one or both adversaries, even on multiple occasions, while others do not? Does rebellion trigger external support or does the causal mechanism run in the other direction? How does external support affect, and in turn complicate, the state’s assurance-deterrence dilemma? The issues of internal ethnic war and state intervention are becoming increasingly salient as we move from a system of inviolable boundaries and multilateral involvement to one of failed states and regional interventions. The roles played by the behavior of host states and minorities, separatist movements, ethnic affinity, and cleavage come to the fore in considering cause and effect as related to rebellion and foreign assistance. Analysis of external involvement into ethnic conflicts can be divided broadly into two categories: studies that focus either on (1) the characteristics of the intervener; or (2) the attributes of host state and minority.1 Studies in the first group focus on motivations and attributes of the intervening state. Collectively speaking, this literature implies that, before intervening, a government will weigh all possible considerations – emotional attachments, benefits, costs, and risks (Regan 2000). In this process, affective and instrumental features both are present. With regard to affect, ethnic ties and affinities play an important role, along with instrumental motives that include strategic, economic, and political calculations (Carment et al. 2006). Characteristics of minorities and state behavior toward them are subjects of the second group of studies. Our work follows this tradition. We try to understand the role of fear and insecurity triggered by state discrimination in eliciting external support, which also entails, as it turns out, a foray into the reasons behind rebellion. States set limits on what groups can obtain, with the degree and form of response from the disadvantaged depending upon the openness of the political system (Gurr 2000: 81). Gurr (2000: 105) asserts that discrimination against certain groups provides them with incentives for ethnopolitical action because of resulting poverty,

162 Y. Akbaba et al. powerlessness, and resentment. What makes discrimination a critical concept is the insecurity and fear it spreads for minorities. It also can be seen as a mechanism for group mobilization by elites. Emphasis on ethnic identities through discrimination creates a security dilemma between and among groups (Posen 1993). There is more to the matter, however, than just an application of an analogy from International Relations. As noted in Chapter 1, the state plays a critical role in mediating or controlling ethnic group security (Saideman 1998). The balance between assurance of security for each group and deterrence of one or more groups disposed toward violence or even rebellion is a delicate balancing act that is in play even before realization of open ethnic conflict (Saideman and Zahar, see Chapter 1). In an ideal situation of trust, the state monopolizes the means of violence and minorities do not compete for security. However, if the state does not protect groups sufficiently, they might compete with each other in the quest for security, with rebellion as one possible result. Attributes of the host state, most notably, security dilemmas created through levels of economic, political and religious discrimination and their effects on international support for a minority group, are of great interest. We seek to extend Posen’s concept of an ethnic security dilemma and combine it with elements of the Minorities at Risk (MAR) model, which includes discrimination as one of the factors that determine the emergence of protest and rebellion by ethnic groups. We add a step to the MAR model that ties discrimination to internationalization of ethnic conflict – i.e. external support. In the MAR-based literature, the road to rebellion starts with discrimination against ethnic groups, which eventually leads to grievances, mobilization and protest. This chapter’s underlying mission is to convey and test an analytical framework that includes meaningful connections between and among state discrimination, the security dilemma, ethnic rebellion, and external support. Thus it addresses one of the fundamental questions motivating this volume: “How do external processes and third party involvement affect the deterrence-assurance dynamics within states?” (Saideman and Zahar, see Chapter 1). In particular, does external support facilitate or dampen rebellion? External support, of course, is not exogenous. Thus, the impact of rebellion on external support also must be brought into the model created in this chapter. Taken together, the analysis includes two interdependent stages. The framework seeks to explain external support and rebellion, respectively. External support is explained by a combination of (1) behavior by the state and minority group; (2) the degree to which a separatist element is entailed by the conflict; and (3) ethnic affinity and cleavage. The same network of variables is used to explain rebellion, with the latter and external support also playing causal roles in relation to each other. This chapter proceeds in four additional parts. The first is a theoretic-

The “chicken or the egg”? 163 ally oriented discussion of the main concepts from the preceding paragraph. The second part is the research design, which includes hypotheses and measurement of variables. For this study, the MAR (Gurr 2002) and external support data sets (Saideman 2002) are used for the time interval from 1990 to 1998. Data analysis appears third. Fourth, and finally, conclusions and directions for future research are offered.

Discrimination, rebellion, and external support: the security dilemma in action Posen (1993) applies a concept from International Relations, the security dilemma, to ethnic conflict. Realism, as an international relations theory, explains that international anarchy leads to security concerns for states. Collapse of an imperial power might create similar fear for different groups in a domestic power vacuum, as observed in the breakup of Yugoslavia and the former Soviet Union (Posen 1993). If groups with different ethnic, religious, or cultural identities suddenly are forced to protect themselves due to disintegration of the state apparatus, their threat perception can shift dramatically. Under these circumstances, a neighboring group can be perceived as a threat, which increases the chance of conflict. Each ethnic group tries to increase its own security which, in turn, creates an unsettled and potentially threatening environment for the others (Posen 1993: 104). Saideman expands the security dilemma concept by applying it not only to collapsed states, but also existing states with ethnic security problems: “If the state cannot protect the interests of all ethnic groups, then each group will seek to control the state or secede so that they can control their own state, decreasing other groups’ security and decreasing the state’s ability to provide security for any group” (Saideman 1998: 135). Although Saideman argues that ethnic groups might feel insecure about many different things, his definition of security includes three aspects: physical, economic, and political. Groups cannot rely on the impartiality of the state and efforts to provide their own security in each aspect become more likely (Saideman 1998). Our approach to the security dilemma as related to external support focuses on two aspects: economic and political insecurity caused by discriminatory actions by the state.2 Discrimination is as an important source of ethnic conflict, but is not prominent in studies focusing on minority fear and insecurity. Although the security dilemma as elaborated by Saideman (1998) explicitly refers to a discriminatory state, it is not linked directly to the concept of discrimination. If groups are insecure because of the vicious cycle created by the security dilemma, they will have a tendency to act in one of three different ways: seek to control the state, create a state they can control (secession) or join a state where their ethnic group is more secure (irredentism) (Ayres and Saideman 2000: 1130). In this process, ethnic groups facing discrimination and a security dilemma can choose to

164 Y. Akbaba et al. ask for outside help – third-party intervention – and internationalize the conflict in order to realize their ultimate goals. Ethnic groups weakened by intentional and purposeful state discrimination will have a hard time fighting for their cause, whether it involves secessionism or irredentism. Therefore, to reduce the capability asymmetry between themselves and the state, they need the support of outsiders. This process, of course, could produce a causal mechanism in which ensuing external support triggers rebellion of the now more capable ethnic group. What, then, does discrimination mean in the context of behavior that creates insecurity and fear? Group discrimination is defined by MAR a deliberate act of maintaining inequalities for members of an ethnic group in terms of material well-being, political access or cultural status in comparison to others (Gurr 2000: 106). Politically active ethnic groups targeted for discrimination are of “greatest concern in international politics” (Gurr and Harff 1994: 5). In 1990, around 80 percent of the politicized ethnic groups suffered from either contemporary or historical economic and/or political discrimination (Gurr and Harff 1994). Violence and even full-scale rebellion become salient alternatives in the thinking of such groups when their plight continues for an extended period. Gurr’s (2000: 106–107) authoritative treatment emphasizes several aspects in conveying how the MAR Project comes to grips with the substantive meaning of discrimination. First of all, discrimination is not a universal act but rather is applied to certain groups selectively. Second, group discrimination is accepted as a matter of social practice and/or the result of public policy. Third, it may affect the collective gains of the group more than the well-being of individual members.3 Fourth, discrimination is usually rooted in the past. Fifth, and finally, MAR includes both advantaged and disadvantaged minorities (Gurr 2000: 106–107). MAR’s conception of discrimination points the way toward the security dilemma. Discrimination is intentional, which means it is addressed selectively to a certain group; this causes fear and collective-based responses can be expected. Group-based coping then threatens the security of other groups and provides an easy path to violence. While discrimination might be manifested along many dimensions of public policy, the two categories emphasized by MAR will be the focus of this study: political and economic. This choice of priorities is consistent with Saideman’s (1998) assessment of economic and political insecurity. Political discrimination refers to limitations imposed on people exercising their political rights or diminished access to political positions as compared to others (Gurr 2000). In 1994–95, approximately one-third of 275 minorities experienced political discrimination (Gurr 2000: 114). Examples include Albanians in Yugoslavia, Hungarians in Slovakia, Tibetans in China, Turks in Bulgaria, Palestinians in Jordan and immigrant workers in Switzerland. Economic discrimination refers to whether members of a group are, or

The “chicken or the egg”? 165 have been, intentionally limited in their access to the conditions of material well-being in comparison with other groups that have unlimited access in their own society (Gurr 1993b). Gurr (2000: 110) reports that, for the period of 1994–95, 177 groups out of 275 faced some kind of economic discrimination. For instance, Arab citizens of Israel are subject to more economic discrimination than even other groups with reduced status, such as immigrant Jews (Gurr 1993a). Although these Arab citizens have more political rights as compared to ordinary citizens of Syria, they still are discriminated against by the society in which they live. Another striking example of this kind of discrimination is Copts in Egypt. Despite being more educated (on average) than Muslims, Copts usually experience greater difficulty in finding employment. An indicator of the impact of economic discrimination is that “a number of Coptic university graduates reportedly converted to Islam to for economic reasons” (Gurr 2000: 139). Discriminatory economic systems, which are based on ethnicity or class, can increase the level of insecurity for a particular group and generate resentment, anger, and disturbance (Brown 2001). A country may have a prosperous and growing economy, but if a particular group is excluded, propensity for conflict may be high. Disadvantaged groups of people can be expected to question the system’s fairness and even its legitimacy. Economic development can enhance stability, but significant inequalities or gaps among groups in a society are capable of producing the opposite effect (Brown 2001). While the preceding discussion shows that discrimination along several dimensions can have an effect on ethnic conflict, a key gap remains in the research. Specifically, the impact of political and economic discrimination, along with other factors, on external support for ethnic groups is in need of further attention (Regan 1998: 756). Our effort to fill the gap will build on the best available model of ethnic conflict, which is put forward by MAR. The MAR model is the product of research initiated by Gurr that dates back to his classic work, Why Men Rebel (1970). The model explains ethnic conflict through intervening factors that include deprivation and political mobilization. The Project stands out as a foundation for research on stateless nations (Meho and Tibbo 2003). While Gurr’s theoretical framework is multifaceted, its key components can be summarized as three basic steps: discrimination, grievance, and mobilization (Fox 2000). In the first step, a minority suffering from political, economic, and cultural discrimination develop fears and insecurity and articulates grievances. The idea behind this step is quite simple, but also profound: people react to acts of discrimination. One of the basic reactions to insecurity is anger, which at this point might take either constructive or destructive forms. Constructive anger produces primarily peaceful actions by minorities. When destructive, anger causes a more direct challenge to sources of discrimination. In both ways, however, consciousness of grievance takes hold among the minority. For instance, contemporary grievances

166 Y. Akbaba et al. among members of the Turkish community in Germany, such as threats to their personal safety, originate from discrimination over the long term. In MAR’s second step, grievance stemming from discrimination is the main element that triggers mobilization of a minority. Overt discriminatory policies give people powerful incentives for action; targeted groups can be expected to try to improve their condition. Discrimination will amplify already existing ethnic identification in a way that motivates members to look for collective solutions. Depending on the strength of group identity, degree of ethnic group cohesion, type of political environment, severity of government violence, extent of external support, and international status of the regime, a minority can be expected to develop strategies for political action (Harff and Gurr 2004). This is called mobilization and more than 200 of 233 groups experiencing discrimination became organized politically between 1945 and 1989 to pursue their collective interests (Gurr and Harff 1994: 6; see also Gurr 1994). In MAR’s third and final step, politically mobilized minorities are ready to take on political actions, like protest and rebellion. External support is anticipated to reduce the asymmetry between the state and rebels and to facilitate mobilization, along with protest and rebellion, by increasing the level of group cohesion and identity of the aggravated groups (Gurr 2000; Fox 2001). External support is defined as “verbal encouragement, financial support, weaponry, military personnel, and other forms of active or passive support the ethnic group receives from outside the state” (Harff and Gurr 2004: 105). Third parties to ethnic conflict can sponsor negotiations, impose sanctions on one side or the other, or even intervene militarily. External material, political and moral support contributes to group cohesion and political mobilization. Thus, it is expected that a high level of external support will not only increase the likelihood of violent minority mobilization by amplifying group capacities and opportunities for action up to the level of rebellion (Harff and Gurr 2004; Gurr 2000) but also complicate the assurance-deterrence dilemma of the state. What, then, is the overall role of external support in the MAR model and security dilemma mentioned above? External support can be crucial to minority success and is the main source of internationalization for ethnic conflict (Saideman 2001a, 2001b). However, few studies have enquired into why some minorities receive support while others do not. Could the type and level/extent of discrimination, degree of fear, and insecurity provide some of these reasons? Human rights abuses and refugee flows would appear to be salient reasons to intervene. A tendency among discriminated groups to internationalize their conflict in the hope of obtaining support is another one. Compared to the generally better-equipped and organized armies of states, ethnic groups have little military power. Thus a quest for third-party support naturally emerges.

The “chicken or the egg”? 167

Hypotheses This research design will test two sets of hypotheses, about external support and rebellion, respectively. The first set of hypotheses focuses on external support. Three hypotheses pertain to some form of behavior by the host state and/or minority groups as the independent variable: H1: The likelihood of external support increases to a limited degree as the host state increases political discrimination against minorities within its borders. H2: The likelihood of external support increases to a limited degree as the host state increases economic discrimination against minorities within its borders. H3: The likelihood of external support first increases and then decreases as the level of violence increases between the host state and minority groups. The first two of these hypotheses are based on intuition about who might take an interest in discrimination. External support may ensue as acts of discrimination are witnessed by outside supporters of the ethnic group, leading to calls for assistance or even spontaneous, unsolicited help in one form or another. H1 and H2 refer to a limited relationship because, however repugnant, discrimination is not anticipated to have the same impact on outside observers as mass killing or even more restricted levels of overt violence. While either the elite or general public in another state might sympathize with a minority experiencing discrimination, the likelihood of active support would not increase dramatically in the same way anticipated as a result of media coverage, for instance, of more violent suffering. Both of the hypotheses put forward a positive relationship between external support and economic and political discrimination. A follow up question is “Are economic and political discrimination equally powerful in their ability to mobilize masses?” Studies that have looked at the factors of economic and political discrimination suggests that theoretically both are equally powerful in mobilizing groups. For instance, Brown et al. (2001) regards discriminatory economic systems and discriminatory political institutions as underlying causes of internal conflict. Gurr’s (2000: 169) analysis of mobilization of ethnic groups includes both economic and political discrimination, although the number of groups facing economic (p. 179) and political (p. 211) discrimination in 1990 and 1991 is slightly different. Mean scores for the same period are 1.73 for economic discrimination and 2.26 for political discrimination. The third hypothesis puts forward a curvilinear connection as a result of conflicting ideas found in the literature, each of which seems theoretically defensible and supported by evidence. Consistent with the discussion a

168 Y. Akbaba et al. moment ago, Saideman (2002) argues that violent conflict will make identities more salient and increase the likelihood that the mass public of another state, notably one in which the minority at risk also exists, will get involved by offering support. Regan (1998), by contrast, argues that external intervention becomes less likely because extremely violent conflict greatly reduces the prospects of success. Risk aversion takes precedence and elites are less likely to offer support. Putting together the arguments from Regan and Saideman thus produces a curvilinear expectation: higher (lower) likelihood of external support at lower (higher) levels of violence between the host state and its minorities. Three hypotheses focus on attributes of the host state and derive from an argument concerning vulnerability: H4: Separatist minority groups are less likely to receive external support than other groups. H5: Minorities that reside in states with a high number of other separatist groups are less likely to receive external support than other minorities. H6: Separatist groups in neighboring countries decrease the likelihood of external support to minorities in host states. The presence of separatist groups, as opposed to those who might have more limited goals, is taken as a sign of particular vulnerability for a state. Moreover, there are neighborhood effects to consider. States in a subregion where separatism is common are more likely, according to the “vulnerability” argument as described by Saideman (2002: 30), to show restraint toward ethnic conflicts in their vicinity in the hope of reciprocity. Case studies suggest that states generally are unwilling to support separatism, even when matters become violent (Carment and Rowlands 2004). Saideman’s (2002: 42) data analysis, however, produces results that are either non-existent or opposite to those expected, which would tend to support a more realist-oriented view of the dynamics of mutual vulnerability. Instead of cooperation through observance of norms, as might be expected by a liberal perspective, the result is something more like the Tragedy of the Commons, where meddling out of self-interest becomes the modal result among the mutually vulnerable states. H4 through H6 are worded as would be anticipated by the conventional wisdom on the subject, but more nuanced results, possibly once again supporting a realist point of view, would not be considered anomalous if they occur. Three hypotheses focus on ethnic affinity and cleavage, factors that have demonstrated some importance in explaining third-party intervention in a series of case studies (Carment et al. 2006): H7: The likelihood of external support increases when ethnic kin of the minority group dominate an adjoining state.

The “chicken or the egg”? 169 H8: The likelihood of external support increases when religious differences distinguish the ethnic minority group from the majority. H9: The likelihood of external support increases when racial differences distinguish the ethnic minority group from the majority. H7 is based on the idea that ethnic affinity will raise the probability of external support. An example of how this might work in practice is the case of Sri Lanka and its Tamil minority, which traditionally looks across the Palk Strait for both moral and even more tangible support from ethnic brethren in India. As ethnic tensions rose within Sri Lanka during the mid1980s, both politicians and the general public in Tamil Nadu pressured the Indian government to recognize the grievances of ethnic kin in Sri Lanka and offer support. India eventually did intervene, with mixed results, in 1987 (Carment et al. 2006). Religious and racial differences variables are incorporated in H8 and H9 in recognition of the potential effects of group-based cleavages. To begin, salience of ethnic identity appears as an important catalyzing factor in models from Gurr’s MAR Project. Gurr (2000) argues that greater salience of ethnocultural identity facilitates mobilization for collective action. Although he accepts that communal identities are multidimensional, he distinguishes some traits such as race and religion from others since they are intrinsically more important. Aside from secular societies, religion is perceived as a strong source of group cohesion in many instances. When incompatibility of two religions is underlined due to any reason, salience of ethnic identity is assumed to increase suddenly for ethnic groups. Fox (2001), for example, finds symmetry in external support from states vis-à-vis the religion of those being assisted. Islamic and Christian states tend to help their own religious brethren elsewhere, although this effect is more pronounced among the former than the latter. Racial differences between people are rivaled perhaps only by gender in immediate visibility, so the grounds for inclusion of this variable in the framework are even more obvious here. Two variables have the status of controls. Africa also is singled out because of the preceding arguments back and forth, from liberal and realist points of view, about mutual vulnerability causing restraint, or, perhaps instead, encouraging external support in the attempt to achieve relative gains. The regime type of the host state also is included due to its strong presence in the literature on conflict processes over the last 15 years (Russett and Oneal 2001; Choi and James 2005). Authoritarian regimes, which are not constrained by institutions or public opinion, can act more easily in ways that cause grievances for minority groups – for example, by simply ignoring their interests or suppressing them. This also can happen with a democratic government, but is regarded as less likely due to existing rules, institutions, and norms. Thus, regime type emerges as something important to recognize in assessing the relationship between discrimina-

170 Y. Akbaba et al. tion, violence, separatism, and external support. Moreover, existing data analysis from MAR links regime type to respective patterns of ethnic conflict: “[e]thnopolitical groups in democratic societies are more likely to use strategies of protest than rebellion” (Gurr 2000: 154). In particular, compared to autocratic regimes, democracies have a smaller number of ethnic wars and rebellions (Gurr 2000). For the second set of hypotheses, which focuses on rebellion, a more brief explanation will be in order because the effects mostly follow intuition from the connections described for the first set of hypotheses. H1 and H2 are straightforward derivatives from the MAR model presented earlier. Acts of discrimination by the state are anticipated to stimulate mobilization, which makes rebellion more likely to occur. H3 is based on a reciprocal effect. Just as rebellion is expected to encourage external support, the latter is anticipated to stimulate the former. H4 and H5, which pertain to whether a group is separatist and resides in a state with a high number of other separatist groups, assert a positive connection with rebellion. Separatism is a more “zero sum” type of movement within a state. If a minority is committing to leaving and taking territory with them, the conflict becomes more fundamental and, therefore, more likely to produce rebellion than otherwise. The same is true with a higher number of secessionist groups because the state must deal with so many challenges to its authority that rebellion becomes more likely to emerge in the absence of especially astute conflict management. The effect for H6 is in the direction opposite to the other two variables pertaining to separatism: Neighbors with their own separatist problems, based on the logic of vulnerability, become less likely to offer support to ethnic groups abroad out of presumed mutual restraint. This is another manifestation of the “African factor” as introduced earlier, namely, risk aversion about intervention among weak states. H7, which concerns ethnic kin in a dominant position in an adjacent state, puts forward a positive connection with the likelihood of rebellion. This is a straightforward reflection of at least perceived greater opportunity for action. Finally, H8 and H9 focus on religious and ethnic differences. In each instance the expected association with rebellion is positive because these variables pertain to the salience of differences between, and among, groups. The more salient the differences, the higher the likelihood of rebellion, all other things being equal, because there is a greater disposition toward mobilization of groups in conflict with each other. Control variables also operate in much the same way in this second stage of the framework. Once again, a check is made for effects related to Africa. African states, of course, are chronically troubled and thus also more likely to experience rebellion, at least on the basis of intuition. The other control, regime type, associates democracy with peace. Thus democracies are anticipated to settle their internal differences through peaceful means, with rebellion as less likely than in autocracies.

The “chicken or the egg”? 171

Data, measurement, and method: data sources Minorities at Risk’s data set 2002 and the external support data set compiled by Saideman (2002) are used in this research design. Saideman’s external support data is compatible with the basic MAR data, where the units of analysis also are minorities that exist within states. Two aspects of the data are worth noting in particular: first, MAR is the only large-scale data set that focuses on minorities at risk. MAR’s 2002 version includes political and economic discrimination against 285 politically active groups throughout the world from 1980 to 2000. Since our concern is not with states that receive support, but instead ethnic groups, MAR data (Gurr 2002; see also Saideman 2002: 34) becomes the obvious choice.4 Second, the third-party intervention or external support variable is not available in the final release of MAR (Gurr 2002). Therefore, Saideman’s external support data is combined with it.5 The time interval we can cover is from 1990 to 1998. Dependent variables We have two dependent variables, the intensity of external support and rebellion. Intensity of support refers to the highest level of support given by a third-party state to a minority group (Saideman 2002). The coding is as follows: (0 – None): No support received; (1 – Low): Ideological encouragement, diffuse support (relatively weak forms of support), other unspecified support; (2 – Moderate): Non-military financial support, access to external communications, markets, transport, including the hosting of nonviolent exile organizations; (3 – Strong): Funds for military supplies, provision of military equipment and supplies, military training in exile, advisory military personnel, peace-keeping observers; (4 – Intense): Blockades, interdiction against regime, cross-border sanctuaries for armed fighters, rescue missions in country, cross-border raids in support of dissidents, active combat units in country. In our data set this is coded as three biennial (1990–91, 1992–93, 1994–95) and three annual (1996, 1997, 1998) variables (see MAR 2002; Saideman 2002). Rebellion measures level of conflict between group and host state. The variable measures level of conflict between ethnic group and host state. The categories are as follows: 0 – None; 1 – Political banditry, sporadic terrorism; 2 – Campaigns of terrorism; 3 – Local rebellions: Armed attempts to seize power in a locale6; 4 – Small-scale guerrilla activity; 5 – Intermediate scale guerrilla activity which has one or two of the defining traits of large-scale activity and one or two of the defining traits of smallscale activity; 6 – Large-scale guerrilla activity; 7 – Protracted civil war that is fought by rebel military units with base areas.

172 Y. Akbaba et al. Independent variables Independent variables included in the analysis are as follows: external support, political and economic discrimination, rebellion (also rebellion squared), separatist groups, other separatist groups in the host state, separatist groups in adjacent state, domination by ethnic kin in adjoining state, religious differences, and racial differences. Rebellion and external support play roles as independent variables in the respective models, but their coding already has been explained above. Political discrimination accounts for how certain groups are (or have been) limited systematically in exercising their political rights or access to political positions as compared to others in their society. This variable, therefore, represents “macro coding of the role of public policy and social practice in maintaining or redressing political inequalities” (MAR 2002: 59). The categories of government involvement are as follows: (0) no discrimination; (1) historical neglect/remedial policies (i.e. substantial underrepresentation in political office and/or participation due to historical neglect or restrictions and explicit public policies are designed to protect or improve the group’s political status); (2) neglect/no remedial policies (i.e. substantial under-representation due to historical neglect or restrictions, no social practice of deliberate exclusion, no formal exclusion and no evidence of protective or remedial public policies); (3) social exclusion/neutral policy (i.e. substantial under representation due to prevailing social practice by dominant groups and formal public policies toward the group are neutral or, if positive, inadequate to offset discriminatory policies); and (4) exclusion/repressive policy (i.e. public policies substantially restrict the group’s political participation by comparison with other groups) (MAR 2002: 59–60). Economic discrimination focuses on how certain groups are (or have been) intentionally limited in their access to the conditions of material well-being in comparison with others in their society (MAR 2002: 75). The economic variable measures the level of discrimination based on government involvement, with categories as follows: (0) no discrimination; (1) historical neglect/remedial policies (i.e. significant poverty and under representation in desirable occupations due to historical marginality, neglect, or restrictions and public policies which are designed to improve the group’s material well being); (2) historical neglect/no remedial policies (i.e. significant poverty and under representation due to historical marginality, neglect, or restrictions and no social practice of deliberate exclusion and few/no public policies aimed at improving the group’s material wellbeing); (3) social exclusion/neutral policies (i.e. significant poverty and under representation due to prevailing social practice by dominant groups and formal public policies toward the group are neutral or, if positive, inadequate to offset active and widespread discrimination); and (4) restrictive policies (i.e. public policies – formal exclusion and/or recurring repres-

The “chicken or the egg”? 173 sion – substantially restrict the group’s economic opportunities by contrast with other groups) (MAR 2002: 75). Economic discrimination data are available for the same individual years as the political discrimination variable in the 1990s (Gurr 2002). As with political discrimination, the MARbased economic discrimination variable is adapted to the host state level by taking the maximum level of discrimination against any of the minorities within the state. Since the goal is to assess the influence of economic discrimination on the likelihood of receiving support, for each case the value one year before the date of intervention is taken to represent economic discrimination for that state. Missing values are handled in the same way as for political discrimination. MAR’s economic and political discrimination variables are coded for most individual years in the 1990s (1990–98) (Gurr 2002). For every case, the maximum level of economic or political discrimination exercised by a particular state against all of the minorities within its boundaries is recorded. In other words, when external support is the dependent variable, a one-year lag is applied to the discrimination variables. Separatism is included in various ways in this study to capture different dimensions of the “vulnerability” argument as described, but not endorsed, by Saideman (2002). Separatist group is a MAR index that includes both historical and active separatist groups. Saideman (2002) recodes MAR’s index into a dichotomous variable. If the group is actively separatist it is coded as 1, otherwise 0. The recoded version is used to focus on the impact of groups that are currently engaged in an effort to gain separation. Other separatist groups in the host state indicates the number of other groups in the same state that are actively separatist (Saideman 2002). This variable assesses vulnerability of a state based on the number of MAR that are engaged in efforts to separate. Separatist groups in adjacent state refers to the number of separatist groups in adjacent states, an indicator of possible spillover or neighborhood effects (Gleditsch 2002). The three MAR-based, separatism-related variables incorporate interrelated aspects of the vulnerability approach, which argues that separatist groups are less likely to receive external support. Ethnic kin dominant in adjoining state is a variable with good intuition and a strong track record behind it (Saideman 2002). This factor, perhaps more than any other, measures the degree of willingness that a state will possess to consider offering support to ethnic brethren. When those of concern are right next door, disposition to act should be much higher than otherwise for any number of reasons. If the group dominates or is in the majority in the state adjacent to the host state it is coded a 1, otherwise 0. Data is taken from Saideman (2002). Religious and racial differences appear as indicators of ethno-cultural distinctiveness in the MAR dataset. These differences can be taken as an indicator of potential cleavage or disposition to conflict within a society. Religious differences are coded as follows: ethno-religious minorities that

174 Y. Akbaba et al. are defined as ethnically and religiously distinct from the rest of a state’s population to the extent designated in each cateogory: (0) unknown or none; (1) different sect within same religion as the dominant group; (2) multiple sects, some different from dominant group; (3) different religion than dominant group. For racial differences, MAR refers to the degree to which a minority can be distinguished by physical appearance. Relative physical visibility is the concern here. The coding is as follows: (0) none or unknown; (1) physically distinguishable subtype of same racial stock; (2) different racial stock from the dominant group with substantial intermixture; (3) different racial stock, little or no intermixture. Control variables African state is a dummy variable in which states in Sub-Saharan Africa are coded 1 and all others 0. Regime type is taken from Saideman (2002), who used Polity 98 data and subtracted each host country’s autocracy score from its democracy score, producing an indicator ranging from –10 (most autocratic) to 10 (most democratic). Method Seemingly Unrelated Regression (SUR) is used to analyze the data. This technique is ideal for addressing our “chicken and egg” problem. SUR produces two equations and takes account of interdependencies in the errors between them.7 Data analysis Table 9.1 conveys the SUR results for external support. Overall, the results are encouraging. Hypotheses from each of the three major categories – behavior, separatism, and affinity and cleavage – are supported. Three hypotheses are supported, three rejected, one produces mixed results and two with competing theoretical bases (i.e. realism versus liberalism) are significant but not as phrased above. Given different levels of expected performance for the hypotheses, these results are favorable to further investment in the model. H1 and H2, which focus on political and economic discrimination, are not supported. Political discrimination is positive and significant for just one of the three years in the table. Expectations had been somewhat low here because political and economic discrimination, while repugnant, do not convey the same intensity as more extreme behavior leading to loss of life and therefore may not stimulate external support to the same degree as other factors. By contrast, quite impressive is the performance of H3. Rebellion is positive and significant for all three years. Interesting also is the negative coefficient for Rebellion squared, although it is significant

The “chicken or the egg”? 175 Table 9.1 SUR analysis of external support for minorities at risk

Political discrimination Economic discrimination Rebellion Rebellion2 Separatist group Other separatist groups in the host state Separatist groups in adjacent state Does ethnic kin dominate adjoining state? Religious differences Racial differences African state Regime type Constant R2 N

1992–93

1994–95

0.12* (0.07) 0.00 (0.08) 0.54** (0.15) –0.03 (0.02) –0.02 (0.20) –0.17** (0.06) 0.06** (0.02) 0.37* (0.20) –0.08 (0.06) 0.10 (0.08) 0.01 (0.22) –0.02 (0.01) 0.04 (0.23) 0.34 191

0.07 (0.05) –0.05 (0.06) 0.66** (0.11) –0.05** (0.02) –0.03 (0.16) –0.08* (0.04) 0.05** (0.02) 0.59** (0.16) –0.01 (0.05) 0.15** (0.06) –0.23 (0.19) –0.03** (0.01) 0.16 (0.20) 0.39 222

1998 –0.08 (0.06) 0.01 (0.06) 0.47** (0.12) –0.01 (0.02) –0.15 (0.15) –0.05 (0.04) 0.06** (0.02) 0.32* (0.16) –0.00 (0.05) 0.01 (0.06) –0.46** (0.18) –0.05** (0.01) 0.71** (0.20) 0.40 248

Notes *p < 0.05, **p < 0.01.

only for 1994–95. This hypothesis is very demanding and its curvilinear form is supported, at least provisionally, by testing. Thus, to return to the earlier discussion, both Saideman and Regan seem to be on the right track, if they are taken to mean that violence can have opposite effects on the likelihood of external support as a function of its level. H4, on whether a group is separatist, is not supported; all coefficients are close to 0. This result is consistent with the data analysis from Bélanger et al. (2005: 454), which focuses on secessionist conflicts from 1990–92 and finds that secessionist movements receive very little external support. Surprising to the liberal argument is the negative coefficient for H5, which also is significant for 1992–93 and 1994–95. The number of separatist groups in a state, a measure of its vulnerability, is associated with a higher likelihood of external support. This result would seem to encourage a realist interpretation, with perceptions of a state’s weakness tending to

176 Y. Akbaba et al. attract interference from outside, setting aside any normative judgments about when external support for ethnic groups might be justified on humanitarian grounds. It is interesting to note that H6, which focuses on whether separatist groups exist in a neighboring state, is significant but positive. Once again, this result favors a realist over a liberal interpretation. The argument that mutual vulnerability will create a neighborhood effect inducing restraint vis-à-vis external support is not consistent with this data. The pursuit of relative gains through exploitation of vulnerability instead seems more in line with the present results. Indeed, the findings for all three of the hypotheses about separatism would tend to support Saideman in his criticism of the vulnerability thesis. H7, on whether ethnic kin are dominant in an adjacent state, is supported strongly by the data, with a positive and significant result for all three years. This result undoubtedly reflects greater willingness to act when ethnic brethren are right next door and exist as a minority. While the state-of-the-art study on intervention by Bélanger et al. (2005: 448) does not find that ethnic ties help to explain external support, there is a sound explanation for this discrepancy. The present study differs from theirs in that it focuses more specifically on whether ethnic kin are dominant in the neighboring state. Thus the positive finding here in all likelihood is as a product of the interaction between physical and ethnic proximity. H8, on religious differences, is not supported. It is very interesting to discover that religious differences do not affect the likelihood of external support. Finally, H9, which pertains to racial differences, does somewhat better. The sign is in the right direction for all three years and significant in 1994–95. Taken together, H8 and H9 suggest that, to the extent that differences matter in stimulating external support, religion takes a back seat to ethnicity. Results are mixed for the first of the two control variables, which pertains to African states. The coefficient is around 0, negative, and positive for the three years respectively, so the straightforward interpretation is that no consistent connection is present. By contrast, regime type performs as expected and with a significant connection in 1994–95 and 1998. Democracies are less likely than autocracies to have external support come in for minorities. This result rings true because, all other things being equal, other states are less likely to intervene in a democracy out of belief that it is more probable that such regimes ultimately will be able to manage their ethnic conflicts with some degree of legitimacy. In their parallel study, Bélanger et al. (2005: 448) find that intervention is less likely for democratic dyads. The result concerning democratic disposition away from external support might be regarded as a monadic analogue to the finding just described, suggesting that a more general connection may exist between regime type and ethnic interventionism. Table 9.2 displays the results for the hypotheses about rebellion. Once again, the results are encouraging, if not overwhelming. Four out of nine

The “chicken or the egg”? 177 Table 9.2 SUR analysis of rebellion for minorities at risk 1992 Political discrimination Economic discrimination External support Separatist group Other separatist groups in the host state Separatist groups in adjacent state Ethnic kin dominant in adjoining state? Religious differences Racial differences African state Regime type Constant R2 N

0.04 (0.11) 0.02 (0.12) 0.84** (0.10) 1.22** (0.27) 0.24** (0.09) –0.02 (0.03) –0.07 (0.31) 0.02 (0.10) 0.05 (0.12) 0.02 (0.34) –0.03 (0.02) –0.26 (0.35) 0.35 191

1994 0.14* (0.09) 0.03 (0.09) 0.79** (0.09) 0.96** (0.24) 0.11* (0.06) –0.004 (0.03) –0.24 (0.26) –0.07 (0.08) 0.02 (0.10) 0.55* (0.29) 0.02 (0.02) –0.65* (0.31) 0.33 222

1998 0.24** (0.09) 0.07 (0.09) 0.84** (0.08) 0.32 (0.23) 0.09 (0.07) 0.00 (0.03) 0.02 (0.25) –0.12 (0.08) –0.06 (0.10) 0.96** (0.28) 0.03 (0.02) –0.85** (0.32) 0.28 248

Notes *p < 0.05, **p < 0.01.

hypotheses are supported from two out of the three clusters, referring to behavior and separatism, respectively. Results are mixed for H1 and H2, which focus on political and economic discrimination, respectively. All of the coefficients are in the right direction, with political discrimination significant for two out of the three years, so H1 is regarded as supported by the data. This is consistent with, but not as strong as, the results obtained by Steinberg and Saideman (forthcoming) for government intervention in the economy. This result is particularly interesting because it suggests that the sheer amount of government manipulation of the economy is a better predictor of ethnic rebellion than direct measurements of discrimination, economic or otherwise. Perhaps this is because discriminatory acts are further back along the causal chain and escalating government manipulation of the economy is closer to the stage of rebellion itself. If so, this would complement the MAR model quite nicely as a further insight about how rebellion occurs. The idea is a candidate for more intensive process tracing that focuses on a

178 Y. Akbaba et al. smaller number of cases in greater depth. In other words, is there a causal chain from discrimination through government economic manipulation to ethnic rebellion? H3, on external support, fares extremely well. Strongly positive coefficients appear for each of the three years. External support increases the opportunity for rebellion both in theory and practice. The support rebels receive from an outside state provides additional capabilities, facilitates the mobilization process, and that, in turn, can increase likelihood of further discrimination. The important impact of external support on rebellion of groups might be related to the balance of forces that has changed due to the international support. Intervention in ethnic conflict puts the strife in a broader context of power relations. For instance, Indian involvement in Sri Lanka changed the domestic power structure and put the issue on the agenda of the region. Similarly, Iranian support to Kurds in Turkey in the 1990s shifted the balance of forces between Kurdish rebels and the Turkish state that increased the intensity of conflict. Hypotheses about separatism do well in an overall sense, with further suspicions arising about vulnerability-based arguments. H4, which pertains to whether a group is separatist, does reasonably well, with coefficients in the right direction for all three years and strong significance for 1992 and 1994. This is consistent with the logic of purely redistributive or zero-sum conflict. The goals of a separatist group are not even consistent with the continuing territorial integrity of the state in question, so ethnic rebellion becomes more likely than otherwise. This line of argument is reinforced by the equally solid performance of H5; when there are other separatist groups in the host state, the same pattern of results holds true. Support for H4 and H5 from the data suggests that as the state’s ability to deter some groups declines, other groups mobilize and seek to engage in violence as well. The third hypothesis on separatism, H6, focuses on whether secessionist groups are in an adjacent state, with the expectation – based on the vulnerability argument – that their presence would inhibit external support. There is no support for this hypothesis, once again raising questions about the overall utility of the vulnerability argument. Hypotheses about ethnic affinity and cleavage, surprisingly, produce no significant coefficients. H7, whether ethnic kin are dominant in an adjacent state, along with H8 and H9, which focus on religious and racial discrimination, respectively, receive no support from the data. These non-results are interesting to ponder in contrast to the better performance in the external support model. In speculating on cause and effect, perhaps affinity and cleavage are more important in stimulating external support and their causal effects vis-à-vis rebellion are, therefore, indirect and accounted for already through the strongly positive coefficient for H1. To see whether that might be the case, it makes sense to engage in process tracing and examine causal processes more directly through more in-depth case work. African state, the first control, is in the right direction each time and

The “chicken or the egg”? 179 significant for 1994 and 1998. This positive result is consistent with the intuition about Africa as a region disposed toward rebellion and even disorder in general. Regime type, the other control variable, is around 0 for each year, which is not something the literature would have anticipated. External support for a rebellion is independent of whether the host state is a democracy. Taken together, several patterns emerge in the findings about external support and rebellion. First, the question of the “chicken or the egg”? visà-vis ethnic rebellion and external support is answered as “both.” Rebellion and external support facilitate each other. Second, the results for external support in particular contradict the vulnerability thesis, most notably the positive association with rebellion of the presence of other separatist groups in the host state and in an adjacent state. This offers credence to realist, rather than liberal, accounts of the dynamics of ethnic strife, at least in this context. Third, ethnic affinity and cleavage, to the extent that they impact upon rebellion, appear to do so indirectly through a positive connection with external support.8 Fourth, among the variables in the model, the presence of other separatist groups in the state should receive special attention because it is the only one that is significant in accounting for variance in both external support and rebellion. Fifth, neither economic discrimination nor religious differences is significant in either stage of the analysis, so these variables might not be considered in more in-depth extensions of this project. Sixth, each control variable works at one stage of the analysis; democracy has the anticipated dampening effect on external support for an ethnic group and states in Africa are more prone to rebellion.

Conclusions This chapter started out with a focus on the question of the “chicken or the egg”? in relation to external support and ethnic rebellion. The answer is that each matters to the other. With respect to external support, the results confirm Downes (2004: 277–279) on the range of possible outcomes for external support, depending on its type and circumstances. In the present context, high levels of support, which include funds for military supplies, advisory military personnel and even cross-border raids and active combat units in a country at the upper scale points, are associated with rebellion. External support can alter the power balance between rebels and the state by bringing a new set of opportunities for the group. Therefore, states need to take into account the fact that, if minorities are not satisfied with the status quo (due to discrimination of other reasons), they can attract outsiders to the issue in order to fight for their cause and to facilitate their mobilization attempts. Results from the data analysis show the dual nature of external support in relation to the greater questions of assurance and deterrence raised at the outside of this chapter and

180 Y. Akbaba et al. in previous studies. Walter (1997, 1999), for example, finds that termination of ethnic warfare is stabilized by credible guarantees from external states. The present study focuses on an earlier stage of such processes. We find that rebellion elicits external support and is stimulated by it as well. Thus, at one stage, external support could have the effect of escalating strife, while at another guarantees by a third party might keep the peace. The simultaneous presence of such results should encourage further research on process tracing. How and why external support affects ethnic conflict across stages of its development are questions that will continue to preoccupy research for some time. Some final speculation is in order here. If, indeed, external support feeds rebellion – and rebellion, in turn, triggers support from third-party states – it is essential to verify the conditions under which external support assists the state in developing assurance policies. These conditions are of great interest in the quest to break rather than reinforce the cycle of violence. A “simple” answer probably would stimulate a new look at the initial wave of studies on intervention. For example, consider this argument: External third parties with an ethnic affinity to a minority, when they intervene, always will side with their brethren and they will end up altering the balance of forces. Thus the chapter concludes with multiple and interesting ideas for further research.

Notes 1 See Suhrke and Noble (1977), Rothschild (1981), Heraclides (1990), Carment (1993), Carment and James (1995), Regan (1996, 1998), Gurr (2000), Regan and Stam (2000), Saideman (2001a, 2001b), Brown (2001), and Carment et al. (1997); Gurr (2000) focuses on international organizations. 2 The role of physical security, a double-edged sword that can be a concern either for an ethnic minority or the state center, is recognized later through incorporation of rebellion into the framework. 3 Gurr (2000: 107) is careful to note that inequalities among groups do not necessarily result from discrimination. 4 Groups are included in the MAR data set if the country in which they reside has a population exceeding 500,000 and the group itself has a population larger than 100,000 or 1 percent of the country. Minorities not at risk are not represented in this data set (MAR 2002). 5 MAR does not include the transnational support variable (except for 1999 and 2000) in its current version because the project is in the process of recoding. 6 Some nuances in the coding should be clarified here. For the coding of local rebellion: “If they prove to be the opening round in what becomes a protracted guerrilla or civil war during the year being coded, code the latter rather than local rebellion. Code declarations of independence by a minority-controlled government here” (MAR 2002: 90). For small-scale guerrilla activity, “All of the following must exist: 1) fewer than 1,000 armed fighters; 2) sporadic armed attacks (less than six reported per year); and 3) attacks in a small part of the area occupied by the group, or in one or two other locales” (MAR 2002: 90). Finally, for large-scale guerrilla activity, “All of the following must exist: 1) more than 1,000 armed fighters; 2) frequent armed attacks (more than six per

The “chicken or the egg”? 181 year); and 3) attacks affecting a large part of the area occupied by the group” (MAR: 90–91). 7 A model may contain several linear equations. It would be unrealistic to expect that the equation errors would be uncorrelated with each other. We have two equations in which the dependent variable of the first appears as a predictor in the second and vice versa. Fitting two separate regression equations independent from each other would be an error here, since we would not be counting for the cross-equation error correlation. A set of equations that has contemporaneous cross-equation error correlation is called a SUR system. At first glance, the equations seem unrelated, but they really are connected through correlation in the errors (Zellner 1962; Srivastava and Giles 1987). 8 However, these tests only address whether a group is different from the majority, not whether potential supporters have ethnic ties to one side or the other.

10 Tackling the anarchy within The role of deterrence and great power intervention in peace operations Sarah-Myriam Martin-Brûlé Introduction State out, outsiders in The monopoly of violence in a determined territory is an essential characterisitic of statehood (Weber 1978). The acceptance or recognition of a given authority structure is one aspect of domestic sovereignty yet the level of control that officials can actually exercise is crucial for that sovereignty to be effective (Krasner 2001). Yet, since the end of the Cold War and with it, the end of superpower support to their allies, there has been a marked increase in the number of states whose monopoly of force has been challenged. A plethora of terms has been put forward to describe phenomenon: fragile states (USAID), wobbly states (David Sogge), crisis states (LSE), states at risk of instability (UK Cabinet Office study), warlord states (Will Reno) (Woodward 2005). These terms refer to a combination of factors but they all point to the ineffectiveness of a government to control its territory, the lack of perceived legitimacy by a significant portion of its population, the high level of domestic insecurity and/or the failure to provide basic public services to its citizens (Failed States Index 2006). In these conditions, the legimitate monopoly of force might be seen in the hands of outside actors. Indeed, if one of the essential roles of government is to deter violence – to possess a monopoly of the legitimate use of force and to wield it so that individuals and groups refrain from engaging in serious conflict – when the state is too weak, the question then becomes: how can outsiders step in to play this essential role? The question is twofold: To what extent are foreigners, and more specifically UN peace operations, better equipped to take over this monopoly of violence? What are the advantages/limits of outside interveners and what are the associated short- and long-term trade-offs for their intervention? In the literature, these questions have been addressed in the framework of the credible commitment theory. According to this theory, the anarchic character of intrastate wars makes it hard for actors to arrange on their own credible guarantees for peace settlement.

Tackling the anarchy within 183 Third parties are presented as necessary to enforce or to verify the implementation of agreements and ultimately, for the success of the peace operation (Walter 2002). The accomplishment of the latter’s mandate is seen as dependent on two kinds of guarantees – conceived as either simulatenous or consecutive, depending on the context: (1) the guarantee of the security of the belligerents who comply with the parameters of the agreement, and (2) the guarantee of compliance to the agreement. Each group has thus two expectations: the deterrence of their adversaries, and the assurance that this deterrence will be effective. The success of peace operations not only depends on the willingness and ability of third parties to accomplish their mission but also in the confidence that the protagonists have in the commitment of interveners (Walter 2002). Confidence seems intrinsically linked to the coercive potential of a mission. The belligerents’ fierceness not only toward each other but toward civilians and even toward the staff of peace operations points to the necessity of coercion for the operation to be effective (Roberts 1995–1996; Stephens 2005; Brunnée and Toope 2004; Mansson 2005). In such conditions, the capacity of interveners to use coercive force is understood as the sine qua non condition to guarantee the security of the belligerents who comply to a peace agreement. Yet, the credibility of the interveners that have the capacity to use this coercive force seems the determinant factor in assuring that the agreement will be respected (Weiss 1994; Hirsh and Oakley 1995; Drysdale 1997; Walter 2001). Still if credible intervener and credible commitment matter, one crucial ingredient is missing: strategy. Strategy is described as a chain of relationships among means and ends that span several levels of analysis from the maneuvers of units in specific engagements through larger campaigns, whole wars, grand strategies and foreign policies (Bett 2000:6). My take is that deterrence and assurance should be seen not as outcomes but as the very means of intervention to restore security and the monopoly of violence within and by the state. The focus is on how deterrent strategies interact with the relative power of intervening countries to improve peace operations’ efficiency by assuring the belligerents that the peace agreement will be implemented and that the protagonists will comply. The cases of peace operations in Sierra Leone and Liberia indicate that deterrence success – the threat of force rather than the use of force – is more likely when a great power is directly involved, leading the intervention. The adoption of a deterrence strategy combined to the intervention of a great power thus seems to contribute to the success of a peace operation, i.e. the accomplishment of the mandate and the efficiency of the peace operation.

Theoretical discussion Failed states are easy cases for testing the credible commitment theory, yet tough cases for assessing deterrence strategy. Both characteristics (easy

184 S.-M. Martin-Brûlé and tough) are associated with the two-sided coin that are failed states. Because failed states have to be rebuilt from scratch, it is easy to prove that any commitment will have an effect. Yet, because of this lack of any security provision, it is the least likely case for deterrence to be effective. I identify two major challenges for peace operations in failed states settings: the prerequisite of re-establishing a degree of public order before undertaking political procedures toward peace building and the coordination of the various members, elements and structures of the operation. This translates into establishing a relation of deterrence and assurance from square one. Hand in hand with credible interveners and credible commitment, credible strategies matter. Yet, if security is not so much a good – the presence or absence of which defines rules of interaction – but a continually renewed bargain between the state and constituent groups, in the case of intervention in intra-state wars within failed states, security then becomes a matter of establishing a new bargain, between the interveners and the constituents groups. Bargaining is about deciding how to divide the gains from joint action (Powell 2002). In economics, bargaining refers to the process by which a buyer and a seller agree on a price. The bargaining model overlaps with deterrence theory (Reiter 2001). In deterrence theory, an actor attempts to deter a potential attacker by promising to make a specific response if the second actor takes some action. Deterrence succeeds when the potential attacker does not attack. In this context, the agreement is on the “price” of actions, i.e.whether a threat is executed or not. Great powers possess two main advantages which render their commitment more credible and hence bolster the probability of success of a peace operation in failed states. First, the credibility of great powers stems from their capacity to mobilize more resources (human and material) and to do so more efficiently. Second, given the magnitude of their human and material investment, the great powers might draw their credibility either from the stakes they have in the success of the operation and/or building on their established reputation.1 Great powers share the dual advantage of being and appearing more committed. They are thus more likely to convince groups of the credibility of their commitment and hence contribute to the probability of success of the operation (Walter 2002:27). Yet, on many occasions the presence of a great power has not proven sufficient in itself to assure the success of peace operations. If credible commitment and credible intervener matters, the strategy employed by the great power might be key to the success of its mission. There are many strategies available to great powers. Among the most common strategies are economic sanctions and/or military interventions. However these strategies have not been efficient. Economic sanctions are a type of strategy which could be argued to be the main determinants of success or failure of peace operations in civil

Tackling the anarchy within 185 wars fueled by traffic of arms and diamonds (Pugh et al. 2004; Collier 2001; Hirsh 2001b). In Sierra Leone, numerous measures have been taken by international financial institutions and international organizations (UN under Chapter 7) in order to abate the economy of war.2 Yet sanctions and restraints have produced mixed results which led many to question the adequacy of these measures in such settings (Orogun 2004). For example, in a context of weak state control combined with rampant corruption such as is the case in Sierra Leone, the adoption of the diamond official certification system has made it easier to launder these gems since it has eluded the need for complicated smuggling through other countries (Cooper 2003: 11). A similar phenomenon is also said to have happened with arms embargoes in Liberia. Despite ten years of embargoes on arms and military equipment, a steady supply has been reported to enter the region and authorities have not complained about the sanctions (Pugh et al. 2004). Hence, the sanctions have been incapable of stopping the violence their very implementation subsidized. Arms embargoes and economic sanctions may be useful and may represent complementary obstacles to violence, but are not in themselves sufficient to ensure the success of peace operations (Oudraat 2000; Tierney 2005). This observation thus calls for the complementarity of strategies and action. Deterrence strategy refers to the use of threats in order to convince a potential aggressor to abstain from undertaking an action (Huth 1999). Deterrence has often been criticized for its elusive character and for the associated difficulty of distinguishing it from other coercive strategies such as compellence (Lebow and Stein 1990). Deterrence and compellence are two types of coercive strategies, i.e. both are threat-based and both seek to influence the behavior of another by using force, or the threat of force, to narrow the range of the adversary’s options – more specifically, the options which diverge from the coercer’s aims (Schaub 2004). Coercion differs from brute force. According to Schelling, the use of force is essentially a continuation or escalation of a threat first articulated before any resort to force. What distinguishes coercion is the use of force in a measured and controlled way to “signal” to the target the threat of further punishment unless it complies. “Coercion depends more on the threat of what is yet to come than on damage already done” (Carment and Harvey 1999). In a strategy of compellence, coercive demands are related to forcing the undertaking of a certain action, whereas in a deterrent strategy, demands relate to preserving the status quo, i.e. refraining from committing an action (Harvey 1999). Deterrence strategy combined with the involvement of a great power allows a peace operation to respond adequately to the problems relative to the context understudy and to efficiently accomplish the mandate. It converges with the key dilemma for states and peace-enforcing efforts which is to be powerful enough to deter belligerents yet sufficiently responsive to reassure the population.

186 S.-M. Martin-Brûlé The adoption of a deterrence strategy combined to the intervention of a great power thus seems to contribute to the success of a peace operation defined in terms of the restoration of the monopoly of violence. Somalia is a telling case in study of peace operations in the midst of intrastate wars. Within a year, three peace operations were undertaken under various commands as well as under various chapters of the United Nations’ Charter (6 and 7): United Nations Operation in Somalia (UNOSOM), Restore Hope and UNOSOM II. UNOSOM was launched under Chapter 6, which limits the use of force to self-defence; Operations Restore Hope and UNOSOM II took place under the aegis of Chapter 7 and were thus authorized to use coercive force. All three were mitigated successes. However, Restore Hope proved to be not only more efficient than UNOSOM, it was more efficient than UNOSOM II (Hill and Malik 1996). My study of these three operations confirmed the following finding: it seemed that it was less the use of coercive force than the resolve to use coercive force and the demonstration of one’s capability to do so, hence the operation’s deterring character that had been the main determinants of success.3 The use of a deterrent strategy had met the prerequisite of reestablishing a certain degree of public order before the delivery of food and care and the coordination of the various members, elements and structures of the operation and the threat to use coercive force had rendered the operation more efficient (Martin-Brûlé 2003). To measure the application of this strategy, I will refer to the prescriptive nature of deterrence theory and will operationalize this variable by breaking it into three parts. First, I will consider the communication between the belligerents and the interveners, among the belligerents and among the interveners (George and Smoke 1989; Harvey 1999).4 The defender ought to clearly define the acts to deter. It may then, either, clearly state the impending sanctions if transgression were to occur or leave the consequences obscure to complexify and heighten the risk assessment from opponents (Lebow and Stein 1990).5 Communication among interveners and among belligerents is necessary to assure the articulation of groups’ actions and principles. Communication between the intervening groups and third parties is also of importance. The transparency of the operation with regard to the international community can also contribute to increase the legitimacy of the operation and to promote national and international political support.6 Second, I will evaluate the peace operations’ capacity and demonstration of resolve to retaliate against a transgression of prohibited actions (and hence to make use of this capacity) (Huth 1999). The authority conferred by the UN mandate (whether it allows or prohibits coercive force, i.e. under Chapter 6 or 7 of the United Nations Charter), importance of the military deployment (quantity and quality of weapons, number of soldiers) and the use of sanctions (number and types of transgressions from the belligerents, number and types of sanctions related to punish the bel-

Tackling the anarchy within 187 ligerents’ transgressions) will serve as indicators of the capacity and demonstration of resolve. Finally, the interveners’ and belligerents’ “knowledge” of each other (based on reputation and known interests) will be assessed to determine to what extent strategies were calibrated to the specific adversary (Zagare and Kilgour 2000). For example, the credibility of the threat perpetrated by the defender will be enhanced or lessened depending on whether this defender had previously failed to sanction or severely punish a proscribed behaviour. Reputation can be assessed in terms of the results of previous encounters between the protagonists, previous similar interventions, the number of sanctions applied after transgression from the defender. The interests of the adversary, whether known or suspected, also factor in the cost–benefit calculus related to an attack (Lebow and Stein 1990). Thus, if the adversaries believe that a defender has many or important interests in the success of his mission, they will anticipate stronger sanctions for the transgression of a proscribed act. The probability of transgressions will thus decrease. Furthermore, awareness of the attackers’ interests may enable the defender to better negotiate with his adversaries. He will be able to stress the “relevant” benefits associated with renouncing to carry the attack. Interest can be gauged in terms of the financial resources invested in the operations, the number of actors engaged in the operation, the domestic political support enjoyed by the interveners and the nature and prioritization of sought benefits (financial benefits, legitimacy, etc.). Deterrence strategy is directly linked to an assurance/re-assurance process. When used by peace enforcers, it represents the delicate balance between deterring violence and channelling dissent among the belligerents and between the belligerents and the interveners. A deterrence strategy has several advantages. It answers the expectations of groups that their adversaries will be deterred while increasing (in comparison with other strategies) the assurance that the deterrence sought will be effective. Privileging non-coercive means, albeit having the authority and capacity to use force, it avoids legitimating warfare, and it contributes to the promotion of pacific means of communication. Using the appropriate balance of power to back the pacific relationship between the interveners and the belligerents will be key to ensure the success of the peace operations. By keeping this balance and prioritizing non-coercive means, while having the capability to do so, it may thus present the benefit of inculcating norms or standards of behavior, reinforce their effectiveness and/or solidify the commitment to them among the actors involved (Freedman 2005, Morgan 2005).7 Deterrence strategy thus has the dual asset of increasing the efficiency and ensuring the accomplishment of the mandate. Even when non-coercive means are responsible for the success of an operation’s mandate, the operation would not be efficient if it was not backed by the possibility to use coercive means. Incidentally, even though efficiency seems a function of

188 S.-M. Martin-Brûlé coercive potential, it could not ensure the accomplishment of an operation’s mandate in isolation from non-coercive means.8 A deterrence strategy further presents the advantage of focusing on the rationale of actors. According to Freedman, deterrence works best when the targets are able to act rationally and when the deterrer and the deterred are working within a sufficiently shared normative framework so that it is possible to inculcate a sense of appropriate behavior in defined situations that can be reinforced by a combination of social pressures and a sense of fair and effective punishment. (Freedman 2004: 5)

Case study Recent UN peace operations in Liberia and Sierra Leone occurred in failed states and have known mitigated success.9 Apart from the importance derived from the magnitude of both operations,10 these two cases represent a dual advantage in testing the hypothesis of a deterrence model. First, they encompass a regional aspect, absent in the case of Somalia, which allows the testing of the hypothesis by contrasting it with an operation undertaken by a regional organization or under a regional great power’s command. Second, they occurred after/during experiences in Somalia, Bosnia and Rwanda. They can thus be suspected of having been influenced by the lessons learned from these operations. To test these findings, I developed a model regrouping Liberia and Sierra Leone peace operations into four sets: peace operations led by a regional great power (peace operations of the Monitoring Group of the Economic Community of West African States – ECOMOG – but led by Nigeria), by a regional organization in collaboration with the United Nations (ECOMOG and United Table 10.1 Typology of peace operations in Liberia and Sierra Leone (1990–2005) No great power intervention

Great power intervention

Deterrence

UNAMSIL I (2000–2001)

No deterrence

UNOMIL (1993–1997) UNOMSIL (1998–1999)

UNAMSIL II (2001–2004) UNMIL (2003– ) ECOMOG (led by Nigeria) In Liberia (1990–1998) In Sierra Leone (1998–)

Notes UNAMSIL: United Nations Assistance Mission in Sierra Leone. UNOMSIL: United Nations Observer Mission in Sierra Leone. UNOMIL: United Nations Observer Mission in Liberia. UNMIL: United Nations Mission in Liberia.

Tackling the anarchy within 189 Table 10.2 Typology of peace operation’s success

Deterrence No deterrence

No great power intervention

Great power intervention

2 Partial success 4 Failure

1 Success 3 Partial failure

Nations), by the United Nations and, finally, by a great power and the United Nations (the United States in Liberia and the United Kingdom in Sierra Leone). In the case of Liberia, as of Sierra Leone, a strategy of coercion was adopted for the first two operations, a strategy of the deterrence for the latter. If all operations are, or have been, mitigated successes, authors agreed on two points: the peace operations conjointly led by the United Nations and a great power (UNAMSIL and the United Kingdom and to a certain extent UNMIL and the United States11) adopting a deterrence strategy were not only more successful than the peace operations using coercion; they were also more successful than the peace operation undertaken by the United Nations or by the regional organization also using a deterrence strategy (Durch et al. 2003). Moreover, ceteris paribus, less violent attacks and fewer confrontations occurred under these operations than throughout the three others. These two observations suggest that, as in the case of Somalia, the success of peace operations might emanate more from its coercive potential rather than from the actual use of force and that a great power intervention might be necessary for the deterrence strategy to be effective. Liberia and Sierra Leone: civil wars within failed states In Liberia as in Sierra Leone, the conflicts can be traced back prior to the colonization period (Adebajo 2003; Hirsh 2001a). However, the wars took a different path during the 1990s, following the withdrawal of the superpower support for their respective allied governments. The Liberian civil war was triggered by a coup led by Charles Taylor in 1989. By 1995, eight parties were recognized to the conflict.12 Since 1990, Liberia has known 14 peace agreements yet all but one (the last one) have failed to be implemented. By 2000, the conflict had generated more than 150,000 deaths and more than 850,000 persons had taken refuge in neighboring states (UNDPKO 2001). The conflict in Sierra Leone would not have escalated as such if it had not been for the spillover of the conflict in Liberia (Hirsh 2001a). On March 23, 1991, in an attempt to replicate his own efforts at toppling the established government, Charles Taylor encouraged elements of his factions to enter Sierra Leone (Adebajo 2003:90). In less than a year, the

190 S.-M. Martin-Brûlé group was occupying half of the country and most of the diamond-producing area. In May 1992, a group of young army officers overthrew the Sierra Leonean government. In March 1996, war-time elections were held. A peace agreement, the Abidjan Accords, was then signed between the rebel RUF and the government, but this was undermined by a coup the next year. In 2000, following a combined intervention of the UN and the United Kingdom, the government was restored with a joint declaration to end the war (Pugh et al. 2004: 93).

Partial failure: Pax Nigeriana in Liberia and Sierra Leone The Nigerian-led ECOMOG peace operations in Liberia and Sierra Leone were partial failures. In both cases, a great power, Nigeria, intervened, yet a deterrence strategy was not adopted. While the belligerents were assured of the will of the intervener to go through with the peace agreement, they did not have the assurance that their adversaries would be deterred nor were they themselves deterred from carrying out attacks. Liberia ECOMOG would not have been possible without Nigeria. The country provided 75 percent of the troops and 90 percent of the funding. Yet, to secure domestic, as well as international support, Nigeria chose to intervene with regional counterparts (Adebajo 2003: 60). ECOWAS authorized the Nigerian-led mission in May 1990. The peace operation had the mandate to make the parties comply with a lasting settlement that would result in the establishment of a democratic government in Liberia. Given the dubious democratic quality of the governments participating to the mission, the Liberian rebels remained skeptical of the “real” intentions behind the operation.13 As the mission unfolded, the belligerents’ rising aggressiveness toward civilians and members of the operation alike called for a stronger mandate. A compellence strategy was adopted (Adidbe 1997). Yet, the use of force with inadequate means generated slip-ups which overshadowed its military success. The collateral damage spurred the resentment of Liberians, so that ECOMOG became seen as another faction to combat. Although numerous troops were assigned to the mission, the battalions remained ill-prepared to confront adversaries on the field. The mandate, a mix between peacekeeping and peace enforcement, was unclear for many interveners (Adebajo 2003). Diverging visions of the goals of the peace operation increasingly hampered the efforts of ECOMOG. The interveners had established communication links with the parties to the conflict. Yet, the quality and the nature of these links undermined the operation. Privileged cooperation agreements with different parties to the conflict such as the National Patriotic Front for Liberia (NPFL) and Armed Forces of Liberia

Tackling the anarchy within 191 (AFL) reinforced the perception of ECOMOG as a warring party. For many Liberians, these links confirmed the biased interest of Nigeria in manipulating the operation as to assure its control over the region. Finally, the isolation of the regional great power from other great powers and from international organizations reinforced the belligerents’ suspicions toward Nigeria’s true intentions. The Nigerian-led ECOMOG peace operation was a partial failure. It succeeded in bringing the parties to the table to sign a new peace treaty (Cotonou Agreement). Yet, the signing of this agreement was less seen as an act of good will on the part of the belligerents than a result of “ECOMOG (having) bombed Charles Taylor to the negotiation table.”14 The event was thus seen as forcing temporary compliance. The unclear mandate and the poor communications between members of the operation and between the interveners and the belligerents hampered the effort to tackle the violence. Peacekeepers remained unsure of how to react to situations on the field and consequently belligerents were left confused about the proscribed actions and appropriate sanctions. Moreover, the use of force had propelled ECOMOG to be seen as party to the conflict. This also nourished animosity among the factions toward the proposed agreement. Furthermore, the magnitude of the Nigerian involvement, the establishment of a privileged group of discussion and the unwillingness of the international community to participate in the mission, fueled the belligerents’ and even the civilians’ perception of the biased nature of the operation (Adibe 1997). In brief, ECOMOG had largely failed to tackle violence among parties and toward civilians. Sierra Leone Despite its mitigated success in Liberia but because of the similarity of the conflict, Nigeria undertook a second peace intervention in June 1997 (Adebajo 2003). Nigeria had signed a defence agreement with the Sierra Leonean government and it, thus, considered its intervention necessary to restore the overthrown government. Prior to the coup, Nigeria had more than 900 troops in Sierra Leone. The intervention was to be undertaken using air and naval forces. At the Organization of African Unity (OAU)’s request, ECOWAS agreed to deploy ECOMOG in Sierra Leone. ECOWAS chose a tripartite strategy: dialogue, sanctions and the use of force (Hirsh 2001b: 146). As it had been the case in Liberia, Nigeria took command of the mission and chose a compellence strategy.15 ECOMOG was composed of 18 battalions and 15,000 supporting units. Of these, Nigeria accounted for 11 battalions, three naval assets and one air force detachment (Berman and Sams 2000: 118–124). Moreover, Nigeria agreed to bear the financial burden of the operation, spending more than one million dollars per day on the mission (Harman 2002). As in Liberia, the magnitude of the

192 S.-M. Martin-Brûlé involvement of the regional great power spurred suspicion among Sierra Leonean civilians and belligerents about the true intent of this operation. Sierra Leone is renowned for its diamond mines. The interests of the regional great power in the mission were deemed closely linked to the rich natural endowments of this country. ECOMOG’s robust action succeeded in bringing a sense of relative security to the country and in restoring the legitimate government in March 1998. Yet, as in the case of Liberia, the overwhelming use of force by the peacekeepers reinforced the perception of ECOMOG as being one of the belligerents and hindered the lasting implementation of the peace agreement (Berman and Sams 2000: 112). ECOMOG had initially provided credible military deterrence which allowed the start of a disarmament and demobilization process. Yet ECOMOG’s lack of familiarity with the terrain, its massive use of force, logistic problems and lack of information led the organization to become increasingly targeted by the belligerents (Hirsh 2001a). By the end of 1998, ECOMOG had failed to accomplish its mandate. Sierra Leone had become “a zone of terror, a noman’s land.” Both ECOMOG and humanitarian agencies were blocked in their operation. The government authority had collapsed (Hirsh 2001a: 75). Closely connected to the warring factions in Liberia, the Sierra Leonean belligerents had learned about the experience of ECOMOG in this latter country. Since the intervention in Sierra Leone was carried by the same organization and was led by the same regional power, the belligerents could react according to their knowledge of the strengths and weaknesses of their adversary. Aware of the interveners’ aversion for casualties,16 the RUF thus undertook the “Operation No Living Thing” with a twofold objective: to convince the Sierra Leonean population that neither their government nor ECOMOG could defend them and to deter the intervention of the regional organization. In both the Liberian and Sierra Leonean cases, the magnitude of the Nigerian involvement combined with the adoption of a compellence strategy resulted in mixed outcomes. On the one hand, the amount of troops and money invested in each mission served as indicators to the belligerents of the will and capacity of the operation to accomplish its mission. On the other hand, the manifested interest in the mission spurred doubts about its “real” aims. The importance of the intervention, the encompassing role of Nigeria combined with Nigeria’s historic trend of interventionism in the region, contributed to nourish suspicion about the real interest of the regional great power in the mission. Many started to believe that the operation was instead part of Nigeria’s grand strategy to extend its control over the region. As in Liberia, the magnitude of the involvement of the regional great power spurred suspicion among Sierra Leonean civilians and belligerents about the true intent of this operation. The clarity of the objectives of the mission was also hampered by the lack of communication

Tackling the anarchy within 193 between the interveners. This also contributed to the mixed signals to both the population and the belligerents regarding the means and the procedures of the operation (nebulosity between peacekeeping and peace enforcement), and hence, about the means and expectations of the interveners. In both peace operations, the overwhelming use of force resulted in the perception that the interveners were party to the conflict, thus hindering both the procedures of the operation and the accomplishment of the mandate. Finally, in both cases, the mandate failed to be accomplished despite the perseverance of the intervenors.

Failure: no deterrence/no intervention of a great power: UNOMIL and UNOMSIL In Liberia as in Sierra Leone, the UN collaborated with the regional intervener and succeeded in giving to the operation an appearance of greater impartiality. However, the partnership and the need for collaboration complicated the communication between the interveners and, as a result, between the interveners and the belligerents. The difference of intervening cultures, of understanding of local dynamics, combined to hinder the articulation of the joint operations discourse and actions. Moreover, after the failure of peace operations in Somalia and Bosnia, the belligerents had learned that casualties could cause interveners to withdraw hastily from an operation. The continuing use of force by the regional organization gave the belligerents the ideal pretext not only to forcefully reply to the coercive actions but to target UN personnel. Revenge toward ECOMOG and discouraging the targeted international personnel worked well for the belligerents. Moreover, that ECOMOG proved incapable of protecting the personnel was seen as an opportunity for the belligerents to discredit the regional organization in the eyes of the international community. By reacting robustly to forceful actions, belligerents gained credibility on the international scene. On the interveners’ side, compellence had deteriorated the situation on the field and a peace agreement seemed even farther away. The failure of both operations stems from the fact that no adversaries were deterred from engaging into violence, and no adversaries were reassured that the agreement would be respected. Liberia: ECOMOG and UNOMIL Created on September 22, 1993, UNOMIL was the first UN peacekeeping mission undertaken in cooperation with a peacekeeping mission set up by another organization (Adibe 1997). The joint peace operation was mandated to implement the disarmament plan of the Cotonou Agreement. ECOMOG was assigned to conduct the main duties and the UN observers were to supervise the process (Olonisakin 2003). ECOMOG

194 S.-M. Martin-Brûlé was allowed the use of coercive force in specific conditions and upon authorization of the UNOMIL mission. UNOMIL was restricted to noncoercive means. The command structure remained blurred for most interveners. As for the strength of the operation, 12,000 to 18,000 were deemed necessary to tackle the situation. Yet, by 1995, ECOMOG had a mere 8,430 troops in Liberia (of which more than 4,900 were Nigerians) and UNOMIL was solely composed of 86 military observers (Adebajo 2003).17 The success of UNOMIL was conceived as intimately linked to the application of the Cotonou Agreement. The collaboration of ECOMOG with the UN furthered the legitimacy of the enterprise. However, the scarce number of peacekeepers, the frictions between ECOMOG and UNOMIL, as well as the massive use of force which principally targeted the opponents to Charles Taylor, sent mixed signals to the belligerents as to the impartiality of the mission. While the joint operation led to the signing of a supplement to the Cotonou Accord (Akosombo Agreement), the fighting endured. Moreover, as the accord was being signed, two of its three signatories were expelled from their headquarters by rival factions. Interveners soon realized that Akosombo had become more a way for the signatories to increase their military strength than to renounce to its use. Once again the agreement was “an illusion.” By September 1994, in-fighting areas covered 80 percent of the country. The rising insecurity reached such an extent that the Security Council amended UNOMIL’s mandate to decrease the size of the mission (Olonisakin 2003). As for ECOMOG, the West African peacekeeping force left the country in 1998. The non-use of force by the UN and its disengagement signaled the impact of the warlords’ actions on the international community. In the belligerents’ perspective, the use of force by ECOMOG confirmed their partiality in the conflict. ECOMOG had accomplished little of the tasks they had been assigned. Sierra Leone: ECOMOG and UNOMSIL On July 1998, the UN established the United Nations Observer Mission in Sierra Leone (UNOMSIL) to support for the actions of ECOMOG in Sierra Leone. It was meant to demonstrate international support for the return of the Government in Sierra Leone. Yet, as it had been the case in Liberia, the cooperation between ECOMOG and UNOMSIL was never optimal.18 The disparity of the two organizations’ resources and the imbalance between their roles on the field heightened tension among the interveners. UNOMSIL was mandated to monitor the military, the security situation in the country and the disarmament and demobilization of former combatants. It was also to assist ECOMOG in the provision of security and in the collection and destruction of arms, to help monitor

Tackling the anarchy within 195 international humanitarian law, as well as the voluntary disarming and demobilizing of the members of the Civil Defence force (UNSC resolution 1181). ECOMOG had fewer resources than the international mission, yet it was assigned the crux of the operation. It was mandated under the authority of Chapter 7 to establish security in the country by flushing out the remnants of the AFRC/RUF, to conduct disarmament and demobilization and to protect the UN personnel. The difference in the risk, tasks and disparities of pay between the regional and international staff continuously created tensions between the two organizations (IPA 2001). With regards to the belligerents, ECOMOG emulated the strategy adopted in Liberia, using compellent force (UNDPKO 2003). Once again, ECOMOG’s robust actions had mixed results. On the one hand, they brought a sense of relative security to the country and allowed the restoration of the legitimate government in March 1998. ECOMOG’s troops at first provided a credible military deterrent which abated violence and which allowed for starting the disarmament and demobilization process in the country. Yet, the systematic use of force, not only for signaling resolve, but, as a means of negotiation, drew ECOMOG to be considered once again a party to the conflict and hence, as a potential target of attacks by the remnants of the junta. This jeopardized the safety of unarmed UN personnel under its protection. The UN withdrew its personnel. This withdrawal hinted to the belligerents which strategy to adopt. If massive casualties could be incurred among Nigerian interveners, the great power would be forced to withdraw its remaining troops. The unprepared and ill-equipped UN force would then not be able to disarm them. The belligerents made clear they would attack the Nigerians if they continued on the field. The belligerents also knew that upcoming elections in Nigeria would render more difficult the sustainability of troops in what was being called “Nigeria’s Vietnam” (Doyle 1999). They demonstrated their resolve by taking several Nigerians as prisoners of war. As ECOMOG took heavy casualties in the fierce fighting, Nigeria indeed decided to withdraw from ECOMOG in January 1999 (UNDPKO 2003). Violence resumed and Sierra Leone was left once again in a civil war. By June 1999, the UN had decreased their troops on the field to a mere 24 observers. As in the case of Liberia, the joint operation and its resulting mixed signaling had not succeeded deterring the belligerents to spoil the peace agreement or to abate the violence. The belligerents had seen more opportunity and more security in violating the accords than in sticking with them.

Partial success: deterrence/no intervention of a great power Both UN operations in Liberia and Sierra Leone accomplished partial success. The singular operation (once ECOMOG retreated from the field) allowed for an easier coordination of the activities on the ground and

196 S.-M. Martin-Brûlé minimized the problems of communication both within the members of the operation and with the belligerents. However, the unwillingness of great powers to commit themselves to the operations spread doubt regarding the resolve of the participants to stay the course of the operation. This opened some room to maneuver for the belligerents. The two cases seemed to be in fact a case of a deterrence strategy being adopted by the warring factions instead of being adopted by the leaders of the peace operation. Liberia ECOMOG’s troops were out of Liberia by 1998. However, the mounting violence in the country in early 2003 led to a renewed intervention. On August 1, 2003, the Security Council unanimously adopted Resolution 1497. It assigned the ECOWAS’ Monitoring Group in Liberia (ECOMIL) a robust mandate under Chapter 7 of the UN Charter in order “to ensure that it has a credible deterrence capability” (Quist-Arcton 2003). Once again a Nigerian-led multinational force would support the June 17 ceasefire and establish the initial conditions for disarmament, demobilization, reintegration and repatriation of ex-combatants. Furthermore, ECOMIL was mandated to help with the establishment and maintenance of security to prepare for the taking-over by the new government and to prepare for the deployment of a UN peacekeeping force by October 1.19 ECOMIL this time received the support of a great power, notably the United States which provided logistical support (ICG 2003c).20 By the end of August the peace operation had about 3,500 men in Liberia, and the leaders of ECOMIL had sent several warnings to the rebels who continued to clash with government militias in remote villages, to stop violating the August 18 peace agreement (IRIN News 2003c). The Nigerian-led peace operation successfully disarmed most rebels in the area it controlled (Monrovia) and did not suffer casualties. Looting and violations, however, continued in the rural regions. ECOMIL’s mission was annexed to the new United Nations Mission in Liberia in September 2003. Sierra Leone: UNAMSIL: Take 1 Faced with the growing insecurity in Sierra Leone in October 1999, the UN decided to replace the Observer Mission in Sierra Leone with a larger operation. The United Nations Assistance Mission in Sierra Leone was first assigned a maximum of 6,000 military personnel (UNDPKO 2001). In February 2000, mounting violence on the field convinced the Security Council to increase its strengths to the, yet unmatched, deployed UN force of 17,000 troops. Resolution 1289 authorized troops acting under Chapter 7 to provide security at key locations (government buildings, airports, etc.), to facilitate the free flow of people, goods and humanitarian assistance along specified thoroughfares. It was to provide security at

Tackling the anarchy within 197 disarmament, demobilization and reintegration sites and to guard weapons, ammunition and other military equipment collected from excombatants and to assist in their subsequent disposal or destruction. UNAMSIL was first conceived to support the mission of ECOMOG. However, as the West African forces retreated, the UNAMSIL mandate was enlarged. After the failures of peacekeeping missions in Somalia, Rwanda and Angola, the stakes in the success of this operation bypassed the Sierra Leonean situation. The SRSG had his reputation at stake and so had the system of peacekeeping as a whole (Denselow 2000; Harman 2002). UNAMSIL was to become a (successful) symbolic mission. Investments were bolstered to $700 million per year (Connaughton 2001). A deterrence stance was adopted: notwithstanding its strong force on the field, the decision makers were determined to privilege non-compelling means. Various mechanisms of communication were established between the interveners and the belligerents. Decision makers wished to make clear that it carried a “robust” Chapter 7 which allowed the mission to “take necessary action to ensure the security and freedom of movement of its personnel and to afford protection to civilians under imminent threat of physical violence.” (Radio Netherland 2004). Furthermore, the SRSG highlighted that contravention of the Lomé Peace Accord, such as the seizure of weapons from UN peacekeepers, mounting illegal road blocks and blocking the free movement of UN troops in the discharge of their mandate, must stop immediately or would “invite forceful response.” (Radio Netherlands 2004) Notwithstanding its Chapter 7 authority, its adoption of a deterrence strategy, its unmatched capability and the variety, despite the quality of the communication means established, UNAMSIL met with mixed results. First, the re-hatting of certain ECOMOG troops blurred the distinction between the types of peacekeepers. The belligerents considered the ECOMOG forces as part of the conflict. Incapable of discerning re-hatted ECOMOG from UN troops, the aggressive behavior of belligerents was generalized to all troops. Second, communication between the interveners hindered the efficiency of the mission. These difficulties were both on the field and on the international level: the rehatted peacekeepers had difficulty adapting quickly to their new mandates and had trouble understanding their new command structure; developed countries were unwilling to contribute large numbers of troops for UNAMSIL. This reticence hinted to the belligerents the fragility of the commitment of interveners to their mission. Moreover, in the light of the past failures of UN peace operations, the warring parties had learned how to destabilize the mission.21 The display of overwhelming force and the communication undertaken by the leaders of the mission initially succeeded to re-establish a seemly

198 S.-M. Martin-Brûlé order in the country. However, violence resumed as the belligerents realized the lack of unified command and the unwillingness of great powers to commit themselves to the mission. The failure of the deterrence strategy peaked in May 2000 as the belligerents succeeded in disarming 500 peacekeepers and in kidnapping ten British peacekeepers. In the absence of great power commitment, deterrence failed.

Success: deterrence and great powers: UNAMSIL II plus UNMIL The interventions of great powers in each operation sent clear signals to the belligerents of the determination of the international community to resolve the conflict. Knowing the stakes of the great powers in their involvement in such peace operations and given the magnitude of the authority and the capabilities allowed to both operations, the belligerents were soon convinced to respect the prohibited actions. The prioritization of non-coercive means albeit the robust strength of both missions combined with the signal sent by the magnitude of the involvement of a great power to each operation also confirmed the determination of international actors to accomplish the mandate of the mission. The rapidity with which the situation has improved in both Sierra Leone and Liberia after the intervention of the great powers seems to confirm the initial hypothesis. Although challenges remain, UNMIL can be qualified as a success. The expedition with which the operation has convinced the most rebellious faction to disarm and to start negotiations, as well as the swiftness of its deployment in the rural areas (something that ECOMOG struggled to do for seven years and which was accomplished in less than six months by the present operation) call for an acknowledgment of a great progress in the accomplishment of the mandate as well as the grand efficiency of the mission. As for Sierra Leone, the success of the last operation can be judged from the fact that the government was restored and that two elections were able to take place following the intervention of the United Kingdom and the application of a deterrence strategy. Sierra Leone: UNAMSIL: take II22 The United Kingdom sponsored the modification of the UNAMSIL mandate on August 4, 2000. UNAMSIL would maintain the security of the Lungi and Freetown peninsulas, and their major approach routes and “deter and, where necessary, decisively counter the threat of RUF attack by responding robustly to any hostile actions or threat of imminent and direct use of force” (UNSC resolution 1313). As it has been the case for the Operation Restore Hope in 1992, deterrence strategy was officially part of the mandate. The presence of the United Kingdom strengthened the deterring character of UNAMSIL. While

Tackling the anarchy within 199 the UN was negotiating the release of its detained personnel, the UK Government took bold unilateral action and rapidly deployed troops to Sierra Leone. Indeed, when the West Side Boys23 kidnapped the British contingent in August 2000, Great Britain responded in such a forceful way which had not been matched since the Gulf War. This served to signal to the belligerents not only the capability but the resolve of the United Kingdom to use great means in order to achieve their ends.24 The troops’ initial objective was to secure the airport and allow for the evacuation of UK nationals. However, they not only ended providing security in Freetown, they engaged in a conspicuous rescue of workmates captured by rogue elements of the Sierra Leone military (inflicting high casualties on their adversaries in the process), and deployed a number of Short-Term Training Teams to speed the process of rehabilitating the Sierra Leone Army. This intervention also came with the promise of “over-the-horizon” support, meaning that the UK forces and military equipment would again be deployed in support of the government of Sierra Leone.25 According to Bruce Jones, “notwithstanding the multiple coordination efforts of the UN, a significant – if unofficial – coordinating agent in Sierra Leone was the British government. Sierra Leone became, in effect, a case of ‘lead state coordination,’ though this was never officially recognized” (2001b). The United Kingdom had many interests in the success of the mission. The operation in Sierra Leone is said to being viewed in London as a test case for the United Kingom’s post-Cold War capabilities (Human Rights Watch 2005). Hence, the United Kingdom took on a number of important coordinating and leading functions. It chaired a donor forum to invigorate international financial support; it took the lead in financing ECOMOG’s mission when ECOMOG re-intervened in Sierra Leone (ECOMIL), it funded demobilization when there was a shortfall; and, it ensured a degree of consistency, forward planning and sustained international commitment to Sierra Leone (Jones 2001b). Despite the rebel forces hostile attitude toward the UN contingents and despite the mission’s authority and capability to use force, UNAMSIL strived to keep open the lines of communication with the civilians and the rebels, hence privileging non-coercive approach. A media campaign was launched highlighting efforts being made and challenges encountered by the mission and “contact groups” consisting of UNAMSIL military and civilian representatives and the most important rebels, the Revolutionary United Front, were established. UNAMSIL troops adjusted their deployment to dispense effective security for the electoral process. The mission also provided public information support to the electoral committee; Radio UNAMSIL broadcasted information on the voter registration process and other educational resources for the electoral process. UNAMSIL’s size and good public image gave the force legitimacy and credibility both to offer effective deterrence (UNA 2002).

200 S.-M. Martin-Brûlé If the commitment of overwhelming force had punctually deterred the belligerents during the previous operation, the intervention of the United Kingdom sent a clear signal to the warring parties of the resolve of the international community to end the fighting. The deterrence strategy adopted by the United Kingdom, the clear communication of the operation’s aims and mandate, of its capacity of actions yet of its will to negotiate, significantly reduced the violence. With fewer casualties and less confrontation with the warring parties, the two ingredients combined (deterrence strategy and great power intervention), had succeeded in the accomplishment in two years of what the ECOMOG and previous peacekeeping operation had been struggling to do for four years. From the low point of May 2000 to 2002, calm elections had been held, the government succeeded in regaining control of the country. Over 50,000 combatants had disarmed and had begun rehabilitation programs. More than 80,000 refugees had returned and 10,000 internally displaced persons had regained their homes (Harman 2002). Liberia The United Nations Mission in Liberia (UNMIL) was deployed following resolution 1508 on September 19, 2003 and is still effective today. Authorized under the authority of the Chapter 7 of the UN’s Charter, it has the encompassing mandate: to observe and monitor the implementation of the ceasefire; to investigate violations; to observe and monitor disengagement and cantonment of military forces of all the parties; to support the work of the Joint Monitoring Committee (JMC); to develop an action plan for the overall implementation of a disarmament, demobilization, reintegration and repatriation (DDRR) program for all armed parties; to facilitate the provision of humanitarian assistance, including by helping to establish the necessary security conditions; to contribute toward international efforts to protect and promote human rights in Liberia (UNDPKO 2004). The total authorized strength of 15,000 military personnel and the proposed budget of $564.61 million made it the largest peace operation in the world at that time (UNDPKO 2004).26 The United States, its troops already overstretched in Afghanistan and Iraq, chose to commit itself monetarily to the mission. The superpower invested more than $200 million in humanitarian and reconstruction aid and another $245 million to the operation making it the main broker of the operation (Dukuly 2004; Reilly 2004). To stress the level of involvement of the great power in the operation, an American, Jacques Klein, had been assigned as the Special Representative of the UN Secretary-General (SRSG). The refusal of the United States to commit troops to the mission has been highly debated. Many attributed to this refusal several setbacks suffered by the operation. The operation took its most serious blow in

Tackling the anarchy within 201 December 2003, when the SRSG started the disarmament process prior to the complete deployment of the mission’s troops. The pressure to rapidly spend the money allocated by the great power (in fear of it being relocated in other much needed theater) forced these premature actions which, in turn, hampered the operation (ICG 2004: 4). This financial contribution has thus seemed to create obstacles that would not be developed if American troops were involved in the field.27 The communication established with Liberia’s civilians and rebel leaders has met success. Radio, newspaper advertising, posters, “traveling song and dance revues,” public support from senior officials from each of the three warring factions has facilitated the unfolding of the mission (Gelfand 2004). Fighters are said to retain a degree of respect, especially for the military, and a recent report concluded that ordinary Liberians like UNMIL, seeing the SRSG as “their advocate” (ICG 2004: 6). The prioritization of non-coercive combined with overwhelming forces and encompassing authority make the strategy a deterrence one as opposed to a compellent one. In effect, peacekeepers have spent much energy communicating with the belligerents, signalling the intent of the operation, and the potential sanctions. Since 2004, the situation in Liberia has grandly improved. A credible election process was accomplished with the inauguration of Ellen JohnsonSirleaf as president on January 16, 2006 (ICG 2006). Indeed, UNMIL can be qualified as a success, the quick progress of the operation, the discrepancy between its robust capacity and its meagre use of coercive force, as well as the small number of casualties, all hint to the positive effects of the deterrence strategy.28 As UNMIL continues to unfold, another track of study may be suggested regarding the impact the relation between the great power’s type of intervention (ex. commitment of troops on the ground versus financial support) and the success of the peace operations.

Conclusion This chapter sought to explain if and how peacekeepers can restore monopoly of violence in failed states coming out of civil wars. Our observations revealed that deterrence was necessary to restore order and that the commitment of a great power was decisive for its preservation. The adoption of a deterrence strategy resulted in a decrease of violence, and the involvement of a great power allowed for the mechanisms of conflict resolution to take root. A great power could succeed in this process by being committed yet, the strategy of deterrence rendered this effort more efficient. Similarly, the strategy of deterrence was useful in limiting violence and in preventing prohibited actions. However, without the intervention of a great power, the chances of lasting compliance by belligerents decreased. This chapter’s findings are thus twofold. An examination of the first two peace operations highlighted that the use of force is not in itself very

202 S.-M. Martin-Brûlé effective and may in fact slow the peace process by making the interveners party to the conflict. Similarly, and as the UN’s missions success demonstrated, the lack of coercive means also hampered the peace operations. This phenomenon can be partly explained by the fact that interveners lose or simply do not gain credibility in the eyes of the belligerents. Hence, a low profile intervention may not only fail to stop violence, it might in fact trigger violence. The perceived “benign” nature of the operation becomes an incentive for belligerents to target international organizations’ resources and workers/peacekeepers, hence increasing insecurity for both civilians and for the intervening troops (Lischer 2003; Anderson 1999). In brief, the use of a deterrence strategy may provide immediate success to the peace operation by decreasing the level of violence, but without the commitment of a great power, the deterrent potential is weak, and so is the possibility of long-term success. The last case studies confirm the hypothesis, yet suggests avenues for greater research. The commitment of the United Kingdom combined with the deterrent strategy adopted by UNAMSIL was revealed to be a clear success. In less than two years, civil war ended, belligerents complied with the parameters of the peace agreement and free elections were held. Even though the operation is still underway, UNMIL can also be qualified as a success. However, the failure of the United States to commit troops on the ground, the disappearance of arms, the continuing and threatening struggle in the Ivory Coast all call for prudence for the long-term assessment. Such clues also point to the need of further research on the type of commitment (either financial or troops commitment) of a great power and its effect on the efficiency of an operation. It points also to a revised vision of the regional organizations’ input. These organizations can prove to be effective because of the stakes they are seen to have in the conflict, in their “unique” understanding of the ongoing disputes, of the proximity of their troops and their willingness to commit themselves to the operation. However, their impartiality and their capability on the ground should not be overestimated. This may even call for the need for more great power intervention in these conflicts, so the time of great power retirement has not yet arrived. Great powers seem to retain this unique capacity and credibility. Deterrence, while costly in terms of mobilization, combined with a great power involvement, may in fact be saving time and money for the great power in intervening in these conflicts.

Notes 1 Once it commits itself to an operation, it is precisely because the great power is more likely to be casualty averse that it will tend to bolster its investment in the mission, giving its troops the means to minimize casualties and hence being the more deterring to belligerents (Hillen 1998). The belligerents may know that great powers are casualty averse and hence know that they will be the most sensitive to the insecurity of the setting, a sensitivity that lowers the threshold

Tackling the anarchy within 203

2

3 4

5 6 7 8

9 10

11

12

13

of tolerance regarding the loss of soldiers lives, and hence presage the bolstering of their equipment and the aggressiveness of their retaliatory attacks. The Kimberley Process Certification Scheme has been implemented in January 2003. Its objective is to guarantee that parcels of rough diamonds can be tracked from their primary source of origin without confusing specific parcels with the country of provenance (Orogun 2004). S.-M. Martin-Brûlé, 2003, “Le rôle de la dissuasion dans les opérations d’assistance humanitaire: le cas de la Somalie 1992–1995,” Master’s Thesis, Université de Montréal. Even though I will address deterrence as a distinctive element, given its presence and importance in all the identified “ingredients” of a deterrence strategy (capacity, resolve, interests, etc.) it will be conceived as a transversal component. Moreover, indirect communication, which sends “cues” to the adversary with regards to the operation’s aims and resolve, will also be taken into account. Even though these means will not be directly addressed in this chapter, they shall be in a subsequent and more exhaustive research. I will refer to such things as the proximity of the Blue Helmets’ headquarters from the belligerents, the number of scheduled/planned meetings with the various parties and among the interveners, the type, number and quality of media on the ground, the frequency of their transmission of information, the pedigree of the operations appointed leaders, etc. The case of Israel’s nuclear capacity is an example of such deterring ambiguity. In this chapter, international community will be understood as the ensemble of states and international actors (organizations, institutions) on the international scene. Morgan (2005: 760), however, notes that to work properly deterrence must come to be seen as employed in support of norms that are widely regarded as legitimate; it must be embedded in a larger framework of norms. A more thorough analysis of deterring interactions between interveners and belligerents would necessitate further taking into account of the “belligerent variables” (ex. might some belligerents or some actions be non-deterrable?). These questions will be further developed in a future research. Efficiency is understood as accomplishing much with an economy of means. Unlike efficacy, efficiency introduces a time factor to the equation of output/input. The budget of the last mission in Sierra Leone was of US$543.49 million, of the last mission in Liberia: $564.61 million, encompassing more than 15,000 troops each, making them successively the largest peace operations of the last four years. See, United Nations Peacekeeping Website, Available on line at, www.un.org/Depts/dpko/dpko/home.shtml (accessed January 15, 2004). UNMIL was undertaken in October 2003. The United States refused to lead the operation, but nonetheless committed itself to be the largest financial contributor to the operation with an investment of more than $US200 million. Moreover, an American, Jacques Klein, was appointed as the Special Representative of the UN Secretary-General for this operation. Regarding the operation’s success, the efficient progress of the operation has lead several authors to conclude to a success. The signatories of the Abuja Agreement in 1995 were: the Armed Forces of Liberia, the National Patriotic Front of Liberia, the Central Revolutionary Council, the Liberian Peace Council, the Lofa Defense Force, the Liberia National Conference, the United Liberation Movement for Democracy in Liberia (ULIMO) comprising two divisions: ULIMO-K and ULIMO-J. Adebajo (2003) highlights that among the intervening states to establish democracy in Liberia, only Gambia could be considered democratic.

204 S.-M. Martin-Brûlé 14 According to Adebajo, “it is no coincidence that the agreement was signed barely two months after ECOMOG’s capture of the key port in Buchanan and Monrovia” (Adebajo 2003:127). 15 The overwhelming use of force was, however, considered problematic. It was reported that “Several diplomats in the 15-nation Security Council deplored Nigeria’s heavy shelling of the capital, and wondered what role the troops would play now. An envoy complained that “ECOMOG’s existing mandate did not allow for the use of force on this scale” (Haq 1998). 16 The West African Countries removed their troops from Liberia following massacres of some of their peacekeepers. 17 ECOMOG was unable to deploy its troops at major points along Liberia’s border. It thus remained incapable of abating the smuggling of arms. 18 Cooperation between ECOMOG and UNOMSIL was hampered by the lack of a proper mechanism for liaison and coordination, which was never put in place. In addition, differences concerning the relative status of military officers, assignment of specific tasks, ways of conducting military operations, all gave rise to unhealthy comparisons and were a source of tension (Adebajo 2003:125). 19 www.state.gov/documents/organization/38116.pdf. 20 The United States positioned a task force of over 2,000 marines off the coast of Liberia and committed $26 million to transport all the contingents and to fund contracted logistics support and equipment for ECOMIL (Kansteiner 2003; UNSC 2003). 21 Richard Connaughton thus reports the comments of a Freetown businessman to the effect that “The United Nations is not very good at using force (. . .) and, believe me, the RUF knows that. That’s why the UN can’t win” (Connaughton 2001:252). 22 The last operation to have occurred in Sierra Leone started before the one in Liberia. To respect the chronological order of my description, I will change the sequence of the case study accordingly. 23 The West Side Boys are described as a maverick group which is not connected to the main rebel group in Sierra Leone, the Revolutionary United Front (BBC News August 2000). 24 See McGreal and MacAskill 2000 and Smith 2000. 25 “A Review of Peace Operations: A Case for Change,” King’s College London, http://ipi.sspp.kcl.ac.uk/rep004/intro.html. 26 Today, 16,065 total uniformed personnel, including 14,832 troops and 205 military observers; 1,028 police supported by 549 international civilian personnel, 844 local staff and 242 UN Volunteers make up the mission (UNDPKO 2006). 27 The presence of American troops in Iraq and Afghanistan would cause the arm forces to be overstretched. Yet, because of its long time link with Liberia and some would say, because of its relative responsibility in the conflict, many would have expected a greater involvement of the United States in this conflict. 28 One could highlight war weariness as an alternative explanation to the success of deterrence strategy. It could however be suggested that the adoption of a deterrence strategy might have indeed accentuated/made the most of this situation of war weariness. The adoption of such strategy would thus be responsible for the punctual success of the operation.

11 Conclusion Dilemmas of insecurity: implications for research and policy Marie-Joëlle Zahar and Stephen M. Saideman

This project started with a fundamental intuition: that all states face the problem of providing security to their populations. This basic task of governments is the underlying thread which brings together functioning societies, failed states, and all polities in between. Indeed, the fundamental empirical difference between failed and failing societies is the extent to which governments succeed at the delicate balance of deterrence and assurance. In reframing security not so much as a good but as an ongoing bargain between states and groups in society, the book sought to cast a different light on the role of security in civil wars. Instead of framing the insecurity of war as a function of anarchy, brought about by the collapse of the state, contributors used the insights of deterrence theory to probe the delicate balance between the need to assure and the impulse to deter. In so doing, we ask a different set of questions of our cases, one that does not neatly set functioning societies apart from conflict-ridden polities but seeks to examine the conditions conductive to the achievement of equilibrium between the conditions of successful deterrence and the requirements of assurance. This concluding chapter draws on the findings of contributors to sketch some potential answers to the questions guiding this research. It not only charts future research directions but also draws policy implications from the insights generated in the previous chapters.

Challenging the security dilemma The project sought to question the adequacy of the security dilemma as a theoretical construct to think about the dynamics of civil war and ethnic conflict. Our concern was based on two observations, none of which fit the expectations of the theory: (1) that intermixed groups are less likely to fight, and (2) that states are one of the key protagonists in a majority of internal conflicts. The volume provides empirical support for these observations further casting doubt on the utility of the approach.

206 M.-J. Zahar and S.M. Saideman Questioning the fit between theory and empirics Security dilemmas make a number of assumptions about the patterns of violence among groups. Several quantitative studies have questioned their take on demographics. In his contribution to the volume, Stathis Kalyvas goes one step further and casts doubt on the expectations of the theory in terms of mechanisms and outcomes. Using micro-level data from the Greek civil war and from the Vietnam conflict, Kalyvas finds that violence is not typically launched by threatened minorities but by confident and aggressive majorities (a finding echoed in the contribution of Davenport and Armstrong). He demonstrates further that extreme insecurity is likely to act as a deterrent to violence as relative power equilibria under anarchy do not necessarily trigger escalatory spirals but are often associated with local peace deals. Most importantly, he asserts that while weakness at the national level might permit violence, the events at the local level may be driven less by an overarching cleavage and more by personal differences and micro-level dynamics. Such findings echo, among others, empirical reports from Afghanistan. Spikes in Taliban violence over the year 2006 and the beginning of 2007 have been linked to a surge in Taliban strength and support among the population, leading to the suggestion that negotiations might be more effective in subduing the insurgency than military power and the hope of an elusive military victory. The negotiation of local deals, notably in Helmand province between local strongmen and the British contingent, while currently an exception rather than the rule, could provide additional evidence in support of these preliminary findings. Equally telling, in this respect, are developments in Iraq where US officials have gone from attempting to militarily subdue insurgents to meeting with them in an attempt to bring them back into politics.1 Strikingly, the US has taken advantage of intra-Sunni differences to gain some support in previously hostile regions. Questioning the connection between states and security Security dilemmas focus on the security of groups left to fend for themselves in an anarchic environment. They portray ethnic conflict as primarily between groups rather than between a group and a government. They are also premised on the assumption that states are the primary agent of order and stability. Empirical evidence provided by several of the contributions to this volume turns this assumption on its head. A basic finding of the volume, confirming the intuitions of research on genocide, is that the state is the key threat to most groups and individuals. As Aydin and Gates remind us, unfettered state power was already foreseen as problematic by Hobbes. He allowed that citizens could disobey the ruler should he fail to meet the basic requirement of the social contract, protection against the

Conclusion 207 state of nature. Christian Davenport and David A. Armstrong II’s chapter provides empirical evidence in support of this intuition. States not only abuse their monopoly of violence, but, in so doing, they are more threatening to citizens than the dangers of war. According to Davenport and Armstrong’s research, politicide/genocide creates eight times more casualties than civil/interstate war. Though this volume did not specifically deal with the broader notion of human security, its findings are consonant with the arguments raised by promoters of the human security concept. An exclusive focus on the concept of national security blinds us to the reality that states are often the gravest source of threats to their own citizens. New perspectives, new methods Contributors to this volume provide strong empirical evidence in support of criticisms of the security dilemma’s utility to understand the dynamics of conflict in civil wars. Not only does this suggest the need to modify our theoretical lenses, it also has implications for the methods and research designs that we use in support of our theoretical claims. Reviewing the debate on conflict fatalities, Davenport and Armstrong demonstrate that our focus on interstate wars, civil war and terrorism as the loci of violence is directly related to a selective attentiveness to actions that challenge political authorities and to its direct corollary, the systematic exclusion of the victims of state action from fatality counts.2 The stark scale of the difference between the order of magnitude of deaths caused by politicides/genocides and casualties from civil/interstate war militates in favor of increased awareness of the manner in which our presuppositions affect research designs and selection criteria. Most research on civil wars has taken insurgents as its point of departure; in counter-distinction, contributors to this book find empirical support for the need to focus on the state instead because it, not its opponents, is the primary locus of violence in civil war situations. Kalyvas’s research, on the other hand, alerts us to the dangers of reading local patterns of violence through overarching theoretical (or national) narratives and, in so doing, distorting the facts to fit our categories of understanding. Looking at conflict-ridden societies like Iraq and Afghanistan, his contribution alerts us to the danger of ignoring the manifold local logics of violence. Iraq is a clear instance of a society where the breakdown of law and order has served as a cover for many to pursue local agendas. The danger in reading their violence as part of the “insurgency” is double: first, this stands to give political legitimacy and cover to acts of revenge, looting, and to the violence of personal/tribal struggles; in so doing, another danger emerges, that concerned outsiders will have difficulty sorting out relevant interlocutors in any effort to bring a negotiated end to the cycle of violence engulfing the country. Taken together, these two contributions provide empirical support for

208 M.-J. Zahar and S.M. Saideman the theoretical intuitions of the introductory chapter: when and under what conditions does the state generate violence and how can it assure citizens that it is no threat to them?

States as threats: the mechanisms of insecurity The chapters have identified two main mechanisms of insecurity. They are state capture and state weakness. These hinder the state’s ability to provide assurances to groups in society and generate violence – though the violence can emanate from the state as well as from regime opponents. State capture The findings of Aydin and Gates support the approach here framing security as an assurance/deterrence game. They find that state capture triggers doubts about the state’s ability (even willingness) to protect and deter, leading to efforts to either control the government or secede. State capture increases rulers’ access to political power and their control over the instruments of violence. By decreasing constraints and thus costs related to the use of force, the capture of the state may facilitate the adoption of deterrent solutions but at the cost of less credible restraints, and, thus, less assurance. Likewise, Petersen and Staniland establish that ethnically dominated militaries will decrease the state’s ability to assure a multiethnic society. Where the state, the military and politically salient groups are all aligned, the state’s ability to project force increases dramatically and the military becomes a catalyst for violence. These findings have direct implications for the way in which Afghanistan, Iraq and other postwar states are rebuilt. Rather than thinking about power sharing (see below), we need to devise ways in which power is divided3 so that no one group or agency can threaten the society, while the government still retains the capacity to deal with those that do not buy in – irredeemable spoilers. This is difficult, but the contributors reveal some clues, which we consider further below. How to create functioning states, inclusive enough to prevent state capture and promote the assurance of all actors, yet strong enough to be able to deter would-be spoilers, should they arise, is the dilemma that bedevils foreign interveners in all post-conflict situations. This dilemma requires a fine-grained understanding of the motivations of regime opponents as well as of their decision processes in as much as these affect a group’s ability and willingness to resort to violence and, by extension, the regime and outsiders’ abilities to devise strategies to deter and assure such groups. Targeting is key for both deterrence and assurance policies. Indeed, much has and is being written in the post-conflict literature to enhance our knowledge of opponents and of the strategies most likely to involve them in peace processes such that these will be

Conclusion 209 sustainable (Stedman 1997; Zahar 2005b, 2006; Greenhill and Major 2007). State weakness State capacity is central to a government’s ability to assure and deter. Weak states tend to have a poor record of control over the monopoly of violence (Herbst 2000). They are also associated with inferior institutions of conflict-management and greater short-termism both of which affect the capacity to assure. In his contribution, Murshed draws a connection between the existence of horizontal inequalities, the weakness of the state, and the emergence of a political security dilemma. Indeed, in the absence of a social contract, states will more likely privilege deterrence at the expense of assurance. Weak states are also prey to capture. Frequently, in the developing world, elites seek to maintain themselves in power and elicit support by implementing policies that discriminate in favor of their power base. This, in part, is a result of their inability to provide equally for all. It is also the result of growing international inequalities bringing horizontal inequalities into sharper focus. In conditions where populations press for demands but rulers cannot meet these, insecure elites will attempt to maintain themselves in power by further strengthening their limited power bases. This, however, further increases the state’s inability to assure its opponents as economics become fused with politics, generating power and wealth for the few and resulting in the exclusion of the many. Moreover, weak states are also characterized by inefficient institutions of commitment further reducing the chances that they will be able to avert violence by offering opponents a credible commitment to change course.4 Both state capture and state weakness result in either actual or perceived discrimination which, as Akbaba, James and Taydas demonstrate in their chapter, not only spreads insecurity and fear but provides a potent mechanism for mobilization. Discrimination fills the gap in the existing literature on conflict resolution in its three dominant strands – greed, predation, and insurgency. It provides clues as to the dynamic whereby group members provide support to their elites. According to Murshed’s economic analysis, horizontal inequalities can be used to overcome collective action problems. Elites will likely succeed in eliciting support from populations in situations of discrimination in public spending and taxation, under conditions of high asset inequality (Wood 2000) and in instances of economic mismanagement and recession.

Constraints on the use of force What, if any, institutions act as constraints on a government’s willingness to privilege deterrence and encourage the pursuit of assurance strategies as

210 M.-J. Zahar and S.M. Saideman well? This volume highlights two effective types of constraints while questioning the efficacy of a third. The composition of the military and the design of institutions, when properly applied, can act as brakes on the ability of the state to yield violence. Power sharing institutions, on the other hand, only seem to work to deter violence in the short term. The military and assurance Militaries are key to the provision of security by the state. Armies are the main deterrent force at the disposal of state authorities; they are entrusted with keeping opponents from using violence. However, as the main coercive instruments of state power, they can also become the main locus of threat where the delicate balance between deterrence and assurance fails to obtain. Petersen and Staniland’s contribution to the volume yields a significant finding with regard to the ability of militaries to act as a brake in relation to ethnic violence; militaries can be structured by adopting specific policies that affect the composition of and recruitment into the armed forces. Focusing on situations of reversal of ethnic status hierarchy, Petersen and Staniland find that resentment can be controlled if the group that was ousted from political power is given assurances in the form of access to (and even control of) the military. The experience of the South African Defence Forces is telling in this respect as one of the elements that explain the peaceful transition from Apartheid to Black rule in South Africa is the continued control of the armed forces by a White officer corps. This, in other words, would constitute an alternative kind of power sharing, more like power balancing where one group holds the reins of military power while the other holds the reins of political power. In such conditions, the military will act as a brake on the state’s ability to use unfettered deterrence against its opponents. Control of such a key instrument of state power will likewise provide these opponents with a credible signal of assurance from the political authorities. Thus, we have yet another reason to regret the American decision to disband the Iraqi army, as it could have played the role of a largely Sunni counterweight to the Shiitedominated government. Petersen and Staniland’s perspective has policy implications for the way in which demobilization, disarmament, and reintegration programs and security sector reform are designed on the morrow of civil wars. Rather than conceiving of national armies and police forces as simply an integrative tool, these need to be designed in such ways as to provide the state with sufficient ability to deter while simultaneously providing opponents with an avenue for assurance. Most DDR programs have almost exclusively been designed with an eye on restoring the state’s ability to deter; instead, this approach takes seriously the other side of the equation, the political economy of DDR. That

Conclusion 211 is, we need to take seriously the winners and losers of demobilization, and that we need to provide security for those who give up their weapons (Spear 2006; Muggah 2006). The approach also addresses what, to date, has been the most serious weakness of DDR and security sector reform programs, notably that these often depend on the continued presence, monitoring, and support of outsiders (Hartzell and Hoddie 2006). On this count, the experience of Bosnia, where opponents were allowed to keep separate armies following the General Framework Agreement for Peace (the Dayton peace agreement), might hold important lessons both positive and negative for the way in which we think of reintegrating opponents in the politics of countries such as Iraq or Afghanistan. Political constraints Research on conflict resolution has made much of the ability of powersharing institutions to overcome the credible commitment problem. Yet, power-sharing is not the panacea it initially appears to be. Reviewing power-sharing experiences in Africa, Donald Rothchild reaches the conclusion that, while successful at short-term deterrence of violence, powersharing institutions fail to reassure weaker parties in the long term. Rothchild traces the failure of power sharing to three separate yet reinforcing problems: the lack of reliable information; the lack of credible commitment; and the lack of external protection. Echoing the findings of Murshed, he suggests that power sharing, though not fully inclusive, weakens the autonomous power bases of salient political actors; in other words, where difficult economic situations prevail and in cases of state weakness where state actors cannot either provide or live up to the social contract that binds them to the population, elites revert to privileging their in-group as a way of securing their hold on power. In such instances, power sharing institutions fail to perform their intended role of mutual assurance. They will either falter or break down. Are alternative institutional arrangements better at constraining state actors from predation upon parts of their societies? Where policy-making requires the consensus of multiple agents, rulers’ access to political power is limited through the presence of accountability groups. Rothchild advocates multiple-majority systems and independent judiciaries as specific types of executive constraints, and Aydin and Gates concur, finding that such executive constraints are most likely to act as a brake on rulers’ tendency to use (in)security to solidify their political standings. Constitutional limits to power, they argue, increase transaction costs and diminish the utility of violent policies. Limited power also breaks information monopolies, one of the problems identified by Rothchild as a cause of the limited ability of power sharing to endure and successfully deter and assure. Indeed, reducing the power of the state over economic outcomes may also serve to reassure the populace and

212 M.-J. Zahar and S.M. Saideman reduce the stakes of competing to control the government (Steinberg and Saideman forthcoming). These findings provide important avenues for future research and they carry clear policy implications. One such implication is that power-dividing might provide a superior route to achieving the delicate equilibrium between deterrence and assurance than power sharing.5 Power-dividing seems better equipped to affect rulers’ preferences while at the same time stemming their capacity to formulate murderous policies. This requires no less than a shift in paradigm as regards conflict resolution. As stated earlier, power sharing is currently the preferred option of the international community as it attempts to rebuild states and societies that have experienced violent internal conflict. Power sharing is premised on the assumption that all parties to a negotiated peace agreement desire peace. The stumbling block is their confidence in each other’s willingness and ability to credibly commit. This good will does not always stand the test of reality. Parties that have been coerced (Serbs in Bosnia and Herzegovina) or cajoled (Kosovar Albanians) into a peace agreement might very well have entered into the deal in the hope of turning the tables on peace once the situation was clearly to their advantage (Stedman 2003). In such instances, power sharing institutions do not have the independent ability to keep humpty dumpy together. Recent research has focused on the need for power sharing across several key dimensions (Hartzell and Hoddie 2007). Contributors to this volume suggest that other institutional arrangements, including arrangements that put the power to assure and the power to deter in different hands, might stand a better chance of achieving long-term stability in deeply-divided societies. To quote Donald Rothchild, “Rather than attempting to thwart majority power by legal and political restrictions on effective action, something that often seems a futile undertaking, multiple-majority measures seek to protect minorities by spreading power.”

Security: a relational bargain A striking difference between this volume’s conceptualization of security and that of research focusing on the security dilemma is the fact that security is conceived of as a dynamic, relational bargain. In this always renewed, precarious equilibrium, the number of actors and their relationships matter for the state’s ability to successfully deter and assure at once. States and politically salient groups The central node of this relational bargain is the interaction between states and other politically salient groups in society. The smaller the number of politically salient groups, the easier it will be for states to craft successful policies of deterrence and assurance. This resonates with the findings of

Conclusion 213 research on cooperation that link success to a limited number of players (Cunningham 2006). The greater the number of groups that express grievances against the state, the more difficult it will be for the latter to craft credible policies of non-violent conflict-management. This holds particularly true in cases where the pie is small and the needs many. Faced with the difficulty of meaningfully addressing inequalities, weak states are tempted to privilege deterrence at the expense of assurance. This echoes the findings of another string of conflict analysis, one that frames state-minority relations in terms of signaling games and suggests that weak states are most likely to privilege deterrence in their interaction with societal groups in an attempt to stem others from expressing their grievances (Jenne 2004). To use the words of Mansoob Murshed, indivisibilities in shares of power and income, made worst by increasing horizontal inequalities, make it difficult for states to reach credible commitments with their opponents. Such findings do not hold particularly bright prospects for the future of weak, conflict-ridden, multinational states. Iraq might seem, in some ways, easier to address than Afghanistan since it has fewer divisions. Yet, there is much less unity within the Kurds, within the Shia, and within the Sunnis. While the US and others may think they are engaged in bargaining with three major actors, the problem of developing a security bargain is complicated (in both good ways and bad) by divisions within each group. Divisions within the Sunnis have permitted a US alliance with relatively moderate elements against the Al Qaeda in Iraq extremists, but competition within the Shia community has made it quite difficult to get the majority to commit to a bargain that assures the two minority groups. Afghanistan is complicated by the panoply of identity groups and cleavages: Pashtun, Tajiks, Uzbeks, Baluchis, and Hazaras. Yet, the harsh experience of Taliban rule has created an opening for Karzai’s government and NATO to provide some security by assuring the population that it will be much less capricious than the previous regime. Even though NATO forces have killed significant numbers of civilians in their effort to fight the Taliban, there is still much support for the government, including the army which has been working closely with NATO.6 This suggests that the effort to provide security remains somewhat successful, particularly when compared to Iraq. Long-term success depends on developing capacity and constraints in both countries. Yet, even states that have the means to develop such policies do not always choose to do so. André Lecours’ research on Spain is particularly telling in this regard. Lecours’ analysis suggests that there are limits to assurance even when the state has the means and willingness to develop such policies. To the extent that state policies promote the Spanish nation, a concept incommensurable with the Basque nation, the Basque nationalist organization ETA will continue to maintain at least some support. The nature of interactions between the state and politically salient

214 M.-J. Zahar and S.M. Saideman groups is also important in understanding the mechanisms that underpin popular mobilization. Indeed, as several contributors argue (Murshed, Lecours, Akbaba/James/Taydas), discrimination spreads insecurity and fear. State policies can, therefore, be used as a potent mechanism for group mobilization. Lecours goes as far as stating that Spanish authoritarianism created ETA because the Basque had little confidence in the state. He also shows that when the state is more even-handed and acts more reliably, those who advocate violence become marginalized, as ETA is today. Groups and external support Security is affected by dynamics that go beyond the borders of the state. External support is a time-honored way of reducing the capability asymmetry that usually favors a state at the expense of insurgent groups at the onset of violent conflict (Saleyhan 2007). But when do outsiders intervene and why? This is a hotly-debated issue in the literature on conflict-management. Some authors suggest that the nature of the relational bargain between the state and politically-salient groups is an important determinant of foreign intervention. Put differently, they argue that discrimination against minorities not only affects the potential for internal mobilization in support of the aggrieved group(s); it also increases mass mobilization in favor of intervention abroad, especially where members of the minority group also exist. Countries with mobilized publics will feel compelled to intervene in such instances.7 Others argue that external support is a function of a host of internal considerations within the states that are likely to intervene. Perceived affinity with the minorities abroad, whether based on ethnic, racial or religious kinship, is likely to militate in favor of intervention (Carment et al. 2006). Interestingly, past work has shown that deterrence dynamics do not seem to operate when it comes to external support. That is, leaders are not deterred by ethnic disputes within their own countries from supporting ethnic groups elsewhere (Saideman 2001b). Consistent with this, the findings of Akbaba, James, and Taydas reinforce our understanding of external intervention as a “chicken and egg” problem. In their words, “rebellion elicits external support and is stimulated by it as well.” The complicated relationship between Shi’a militant groups in Iraq and the Iranian regime is a case in point. While segments of the Iraqi Shi’a forces were always opposed to the American intervention, their links to the Iranian regime and the latter’s dissatisfaction with the deployment of US troops at its border make it extremely difficult to disentangle the relationship between Shi’a insurgency and Iran’s support. External intervention in support of rebellion can go a long way in limiting or enhancing the state’s ability to harm its citizens. At this stage, the net impact of such external support in terms of the overall ability of the host

Conclusion 215 state to strike a balance between deterrence and assurance is less clear. Are there conditions under which external support for the rebellion does not simply deter the state from using violence but also assists in the development of assurance policies (Jenne 2007)? The empirical record is less than encouraging, especially if we consider the polarization that resulted from international support for rebellions in places like Darfur. Indeed, external intervention in support of the rebellion has neither succeeded in stemming the Sudanese state’s over-reliance on force nor has it encouraged the Khartoum government to develop policies to assure the rebels. However, this opens up new lines of inquiry into the types of external intervention most likely to achieve such objectives. As argued by Andrew Strohlein and Gareth Evans with regards to Darfur, a full scale military invasion to implement the Responsibility to Protect doctrine would probably do more harm than good. Instead, they favor the current focus on tough economic sanctions targeted to hurt the regime’s interests “and thus pressure it into doing two things: accepting an effective international civilian protection force and making serious efforts to achieve a comprehensive and sustainable peace settlement with the rebels” (Strohlien and Evans 2007). States and the international environment The relational bargain between states and their politically-salient groups is not only affected by external support for the latter; it is also affected by the insertion of the state in an international environment that either enables or limits their ability to attain the difficult equilibrium between deterrence and assurance. In other words, the state’s ability to produce security while at the same time exercising restraint is also affected by structural conditions beyond its control. The data from Africa is particularly illustrative of these structural constraints. In his contribution to this volume, Donald Rothchild reminds us that African experiments with power sharing were “constrained by the painfully difficult economic situation that follows a civil war.” He asserts that the end of civil wars was followed by significant increases in relief but not in development assistance. Echoing the work of Paul Collier (2000), Rothchild and Murshed linked poverty to renewed risks of violent breakdown of conflict-management institutions. However, they both go beyond Collier’s work in suggesting that breaking the cycle of violence requires a fundamental shift from the current development-donor focus on poverty alleviation. Murshed contends that, much as a social contract with agreed upon rules for redistribution is required to contain dissent within developing countries, an international development contract is needed to sustain world peace. He asserts that growing global inequality has encouraged weak states to “adopt a more repressive stance with their citizens rather

216 M.-J. Zahar and S.M. Saideman than an accommodative approach.” Deterrence has come to dominate assurance as the privileged means of dealing with dissent, particularly in the aftermath of 9/11. To change this trend requires a shift in paradigm and a willingness to conceive of development aid as a privileged means of assistance to sustainable conflict-management within resource-strapped polities. That, however, is far from the current orthodoxy in international development circles (Uvin 2002; Collier 2003).

Security: a difficult balancing act At this point in the volume, it might seem almost commonplace to state that security is a difficult balancing act and that it is, at best, a precarious equilibrium, always threatened, but not always successfully renewed. It is important to underline, however, that this bargain is not equally difficult for all. Countries like Spain can boast several advantages over their African counterparts as they develop policies to deter and assure groups in their midst. For one, their institutions of conflict-management are more effective. They also have a greater redistributive capacity and more economic means to address some, though not all, of the grievances of their restive minorities. Reconceptualizing security as a bargain rather than a dilemma enabled us to provide a comparative vantage point from which to look at functioning societies and failed states alike. It is, however, important to highlight the specific challenges that weaker, conflict-ridden societies face in reestablishing the precarious equilibrium between deterring intra-state violence and assuring government restraint. It is to this task that we now turn. For conflict-ridden states When states descend into violence, their ability to restore the security bargain is seriously compromised by at least three empirical facts: the unravelling of state capacity, the failure to deter, and the lack of credibility to assure restraint. Put together, these three conditions provide a compelling logic for foreign intervention to assist in conflict resolution. Civil wars are at the same time an expression of state weakness and a harbinger of further weakening. As stated in the introduction to this volume, greater state capacity is one of the causal mechanisms tying economic strength to civil peace (Fearon and Laitin 2003). Much like the issue of external support to the regime’s opponents, this is a “chicken and egg” problem. State weakness increases vulnerability to breakdown which, in turn, weakens the state further. The weaker the state is, the less credible its capacity to assure and its ability to deter.8 Herein lies the conundrum that faces conflict-ridden states as they attempt to stem the violence. The outbreak of violence spells the failure of a state’s assurance policies. Research on the onset of civil war has cogently argued that opponents

Conclusion 217 should be deterred by the initial capability gap between themselves and state authorities. If and when opponents choose to take up weapons, it is because the present is so unappealing that the risks are worthwhile (Fearon 1995). The outbreak of violence also spells the failure of a state’s deterrence policy. This affects the relational bargain between the state and politically-salient groups within society. The weaker the state becomes as a result of ongoing warfare, the more likely and less costly successful defiance will seem. If fragility facilitates the eruption of violent conflict (Zartman 1995), the duration of conflict negatively impacts the deterrent credibility of the state. The longer the conflict, the more difficult it will be for the state to reestablish unaided a secure monopoly over violence. The duration of conflict is also likely to harden the positions of the protagonists making substantive compromise increasingly difficult. The state’s ability to assure is consequently further weakened. Though our starting points differ, these conclusions are similar to Walter’s (1997) when she argues that societies coming out of conflict are particularly prey to a credible commitment problem. Re-establishing the security bargain that underpins the functioning of normal societies requires that the state be able to deter intra-state violence and assure government restraint. The dynamics of civil war affect the capacity to deter and the ability to assure in ways that prevent such conditions from obtaining without some sort of external assistance. As well as for foreigners If conflict-ridden states cannot re-establish the security bargain on their own, can foreigners step in and help? The contribution of Martin-Brûlé tackles this question as she reflects on the conditions of and limitations to international peacekeepers’ efforts to “tackle the anarchy within,” in other words restore the monopoly of state violence as a first step toward the normalization of politics in a war-torn society. Martin-Brûlé reframes peacekeeping efforts as a dual attempt to deter protagonists from using violence during a war-to-peace transition, and to assure each one of the compliance of the other to the terms of the peace agreement. Her conclusions are sobering. Only major powers, she asserts, really possess the means of wielding credible deterrence in post-conflict situations. Given the reluctance of major powers to contribute troops to peacekeeping operations, this should cause analysts to pause. It does, however, provide limited optimism for those instances where major powers have agreed to deploy troops and may explain the relative success of the Bosnias of this world compared to the breakdown of peace in many post-conflict African states. International peacekeepers may help keep the peace, they do not necessarily help to rebuild the state’s capacity to deter and assure. At most, their deployment provides a window of opportunity for the rebuilding of state institutions. But such a process requires time and results are not

218 M.-J. Zahar and S.M. Saideman guaranteed. Six years into the deployment of the International Security Assistance Force in and around Kabul, progress is still slow and painful at the same time as populations in the troop-providing countries become increasingly restless calling for the repatriation of soldiers. Though the presence of peacekeepers is no guarantee of success, their early exit does, however, appear to be a sure sign of failure. In polities coming out of conflict, state institutions are, typically, too weak to deter and/or assure potential rebels. The presence of an outside force is a necessary, though not sufficient, condition for the restoration of the security bargain.

Policy implications and future research directions This volume has sought to provide an alternative framework to think about security in conflict-ridden societies. By reframing security as a balancing act between the need to deter intra-state violence and the necessity of assuring groups of the government’s restraint, authors have made at least three major contributions to the field of conflict analysis. First, they have recast the role of the state from hero or entirely absent to sometimes villain. In so doing, they have taken up the challenge of Davenport and Armstrong to question the assumption that state authorities are the primary agents of order and stability. Indeed, the volume is about the gap between what ought to be and what really is in this realm. Second, authors have opened new lines of inquiry into the conditions favorable to striking the delicate balancing act of security. They have specified conditions in which states are likely to err on the side of deterrence, when they are weak or captured by narrow interest groups. They have also investigated mechanisms to limit the harmful effects of states’ predatory behaviour and suggested counterintuitive policies regarding the composition and recruitment into national armies, as well as the adoption of power-dividing rather than power sharing institutions. Contributors have also linked the ability of the state to pursue meaningful assurance policies to structural constraints emanating from the international environment. Finally, authors have also provided us with a framework that overcomes the analytical distinction between functioning states on the one hand and fragile, failing and failed states (Rotberg 2003) on the other. By recasting security as a bargain, all states can now be conceived of as belonging on a continuum with states which succeed at the balancing act on one end and states that fail on the other. This opens up possibilities for comparative work on states across the security spectrum. Do states with similar regime types for example succeed equally in the difficult balancing act of deterrence and assurance? If not, why? What about states at the same level of economic development? These and other such questions hold the promise of refining the framework and furthering research on conflict analysis, management and resolution. What does this volume tell us about the way in which concerned out-

Conclusion 219 siders should step in and help? The volume reinforces some lessons that we already know while offering different solutions than the ones currently favored by politicians and political scientists. Some conflict situations are more intricate and challenging than others. A flare-up in nationalist violence in the Basque country, if it were to ever happen, would not be as difficult to address as the current deterioration of security in Iraq or Afghanistan. This is in part due to different starting points. The weaker the state (Afghanistan) or the more narrowly-based state elites (Iraq), the greater the chances that governments will use deterrence as a means of addressing the grievances of groups in their midst. Discrimination being a powerful tool for mobilization, aggrieved communities are likely to take advantage of the weakening of the state to press their demands and, should the state or regime lose its ability to deter (as happened in Iraq and Afghanistan as a result of American-led military intervention), they might become emboldened to pursue their agendas violently. Peacekeeping requires an ability to deter spoilers and to assure signatories of the peace process of each other’s compliance with the terms of peace agreements. In so doing, major powers possess clear advantages over other types of interveners. Not only do they possess the military means to credibly deter spoilers; they also possess the deep pockets and the political influence needed to grease the wheels of peace by manipulating incentives. However, the ability to deter is no guarantee of success. Regardless of their military capabilities, major powers can be hamstrung in their efforts to restore peace by other considerations. For example, where an initial deterrence posture is either not adopted or fails, major powers have to consider the use of force. Here, considerations of commitment, willingness to incur casualties, and other such political considerations might hamper the troops’ operations. In recent years, the use of the Responsibility to Protect and other such justifications for intervention based on considerations of human security have posed a challenge to major powers. Indeed, these restrict the ability of soldiers to carry out operations in civilian zones where the risk of “collateral damage” is too great and unacceptable. Such considerations have actually hampered American efforts to restore stability in Iraq in the first few months following the fall of Saddam Hussein’s regime (Zahar 2005a). Beyond the obvious consequences in terms of the presence of a security gap, these constraints have eroded the credibility of American deterrence, emboldening various groups to use violence in the pursuit of their agendas and leading to the mayhem that Iraq is experiencing today. Restoring security in countries such as Iraq and Afghanistan might require radically different solutions than those currently advocated by policy-makers. If the insights of contributors to this volume hold, then the fear and resentment of would-be opponents can only successfully be stemmed by giving them a stake in the future institutions of the state. This is not the old power sharing panacea however. Rather, contributions to

220 M.-J. Zahar and S.M. Saideman this volume suggest radical solutions such as maintaining former elites in positions of military power while transferring political power to the groups that replaced them. On this count, the US administration had it only partially right in Iraq. While they opened the way for Shiites and Kurds to assume positions of political power, they dismembered the majority Sunni-led Iraqi army. Power dividing solutions would also suggest a radical reworking of the electoral systems in Iraq and Afghanistan. This has already been advocated in Afghanistan to encourage cooperation, prevent minorities from capturing the state, and decrease the potential for vote manipulation and intimidation.9 Most radical, however, is an implicit prescription of the various contributions that, to the extent that resentment and fear fuel insurgency, spoilers, no matter how unpalatable to outsiders, might have to be brought into the security bargain with the state for peace to be self-sustaining. In sum, this volume brings the government back into the equation – its absence, its presences, its activities determine whether individuals and groups think they have a chance to change who governs and whether it is worth it. In most countries, there is not anarchy where groups compete in the absence of government. Indeed, even where states fail, it is precisely the processes that cause the state’s failure that shape the subsequent events. So, the key questions ultimately turn on whether governments can assure their citizens that their agents will not abuse their authority, but remain alert and capable enough to protect them against actors that might prey upon them. Similarly, when considering peace operations, we need to focus on the capabilities, credibility and resolve of outsiders to use force, if necessary, but also to restrain themselves as well. We take for granted this successful balancing act, this constant renewing and revising bargain, in advanced democracies, but these are exactly the challenges, successfully met in some cases, that need to be understood if we want other countries to succeed.

Notes 1 Edward Wong, “US Envoy Says He Met With Iraq Rebels,” New York Times, March 26, 2007. 2 By extension, one could think about the reaction of US Administration circles to a study published in The Lancet which sought to account for Iraqi civilian deaths, resulting either directly or indirectly from the US military intervention in Iraq. L. Roberts, R. Lafta, R. Garfield, J. Khudhairi, and G. Burnham, “Mortality Before and After the 2003 Invasion of Iraq: Cluster Sample Survey,” The Lancet, 364 (9448), November 20, 2004, pp. 1857–1864. 3 For an excellent discussion of the limits of power sharing see Roeder and Rothchild (2005). 4 This contributes to articulating conditions under which a credible commitment problem might develop, leading inexorably to a violent confrontation between the state and insurgents as described by Fearon (1998). 5 On the logic behind power dividing see Roeder and Rothchild (2006).

Conclusion 221 6 Polls taken by different organizations provide some evidence to suggest that the NATO-led effort is still supported by the Afghani public despite anecdotal stories to the contrary (Asia Foundation 2007; CBC 2007). 7 For an argument focusing on intervention being affected by the intervening state’s assessment of its likelihood of success see Regan (1998). 8 Authoritarian states might be more successful at deterrence but they fail to assure. Democratic states usually do both successfully. Anocracies are particularly vulnerable to breakdown because they do not have the sufficient capacity to deter nor do they possess the sufficient credibility to assure. 9 Barnett R. Rubin, “Afghanistan: The Wrong Voting System,” International Herald Tribune, March 16, 2005, consulted at www.iht.com/articles/ 2005/03/15/opinion/edrubin.php on April 18, 2007.

References

Acemoglu, D. and Robinson, J.A. (2001) ‘A Theory of Political Transitions,’ American Economic Review, 91(4): 938–963. Addison, T. and Murshed, S.M. (2002a) ‘On the Economic Causes of Contemporary Civil Wars,’ in S.M. Murshed (ed.), Issues in Positive Political Economy, London: Routledge. Addison, T. and Murshed, S.M. (2002b) ‘Credibility and Reputation in Peacemaking,’ Journal of Peace Research, 39(4): 487–501. Addison, T., Le Billon, P., and Murshed, S.M. (2002) ‘Conflict in Africa: The Cost of Peaceful Behaviour,’ Journal of African Economies, 11(3): 365–386. Adebajo, A. (2003) Liberia’s Civil War: Nigeria, ECOMOG and Regional Security in West Africa. Boulder, CO: Lynne Rienner. Adibe, C. (1997) ‘The Liberian Conflict and the ECOWAS–UN Partnership,’ Third World Quarterly, 18: 471–488. Africa Confidential (2004) ‘Maintaining Group Power: Liberia: Silencing the Guns,’ Africa Confidential, 45: 3 (February 6). Africa Research Bulletin (2003) ‘Cote d’Ivoire: Can Marcoussis Be Implemented?,’ Africa Research Bulletin, 40: 2 (February). Africa Research Bulletin (2004a) ‘Liberia: LURD Rift,’ Africa Research Bulletin 41:1 (January). Africa Research Bulletin (2004b) ‘Burundi: Broken Ceasefire,’ Africa Research Bulletin, 41(4) (April). Africa Research Bulletin (2004c) ‘Sudan: Last Minute Hitches,’ Africa Research Bulletin, 41(5) (May). Agence France Presse (2003) ‘Kabila Calls Rivals to Kinshasa 14 April for Formation of New Government’, April 2. Agence France Presse (2003) ‘Congo-K: MLC Rebels Arrive in Kinshasa to Take Up Posts in Transitional Cabinet,’ July 2. Agence France Presse (2003) ‘Most of New Transitional Government Ministers Sworn in 15 July,’ July 15. Agence France Presse (2003) ‘New DRC Government’s First Meeting ‘Delayed’ as Some Ministers Not Sworn In,’ July 19. Agence France Presse (2003) ‘Cote d’Ivoire: New Forces Pull Out of Unity Government,’ September 23. Agence France Presse (2004) ‘L’ex-principale rebellion au Burundi: les discussions vont se poursuivre (Zuma),’ July 27. Agence France Presse (2004) ‘Mandela a apporté son appui aux négociations sur le

References 223 Burundi à Pretoria’ [Mandela brought his weight to the negotiations on Burundi in Pretoria], July 21. Anderson, M.B. (1994) ‘International Assistance and Conflict: An Exploration of Negative Impacts,’ The Local Capacities for Peace Project, Issues Series no. 1. Anderson, M. (1999) Do No Harm – How Can Aid Support Peace – Or War, London: Lynne Rienner. Arendt, H. (1951) The Origins of Totalitarianism, New York: Harcourt. Asia Foundation (2007) ‘Afghanistan in 2007: A Survey of the Afghan People,’ available at www.asiafoundation.org/pdf/AG-survey07.pdf, accessed October 30, 2007. Axworthy, Lloyd (2001) ‘Human Security and Global Governance: Putting People First,’ Global Governance, 7: 19–23. Ayoob, M. (2001) ‘State Making, State Breaking, and State Failure,’ in C.A. Crocker, F.O. Hampson, and P. Aall (eds), Turbulent Peace: The Challenges of Managing International Conflict, New York: USIP. Ayres, R.W. and Saideman, S.M. (2000) ‘Is Separatism as Contagious as the Common Cold or as Cancer? Testing the International and Domestic Determinants of Secessionism,’ Nationalism and Ethnic Politics 6 (Fall): 92–114. Azam, J.-P. (2005) ‘Can Lasting Peace Be Imported,’ Paper presented at the Montreal Research Group on Ethnic Conflict Workshop, Montreal, Canada, June 17. Azarya, V. (1994) ‘Civil Society and Disengagement in Africa,’ in J. Harbeson, D. Rothchild, and N. Chazan (eds), Civil Society and the State in Africa, Boulder, CO: Lynne Reinner. Aznar, J.M. (2004) Presidential Address. ‘Seven Theses on Today’s Terrorism,’ September 21, available at http: data.georgetown.edu/president/aznar/inauguraladdress.html. Baldwin, R. and Martin, P. (1999) ‘Two Waves of Globalisation: Superficial Similarities, Fundamental Differences,’ in H. Siebert (ed.), Globalisation and Labour, Tubingen: JCB Mohr. Ball, P., et al. (1999) State Violence in Guatemala, 1960–1996: A Quantitative Reflection. Washington, DC: American Association for the Advancement of Science. Ballantine, K. and Sherman, J. (eds) (2003) The Political Economy of Armed Conflict: Beyond Greed and Grievance, Boulder, CO: Lynne Rienner. Banks, A.S. (2001) Cross-National Time-Series Data Archive, available at www.databanks.sitehosting.net/, accessed October 28, 2007. BBC News (2000) ‘Army Defends Sierra Leone Leaflets,’ BBC News, Thursday June 1, available at news.bbc.co.uk/1/hi/uk_politics/772810.stm. BBC News (2000) ‘Who are the West Side Boys?,’ BBC News, Thursday August, available at news.bbc.co.uk/1/hi/world/africa/901209.stm. BBC News (2001) ‘World: Africa New Threat in Sierra Leone Conflict,’ BBC News, Sunday, January 10, available at news.bbc.co.uk/1/hi/world/ africa/252210.stm. BBC News (2003a) ‘Trouble Mounts for Ivory Coast Leader,’ BBC News, January 29, available at news.bbc.co.uk/2/hi/africa/2706431.stm, accessed August 8, 2005. BBC News (2003b) ‘Rebels Boycott Ivory Coast Peace Talks,’ BBC News, February 10, available at news.bbc.co.uk/2/hi/africa/2743323.stm, accessed August 8, 2005.

224 References BBC News (2004) ‘DR Congo Rebels Leave Government,’ BBC News, August 23, available at news.bbc.co.uk/2/hi/africa/3592174.stm, accessed August 8, 2005. BBC News (2005) ‘Darfur is “Key to Peace in Sudan,” ’ 8 February, available at newswww.bbc.net.uk/2/hi/africa/4238279.stm, accessed August 8, 2005. BBC On-line, ‘Country Time-line: Burundi,’ available at http:news.bbc.co.uk/ 1/hi/world/africa/country_profiles/1068991.stm Accessed August 8, 2005. BBC On-line, ‘Country Time-line: Côte d’Ivoire,’ available at news.bbc.co.uk/1/ hi/world/africa/country_profiles/1043014.stm, accessed August 8, 2005. BBC On-line, ‘Country Time-line: DR Congo,’ available at news.bbc.co.uk/1/hi/ world/africa/country_profiles/1072684.stm, accessed August 8, 2005. BBC On-line, ‘Country Time-line: Liberia,’ available at news.bbc.co.uk/1/hi/ world/africa/country_profiles/1043567.stm, accessed August 8, 2005. BBC On-line, ‘Country Time-line: Rwanda,’ available at news.bbc.co.uk/1/hi/ world/africa/country_profiles/1070329.stm, accessed August 8, 2005. BBC On-line, ‘Country Time-line: Sudan,’ available at news.bbc.co.uk/1/hi/ world/middle_east/827425.stm, accessed August 8, 2005. BBC On-line, ‘Darfur war threatens Sudan deal,’ June 28, 2004, available at news.bbc.co.uk/1/hi/world/africa/3845789.stm, accessed August 8, 2005. Bélanger, L., Duchesne, E., and Paquin, J. (2005) ‘Foreign Interventions and Secessionist Movements: The Democratic Factor,’ Canadian Journal of Political Science, 38: 435–462. Bellamy, A.J. and Williams, P.D. (2005) ‘Who’s Keeping the Peace? Regionalization and Contemporary Peace Operations,’ International Security, 29(4): 157–195. Berman, E.G. and Sams, K.E. (2000) Peacekeeping in Africa: Capabilities and Culpabilities, United Nations Institute for Disarmament Research. Besançon, M. (2005) ‘Relative Resources: Inequality in Ethnic Wars, Revolutions and Genocides,’ Journal of Peace Research, 42(4), 393–415. Bett, R.K. (2000) ‘Is Strategy an Illusion?,’ International Security, 25(2) (Fall): 5–50. Bougarel, X. (1996) Bosnie, anatomie d’ un conflit, Paris: La Découverte. Boyd, R. (2001) ‘Thomas Hobbes and the Perils of Pluralism,’ Journal of Politics, 63(2): 392–413. Brams, S.J. (2005) ‘Fair Division,’ in B.R. Weingast and D. Wittman (eds), Oxford Handbook of Political Economy, New York: Oxford University Press. Bratton, P.C. (2005) ‘When is Coercion Successful? And Why Can’t We Agree on it?,’ Naval War College Review, Summer, 58(3): 99–120. Brown, M.E. (2001) ‘The Sources of Nationalism and Ethnic Conflict,’ in M.E. Brown, O.R. Coté Jr, S.M. Lynn-Jones and S.E. Miller (eds), Nationalism and Ethnic Conflict, Cambridge, MA: MIT Press. Brunnée, J. and Toope, S.J. (2004) ‘Canada and the Use of Force: Reclaiming Human Security,’ International Journal, 59(2): 247–260. Bryans, M., Jones, B.D., and Stein, J.G. (1999) ‘Mean Times: Humanitarian Action in Complex Political Emergencies-Stark Choices, Cruel Dilemmas,’ Report of the NGOs in Complex Emergencies Project, 1, 3 (January). Bueno de Mesquita, B. and Root, H. (2000) ‘When Bad Economics is Good Politics,’ in B. Bueno de Mesquita and H. Root (eds), Governing for Prosperity, New Haven, CT: Yale University Press. Bueno de Mesquita, B., Morrow, J.D., Siverson, R.M., and Smith, A. (1999)

References 225 ‘Policy Failure and Political Survival: The Contribution of Political Institutions.’ Journal of Conflict Resolution, 43(2): 147–161. Bueno de Mesquita, B., Morrow, J.D., Siverson, R.M., and Smith, A. (2002) ‘Political Institutions, Policy Choice, and the Survival of Leaders,’ British Journal of Political Science 32(4): 559–590. Bueno de Mesquita, B., Morrow, J.D., Siverson, R.M., and Smith, A. (2003) The Logic of Political Survival, Cambridge, MA: MIT Press. Burton, M., Gunther, R., and Higley, J. (1992) ‘Elite Transformations and Democratic Regimes,’ in J. Higley and R. Gunther (eds), Elites and Democratic Consolidation in Latin America and Southern Europe. London: Cambridge University Press. Buzan, B., Waever, O., and de Wilde, J. (1998) Security: A New Framework for Analysis, Boulder, CO and London: Lynne Rienner. Carment, D. (1993) ‘The International Dimensions of Ethnic Conflict: Concepts, Indicators and Theory.’ Journal of Peace Research, 30: 137–150. Carment, D. and Harvey, F. (1999) Preventing, Managing and Resolving Ethnic Conflict, New York: Praeger. Carment, D. and James, P. (1995) ‘Internal Constraints and Interstate Ethnic Conflict: Toward a Crisis-Based Assessment of Irredentism,’ Journal of Conflict Resolution, 39: 82–109. Carment, D. and Rowlands, D. (2001) ‘Evolutionary Models, Third-Party Intervention, and Ethnic Conflict: Does Tough Love Really Work?,’ in P. James and D. Goetze (eds), Evolutionary Theory and Ethnic Conflict, Westport, CT: Praeger. Carment, D. and Rowlands, D. (2004) ‘Vengeance and Intervention: Can Third Parties Bring Peace Without Separation?,’ Security Studies, 13(4): 1–28. Carment, D., James, P., and Rowlands, D. (1997) ‘Ethnic Conflict and Third Party Intervention: Riskiness, Rationality and Commitment,’ in P. Weitsman and G. Schneider (eds), Enforcing Cooperation: Risky States and Intergovernmental Management of Conflict, New York: Macmillan. Carment, D., James, P., and Taydas, Z. (2006) Who Intervenes? Ethnic Conflict and Interstate Crisis. Columbus, OH: Ohio State University Press. CBC (2007) ‘51% of Afghans Feeling Good About Country’s Direction in Poll,’ October 18, available at www.cbc.ca/world/story/2007/10/18/afghan-poll.html, accessed October 30, 2007. Cederman, L.-E. and Girardin, L. (2007) ‘Beyond Fractionalization: Mapping Ethnicity onto Nationalist Insurgencies,’ American Political Science Review, 101(1) (February): 173–185. Central Intelligence Agency (CIA) (1979) National Basic Intelligence Factbook, Washington, Office of Geographic and Cartographic Research. Chang, J. and Halliday, J. (2005) Mao: The Unknown Story. London: Jonathan Cape. Choi, S. and James, P. (2005) Civil-Military Dynamics, Democracy, and International Conflict: A New Quest for International Peace, New York: Palgrave Macmillan. Cingranelli, D.L. and Richards, D.L. (2004) The Cingranelli–Richards (CIRI) Human Rights Dataset 1980–2004. Clark, R.P. (1984) The Basque Insurgents. ETA, 1952–1980, Madison, WI: The University of Wisconsin Press.

226 References Cochran, E. (2000–01) ‘The Pivotal State: Post-Apartheid South Africa,’ Parameters: The US Army War College Quarterly, 30(4) (Winter): 83–101. Cohen, F. (1997) ‘Proportional Versus Majoritarian Ethnic Conflict Management in Democracies,’ Comparative Political Studies, 30(5): 607–630. Collier, P. (2000) ‘Rebellion as a Quasi-Criminal Activity,’ Journal of Conflict Resolution, 44(6): 839–853. Collier, P. (2001) ‘Economic Causes of Civil Conflict and Their Implication for Policy,’ in C.A. Crocker, F.O. Hampson, and P. Aall (eds), Turbulent Peace: The Challenges of Managing International Conflict, New York: USIP. Collier, P. (2003) Breaking the Conflict Trap: Civil War and Development Policy, Washington, DC: World Bank Publications. Collier, P. and Hoeffler, A. (2001) ‘Greed and Grievance in Civil War,’ World Bank Policy Research Working Paper 2355, Washington, DC. Collier, P. and Hoeffler, A. (2004) ‘Greed and Grievance in Civil Wars,’ Oxford Economic Papers, 56: 663–595. Collier, P., Elliot, L., Hegre, H., Hoeffler, A., Reynal-Querol, M. and Sambanis, N. (2003) Breaking the Conflict Trap: Civil War and Development Policy, Oxford: Oxford University Press. Collier, P., Hoeffler, A., and Söderbom, M. (2004) ‘On the Duration of Civil War,’ Journal of Peace Research, 41(3), 253–273. Connaughton, R. (2001) Military Intervention and Peacekeeping: The Reality, Burlington: Ashgate. Conversi, D. (1997) The Basque, the Catalans and Spain. Alternatives Routes to Nationalist Mobilisation, London: Hurst. Cooper, N. (2003) ‘Liberal Governance, War Economies and the Emerging Control Agenda,’ Paper for the Conference on Resource Politics and Security in a Global Age University of Sheffield, June 26–28. Coser, L. (1956) The Functions of Social Conflict, Glencoe, IL: Free Press. Cunningham, D.E. (2006) ‘Veto Players and Civil War Duration,’ American Journal of Political Science, 50(4): 875–892. Dallaire, R. (2004) Shake Hands with the Devil: The Failure of Humanity in Rwanda. New York: Carroll & Graf Publishers. Dallin, A. and Breslauer, G.W. (1970) Political Terror in Communist Systems, Stanford, CA: Stanford University Press. Daragahi, B. (2006) ‘The Day the Hatred Boiled Over in Balad,’ Los Angeles Times, November 7. Davenport, C. (2004) ‘Human Rights and the Promise of Democratic Pacification,’ International Studies Quarterly, 48(3): 539–560. Davenport, C. and Armstrong, D.A. II (2004) ‘Democracy and the Violation of Human Rights: A Statistical Analysis from 1976–1996,’ American Journal of Political Science. 48(3): 538–554. Davenport, C. and Ball, P. (2002) ‘Views to a Kill: Exploring the Implications of Source Selection in the Case of Guatemalan State Terror,’ Journal of Conflict Resolution, 46(3): 427–450. Davenport, C. and Stam, A. (2004) ‘Genodynamics,’ www.genodynamics.com, accessed October 28, 2007. Davenport, C., Armstrong, D.A. II, and Lichbach, M.I. (2006) ‘From Mountains to Movements: Dissent, Repression and Escalation to Civil War,’ Unpublished Manuscript.

References 227 Davenport, C., Johnston, H. and Mueller, C. (eds) (2004) Repression and Mobilization, Minneapolis, MN: University of Minnesota Press. David, S.R. (1997) ‘Internal War: Causes and Cures,’ World Politics, 49(4): 552–576. De Figueiredo, R.J.P. Jr. and Weingast, B.R. (1999) ‘The Rationality of Fear: Political Opportunism and Ethnic Conflict,’ in B.F. Walter and J. Snyder (eds), Civil Wars, Insecurity, and Intervention, New York: Columbia University Press. De Jouvenal, B. (1945) On Power: Its Nature and the History of Its Growth, Boston, MA: Beacon Press. Denselow, R. (2000) ‘Disarming Sierra Leone’s Rebels,’ BBC News. news.bbc.co.uk/1/hi/world/africa/694209.stm, accessed October 28, 2007. Des Forges, A. (1999) ‘Leave None to Tell the Story: Genocide in Rwanda,’ New York: Human Rights Watch. available at hrw.org/reports/1999/rwanda/, accessed August 8, 2005. Dougherty, C. (2004) ‘Peace Precarious in Eastern Congo: Tensions Simmer a Year After Pact Ended Civil War,’ The Washington Times, May 5, available at www.publicinternationallaw.org/docs/PNW4/PNW.10May_04.html, accessed August 8, 2005. Douglass, W. and Zulaika, J. (1990) ‘On the Interpretation of Terrorist Violence: ETA and the Basque Political Process,’ Comparative Studies in Society and History, 32: 238–257. Downes, A.B. (2001) ‘The Holy Land Divided: Defending Partition as a Solution to Ethnic Wars,’ Security Studies, 10: 58–116. Downes, A.B. (2004) ‘The Problem with Negotiated Settlements to Ethnic Civil Wars,’ Security Studies 13(4): 230–279. Doyle, M. (1999) ‘World: Africa Nigerian Election “Threatens” Sierra Leone,’ BBC News, Saturday, February 27, news.bbc.co.uk/1/hi/world/africa/ 287236.stm, accessed October 28, 2007. Doyle, M.W. and Sambanis, N. (2000). ‘International Peacebuilding: A Theoretical and Quantitative Analysis,’ American Political Science Review, 94(4): 779–801. Drake, P.W. and McCubbins, M.D. (1998) ‘The Origins of Liberty,’ in P.W. Drake and M.D. McCubbins (eds), The Origins of Liberty: Political and Economic Liberalization in the Modern World, Princeton, NJ: Princeton University Press, 3–15. Drysdale, J. (1997) ‘Foreign Military Intervention in Somalia: The Root Cause of the Shift from UN Peacekeeping to Peacemaking and Its Consequences,’ in W. Clarke and J. Herbst (eds), Learning from Somalia: The Lessons of Armed Humanitarian Intervention, Oxford: Westview Press. Dukulé, A.W. (2004) ‘The US and Liberia: Can We Prevent a Repeat of the Haiti Scenario?,’ The Perspective, March 4, www.theperspective.org/2004/mar/usandliberia.html, accessed October 28, 2007. Dukuly, A. (2004) ‘Politics-Liberia: Former Rebels Surrender Weapons in Return for Money,’ Inter-Press Service News Agency, April 17, www.ipsnews.net/ interna.asp?idnews=23360, accessed October 28, 2007. Durch, W.J., Holt, V.K., Earle, C.R., and Shanahan, M.K. (2003) The Brahimi Report and the Future of UN Peace Operations, Washington, DC: Henry L. Stimson Center. Duvall, R. and Stohl, M. (1988) ‘Governance by Terror,’ in M. Stohl (ed.), Politics of Terrorism, 3rd edn, New York: Marcel Dekker.

228 References Dwyer, L. and Santikarma, D. (2003) ‘When the World Turned to Chaos: 1965 and Its Aftermath in Bali, Indonesia,’ in R. Gellately and B. Kiernan (eds), The Specter of Genocide: Mass Murder in Historical Perspective, Cambridge: Cambridge University Press. Eck, K. and Hultman, L (2007) ‘One-Sided Violence Against Civilians in War,’ Journal of Peace Research, 44(2): 233–246. Eckstein, H. (1964) Internal War: Problems and Approaches. New York: Free Press. Economist Intelligence Unit (1997) ‘Country Profile, Burundi.’ Economist Intelligence Unit (2005) ‘Country Profile, Burundi.’ Economist Intelligence Unit (2005) ‘Country Profile, Côte d’Ivoire.’ Economist Intelligence Unit (2004) ‘Country Profile, DR Congo.’ Economist Intelligence Unit (2004) ‘Country Profile, Liberia.’. Economist Intelligence Unit (2005) ‘Country Profile, Sudan.’ Failed States Index (2006) ‘Foreign Policy and Fund for Peace,’ May/June, www.foreignpolicy.com, accessed May 2006. Fearon, J.D. (1995) ‘Rationalist Explanations for War,’ International Organization, 49(3) (Summer): 379–414. Fearon, J.D. (1998) ‘Commitment Problems and the Spread of Ethnic Conflict’, in D.A. Lake and D. Rothchild (eds), The International Spread of Ethnic Conflict, Princeton, NJ: Princeton University Press: 107–126. Fearon, J.D. (2004) ‘Why Do Some Civil Wars Last So Much Longer Than Others?,’ Journal of Peace Research, 41(3): 275–301. Fearon, J.D. and Laitin, D. (2002) ‘Group Concentration and Civil War,’ Paper prepared for presentation at the 2002 Annual Meeting of the American Political Science Association, Boston, MA, August 29–September 2002. Fearon, J.D. and Laitin, D. (2003) ‘Ethnicity, Insurgency and Civil War,’ American Political Science Review, 97(1): 75–90. Federal Republic of Nigeria (1999) Official Gazette, 86(27), May 5, Lagos: Federal Government Press. Fellman, M. (1989) Inside War: The Guerrilla Conflict in Missouri During the American Civil War, New York: Oxford University Press. FM 100–23. (1994) Peace Operations, Headquarters, Department of the Army, available at www.dtic.mil/doctrine/jel/service_pubs/fm100_23.pdf, accessed January 4, 2005. Forbes, H.D. (1997) Ethnic Conflict: Commerce, Culture, and the Contact Hypothesis, New Haven, CN: Yale University Press. Foreign Broadcast Information Service (2003) ‘Côte d’Ivoire: “Rebel” ministers continue participation in the government despite leader’s decision,’ Trans. Paris Jeune Afrique-L’Intelligent, November 16–22: 8–9. Fox, J. (2000) ‘The Effects of Religious Discrimination on Ethno-Religious Protest and Rebellion,’ Journal of Conflict Studies, XX(1). Fox, J. (2001) ‘Patterns of Discrimination, Grievances, and Political Activity Among Europe’s Roma: A Cross-Sectional Analysis,’ Journal on Ethnopolitics and Minority Issues in Europe, Winter (e-journal). Freedman, L. (2004) Deterrence, Cambridge: Polity Press. Freedman, L. (2005) ‘Deterrence: A Reply,’ The Journal of Strategic Studies, 28(5) (October): 789–801. Freud, S. (1915/1984) On Metapsychology: the Theory of Psychoanalysis. Harmondsworth: Penguin.

References 229 Gagnon Jr, V.P. (1994/95) ‘Ethnic Nationalism and International Conflict: The Case of Serbia,’ International Security, 19(3): 130–166. Gagnon, V.P. (2004) The Myth Of Ethnic War: Serbia and Croatia in the 1990s, Ithaca. NY: Cornell University Press. Gates, S., Hegre, H., Jones, M., and Strand, H. (2006) ‘Institutional Inconsistency and Political Instability: The Duration of Polities,’ American Journal of Political Science, 50(4): 893–908. Geddes, B. (1999) ‘Authoritarian Breakdown: Empirical Test of a Game-Theoretic Argument,’ Paper presented at the 95th Annual Meeting of the American Political Science Association, Atlanta, GA, September 2–5. Gelfand, L. (2004) ‘Ivorian or Liberian, Few Fighters Show Up to Disarm in Buchanan,’ Reliefweb, May 4, available at www.reliefweb.int/w/ Rwb.nsf/6686f45896f15dbc852567ae00530132/a1c7e267381a0d6885256e8a0 0595a93?OpenDocument, accessed October 28, 2007. Gelfand, L. (2004) ‘UN Opens Last Phase of Liberian Disarmament Amid Questions Over Missing Guns,’ Reliefweb, April 30, available at www.reliefweb. int/ rw/rwb.nsf/db900sid/OCHA-64DAGA?OpenDocument&query=UN%20opens %20last%20phase%20of%20Liberian%20disarmament%20amid%20questions% 20over%20missing%20guns, accessed October 28, 2007. George, A. and Smoke, R. (1989) ‘Deterrence and Foreign Policy,’ World Politics, 41 (January): 170–182. Gibney, M. (2004) Political Terror Scale Data 1980–2003, available at: www.unca.edu/politicalscience/faculty-staff/gibney_docs/pts.xls Gillespie, R. (1999) ‘Peace Moves in the Basque Country,’ Journal of Southern Europe and the Balkans, 1: 119–136. Gleditsch, K.S. (2002) All International Politics Is Local: The Diffusion of Conflict, Integration, and Democratization, Ann Arbor, MI: University of Michigan Press. Gleditsch, N.P., Wallensteen, P., Eriksson, M., Sollenberg, M., and Strand, H. (2002) ‘Armed Conflict 1946–2001: A New Dataset,’ Journal of Peace Research, 39(5): 615–637. Grandin, G. (2003) ‘History, Motive, Law, Intent: Combining Historical and Legal Methods in Understanding Guatemala’s 1981–1983 Genocide,’ in R. Gellately and B. Kiernan (eds), The Specter of Genocide: Mass Murder in Historical Perspective, Cambridge: Cambridge University Press. de la Granja Sainz, J.L. (2002) El nacionalismo vasco: un siglo de historia, Madrid: Tecnos. Greenhill, K.M. and Major, S. (2007) ‘The Perils of Profiling: Civil War Spoilers and the Collapse of Intrastate Peace Accords,’ International Security, 31(3): 7–40. Grossman, H. (1991) ‘A General Equilibrium Model of Insurrections,’ American Economic Review, 81: 912–21. Grossman, H. and Kim, M. (1996) ‘Predation and Production,’ in M. Garfinkel and S. Skaperdas (eds), The Political Economy of Conflict and Appropriation, New York: Cambridge University Press. Gunther, R. and Diamond, L. (2001) ‘Types and Functions of Parties,’ in L. Diamond and R. Gunther (eds), Political Parties and Democracy, Baltimore, MD: Johns Hopkins University. Gurr, T.R. (1970) Why Men Rebel, Princeton, NJ: Princeton University Press. Gurr, T.R. (1993a) ‘Resolving Ethnopolitical Conflicts: Exit, Autonomy or Access,’

230 References in T.R. Gurr (ed.), Minorities at Risk: Origins and Outcomes of Ethnopolitical Conflicts, Washington, DC: United States Institute of Peace Press. Gurr, T.R. (1993b) Minorities at Risk: A Global View of Ethnopolitical Conflicts, Washington, DC: US Institute of Peace Press. Gurr, T.R. (1994) ‘Peoples Against States: Ethnopolitical Conflict and the Changing World System,’ International Studies Quarterly, 38: 347–377. Gurr, T.R. (2000) Peoples Versus States: Ethnopolitical Conflict and Accommodation at the End of the 20th Century, Washington, DC: US Institute of Peace Press. Gurr, T.R. (2002) Minorities at Risk Dataset and Codebook, Center for International Development and Conflict Management. Gurr, T.D. and Harff, B. (1994) Ethnic Conflict in World Politics, Boulder, CO: Westview. Hafner-Burton, E. (2005) ‘Trading Human Rights: How Preferential Trade Agreements Influence Government Repression,’ International Organization, 59 (Summer): 593–629. Hamilton, A. (n.d.) ‘Concerning the Power of Congress to Regulate the Election of Members,’ The Federalist: A Commentary on the Constitution of the United States, no. 59. New York: Random House: 383–389. Hamilton, A. or Madison, J. (n.d.) ‘The Senate,’ The Federalist: A Commentary on the Constitution of the United States, no. 62. New York: Random House: 400–407. Hamilton, C. (2000) ‘Re-membering the Basque Nationalist Family: Daughters, Fathers, and the Reproduction of the Radical Nationalist Community,’ Journal of Spanish Cultural Studies, 1: 153–171. Haq, F. (1998) ‘Politics – Sierra Leone: UN Help Needed,’ World News. Hardin, R. (1995) One for All: The Logic of Group Conflict, Princeton, NJ: Princeton University Press. Harff, B. (2003) ‘No Lessons Learned from the Holocaust: Assessing Risks of Genocide and Political Mass Murder Since 1955,’ American Political Science Review, 97(1): 57–74. Harff, B. and Gurr, T.R. (1989) ‘Victims of the State: Genocides, Politicides and Group Repression Since 1945,’ International Review of Victimology, 1: 23–41. Harff, B. and Gurr, T.R. (2004) Ethnic Conflict in World Politics, 2nd edn, Boulder, CO: Westview. Harman, D. (2002) ‘Sierra Leone: The Path from Pariah to Peace,’ Christian Science Monitor, September 18. Hartzell, C. and Hoddie, M. (2006) ‘From Anarchy to Security: Comparing Theoretical Approaches to the Process of Disarmament Following Civil War,” Contemporary Security Policy, 27(1). Hartzell, C. and Hoddie, M. (2007) Crafting Peace: Power Sharing and the Negotiated Settlement of Civil Wars, University Park, PA: Penn State University Press. Hartzell, C., Hoddie, M., and Rothchild, D. (2001) ‘Stabilizing the Peace After Civil War: An Investigation of Some Key Variables,’ International Organization, 55(1) (Winter): 183–208. Harvey, F. (1999) ‘Practicing Coercion: Revisiting Successes and Failure Using Boolean Logic and Comparative Methods,’ Journal of Conflict Resolution, 43(6) (December): 840–871. Hathaway, O. (2002) ‘Do Human Rights Treaties Make a Difference?,’ The Yale Law Journal, 111(8): 1935–2042.

References 231 Hegre, H. (2004) ‘The Duration and Termination of Civil War,’ Journal of Peace Research, 41(3): 243–252. Heraclides, A. (1990) ‘Secessionist Minorities and External Involvement,’ International Organization, 44: 341–378. Herbst, J. (2000) States and Power in Africa: Comparative Lessons in Authority and Control, Princeton: Princeton University Press. Herz, J.H. (1951) Political Realism and Political Idealism: A Study in Theories and Realities, Chicago, IL: University of Chicago Press. Heston, A., Summers, R. and Aten, B. (2002) Penn World Table Version 6.1, Center for International Comparisons at the University of Pennsylvania (CICUP). Higgs, J. (2000) ‘Creating the South African National Defence Force.’ Joint Forces Quarterly, Summer: 45–50. Hill, S. and Malik, S.P. (1996) Peacekeeping and the United Nations, Brookfield, VT: Dartmouth: 136–141. Hillen, J. (1998) Blue Helmets: The Strategy of UN Military Operations, London: Brassey’s. Hirsh, J.L. (2001a) Sierra Leone: Diamonds and the Struggle for Democracy, International Peace Academy Occasional Paper Series. Hirsh, J.L. (2001b) ‘War in Sierra Leone,’ Survival, 43 (Autumn): 145–162. Hirsh, J.L. and Oakley, R.B. (1995) Somalia and Operation Restore Hope: Reflections on Peacemaking and Peacekeeping, Washington, DC: United States Institute of Peace Press. Hobbes, T. (1950) Leviathan, New York: Dutton. Hoddie, M. and Hartzell, C. (2005) ‘The Role of Power Sharing in Peace Settlements: Initiating the Transition From Civil War,’ in P. Roeder and D. Rothchild (eds), Sustainable Peace: Power and Democracy After Civil Wars, Ithaca, NY: Cornell University Press. Horowitz, D.L. (1985) Ethnic Groups in Conflict, Berkeley, CA: University of California Press. Horowitz, D.L. (1991) ‘Making Moderation Pay: The Comparative Politics of Ethnic Conflict Management,’ in J.V. Montville (ed.), Conflict and Peacemaking in Multiethnic Societies, Lexington, MA: Lexington Books. Horowitz, D.L. (2001) The Deadly Ethnic Riot, Berkeley, CA: University of California Press. Horowitz, D.L. (2004) ‘Electoral Systems and Their Goals: A Primer for DecisionMakers,’ Journal of Democracy, 14(4): 115–127. Howe, H. (1994) ‘The South African Defence Force and Political Reform,’ Journal of Modern African Studies, 32(1): 29–51. Human Rights Watch (1999) ‘Democratic Republic of Congo, Casualties of War: Civilians, Rule of Law and Democratic Freedoms,’ available at hrw.org/reports/ 1999/congo/, accessed October 28, 2007. Human Rights Watch (1999) Leave None to Tell the Story: Genocide in Rwanda – The Planning and Execution of Mass Murder, March, available at www.hrw.org/reports/1999/rwanda/, accessed October 28, 2007. Human Rights Watch (2005) www.hrw.org, accessed October 28, 2007. Human Security Report (2005) University of British Columbia, available at www.humansecurityreport.info/, accessed October 28, 2007. Hultman, L. and Eck, K. (2005) ‘Violence Against Civilians in War: Insights from

232 References New Fatality Data,’ Paper presented at the 2005 Annual Meeting of the International Studies Association, Honolulu, Hawaii. Huth, P.K. (1999) ‘Deterrence and International Conflict: Empirical Findings and Theorical Debates,’ Annual Review of Political Science, 2: 25–48. Huth, P.K. and Russet, B. (1984) ‘What Makes Deterrence Work? Cases From 1900 to 1980,’ World Politics, 36 (July): 496–526. Inter-Congolese Dialogue documents, available at www.iss.co.za/AF/profiles/ DRCongo/icd/, accessed August 8, 2005. International Crisis Group (ICG) (2001) ‘Sierra Leone Time for a New Military and Political Strategy,’ ICG Africa Report, no. 28, April 11, Freetown/London/ Brussels, available at www.crisisweb.org//library/documents/report_archive/ A400278_11042001.pdf, accessed October 28, 2007. International Crisis Group (ICG) (2002) ‘Storm Clouds Over Sun City: The Urgent Need to Recast the Congolese Peace Process,’ ICG Africa Report, no. 44, available at www.crisisgroup.org/library/documents/report_archive/A400652_ 14052002.pdf, accessed August 8, 2005. International Crisis Group (ICG) (2003a) ‘Côte d’Ivoire: “The War is Not Yet Over,” ’ ICG Africa Report, no. 72, Freetown, Brussels: ICG. International Crisis Group (ICG) (2003b) ‘Liberia: Security Challenges,’ ICG Africa Report, no. 71, Freetown/Brussels available at www.crisisweb.org//library/ documents/africa/071_liberia_security_challenges.pdf, accessed October 28, 2007. International Crisis Group (ICG) (2003c) ‘Liberia: Security Challenges,’ ICG Africa Report, no. 71, Freetown/Brussels available at www.crisisweb.org// library/documents/africa/071_liberia_security_challenges.pdf, accessed November 3, 2007. International Crisis Group (ICG) (2004a) ‘Fin de Transition au Burundi: Franchir le Cap’, ICG Africa Report, no. 81, July 5, available at www.crisisweb.org/ home/index.cfm?l=2&id=2841, accessed August 8, 2005. International Crisis Group (ICG) (2004b) ‘Rebuilding Liberia: Prospects and Perils,’ ICG Africa Report, no. 75, Freetown/Brussels, January 30, available at www.crisisweb.org//library/documents/africa/west_africa/075_rebuilding_liberia _prospects.pdf, accessed October 28, 2007. International Crisis Group (ICG) (2006) ‘Liberia: Staying Focused,’ ICG Africa Report, no. 36, January 13, available at www.crisisgroup.org/library/documents/africa/west_africa/b036_liberia_staying_focused.pdf, accessed October 28, 2007. International Peace Academy (IPA) (2001) ‘Refashioning the Dialogue: Regional Perspectives on the Brahimi Report on UN Peace Operations,’ International Peace Academy and Center on International Cooperation, available at www.ciaonet.org, accessed October 28, 2007. Integrated Regional Information Networks (IRIN) News (2003a) ‘DRC: Rivals Sign All-inclusive Peace Deal,’ December 17, available at www.irinnews. org/print.asp?ReportID=31415, accessed August 8, 2005. Integrated Regional Information Networks (IRIN) News (2003b) ‘Liberia: ECOMIL Says It Will Only Deploy Troops Close to Capital,’ available at www.irinnews.org/print.asp?ReportID=36609, accessed September 15, 2004. Integrated Regional Information Networks (IRIN) News (2003c) ‘LIBERIA: Main rebel group declares end of hostilities in Liberia,’ available at www.irinnews.org/Report.aspx?ReportId=46347.

References 233 Integrated Regional Information Networks (IRIN) News (2004a) ‘DRC: 2003 Chronology of Events,’ January 5, available at www.irinnews.org/ report.asp?ReportID=38718&SelectRegion=Great_Lakes&Select Country=DRC, accessed August 8, 2005. Integrated Regional Information Networks (IRIN) News (2004b) ‘Sudan: Thinktank Links Lack of Progress in Peace Process to Darfur Conflict,’ March 26, available at www.irinnews.org/report.asp?ReportID=40275&SelectRegion= East_Africa&SelectCountry=SUDAN, accessed August 8, 2005. Integrated Regional Information Networks (IRIN) News (2004c) ‘Peace Unsustainable Without Democratization – Think-tank,’ June 11, available at www.irinnews.org/report.asp?ReportID=41617, accessed August 8, 2005. Jakobsen, P.V. (1996) ‘National Interest, Humanitarianism or CNN: What Triggers UN Peace Enforcement After the Cold War?,’ Journal of Peace Research, 33: 205–215. Jenne, E.K. (2004) ‘A Bargaining Theory of Minority Demands: Explaining the Dog that Did not Bite in Yugoslavia,’ International Studies Quarterly, 48(4): 729–754. Jenne, E.K. (2007) Ethnic Bargaining: The Paradox of Minority Empowerment, Ithaca, NY: Cornell University Press. Jervis, R. (1978) ‘Cooperation Under the Security Dilemma,’ World Politics, 30(1): 167–214. Jervis, R. (1989) The Meaning of Nuclear Revolution: Statecraft and the Prospect of Armageddon, Ithaca, NY: Cornell University Press. Jeune Afrique-L’Intelligent (2003) ‘Burundi: Hutu-Tutsi power-sharing accord, peace process examined,’ November 23. Jones, B.D. (2001a) Peacemaking in Rwanda: The Dynamics of Failure, Boulder, CO: Lynne Rienner. Jones, B.D. (2001b) ‘The Challenges of Strategic Coordination: Containing Opposition and Sustaining Implementation of Peace Agreements in Civil Wars,’ New York: IPA Policy Paper, Series on Peace Implementation. Kalyvas, S.N. (2003) ‘The Ontology of “Political Violence”: Action and Identity in Civil Wars,’ Perspectives on Politics, 1(3): 475–494. Kalyvas, S.N. (2006) The Logic of Violence in Civil War, Cambridge: Cambridge University Press. Kalyvas, S.N. and Kocher, M.A. (2005) ‘Violence and Control in Civil War: An Analysis of the Hamlet Evaluation System (HES),’ Unpublished Paper. Kalyvas, S.N. and Kocher, M.A. (2007) ‘Ethnic Cleavages and Irregular War: Iraq and Vietnam,’ Politics and Society, 35(2): 183–223. Kansteiner, W.H. (2003) ‘U.S. Policy Toward Liberia,’ Testimony to House Committee on International Relations Subcommittee on Africa, Washington, DC, October 2, available at www.state.gov/p/af/rls/rm/24839.htm, accessed October 28, 2007. Kasfir, N. (2005) ‘Sudan’s Darfur: Is It Genocide?,’ Current History, 104(682) (May): 195–202. Kasmir, S. (2003) ‘ “More Basque than You!”: Class, Youth, and Identity in an Industrial Basque Town,’ Identities: Global Studies in Culture and Power, 9: 39–68. Kasrils, R. (1998) ‘Progress in Transformation,’ in J. Cillers (ed.), Continuity in Change: The South African Army in Transition, Pretoria, South Africa: Institute for Security Studies.

234 References Katz, M.N. (1996) ‘Collapsed Empires,’ in C.A. Crocker, F.O. Hampson, and P. Aall (eds), Managing Global Chaos: Sources and Responses to International Conflict, Washington, DC: United States Institute of Peace Press. Kaufmann, C.D. (1996) ‘Possible and Impossible Solutions to Ethnic Civil Wars,’ International Security, 20(4): 136–175. Keegan, J. (1979) ‘India,’ in J. Keegan, (ed.), World Armies, London: Macmillan Press. Khadiagala, G.M. (2002) ‘Implementing the Arusha Peace Agreement on Rwanda,’ in S.J. Stedman, D. Rothchild, and E. Cousens (eds), Ending Civil Wars, Boulder, CO: Lynne Rienner. Kheng, C.B. (1983) Red Star Over Malaya: Resistance and Social Conflict During and After the Japanese Occupation of Malaya, 1941–1946, Singapore: Singapore University Press. Kippenberg, J. (2005) ‘Rwanda Still in Our Human Rights Blind Spot,’ Human Rights Watch, available at hrw.org/english/docs/2004/07/25/rwanda9189.htm, accessed October 28, 2007. Korey, W. (1998) NGOs and the Universal Declaration of Human Rights, New York: Palgrave. Krain, M. (1997) ‘State-Sponsored Mass Murder: The Onset and Severity of Genocides and Politicides,’ Journal of Conflict Resolution. Krain, M. (2005) ‘International Intervention and the Severity of Genocides and Politicides,’ International Studies Quarterly, 49(3): 363–387. Krasner, S. (2001) ‘Sovereignty,’ Foreign Policy, 122 (January–February): pp. 20–22, 24, 26, 28–29. Kuperman, A.J. (2000) ‘Rwanda in Retrospect,’ Foreign Affairs, 79(1) (January/February): 94–118. Kuperman, A.J. (2001) The Limits of Humanitarian Intervention, Washington, DC: Brookings Institution Press. Kydd, A. and Walter, B.F. (2002) ‘Sabotaging the Peace: The Politics of Extremist Violence,’ International Organization, 56 (Spring): 263–296. Lacina, B. and Gleditsch, N.P. (2005) ‘Monitoring Trends in Global Combat: A New Dataset of Battle Deaths,’ European Journal of Population, 21: 145–166. Lacina, B., Gleditsch, N.P., and Russett, B. (2005) ‘The Risk of Death in Battle: A Response to Sarkees, Wyman and Singer,’ International Studies Association Annual Convention, Honolulu, HI. Lake, D.A. and Rothchild, D. (1998) The International Spread of Ethnic Conflict, Princeton, NJ: Princeton University Press. Lamb, G. (2004) ‘Transformation of the Military Shows What New SA Has Achieved,’ Cape Town, South Africa: Centre for Conflict Resolution, February 9. Lebow, R.N. (2005) ‘Deterrence: Then and Now,’ The Journal of Strategic Studies, 28(5) (October): 765–773. Lebow, R.N. and Stein, J.G. (1990) ‘Deterrence: The Elusive Dependent Variable,’ World Politics, 42 (April): 337–369. LeMarchand, R. (1996 [1994]) ‘Managing Transition Anarchies: Rwanda, Burundi, and South Africa in Comparative Perspective,’ Journal of Modern African Studies, 32(4) (December): 581–604. Levy, J.S. (1992) ‘Prospect Theory and International Relations: Theoretical Applications and Analytical Problems,’ Political Psychology, 13(2): 283–310. Levy, J.S. (2000) ‘Loss Aversion, Framing Effects, and International Conflict: Per-

References 235 spectives from Prospect Theory,’ Handbook of War Studies II. Ann Arbor, MI: University of Michigan Press. Lissak, M. (1976) Military Roles in Modernization: Civil-Military Relations in Thailand and Burma, Beverly Hills, CA: Sage Publications. Lijphart, A. (1977) Democracy in Plural Societies: A Comparative Exploration, New Haven, CT: Yale University Press. Lijphart, A. (1985) Power-Sharing in South Africa, Berkeley, CA: Institute of International Studies, University of California. Lijphart, A. (2004) ‘Constitutional Design for Divided Democracies,’ Journal of Democracy, 15(2): 96–109. Lischer, S.K. (2003) ‘Collateral Damage: Humanitarian Assistance as a Cause of Conflict,’ International Security, 28 (Summer): 79–109. Long, J.S. (1997) Regression Models for Categorical and Limited Dependent Variables, Thousand Oaks, CA: Sage. McDermott, R. (2004) ‘Prospect Theory in Political Science: Gains and Losses from the First Decade,’ Political Psychology, 25(2): 289–312. McGreal, C. and MacAskill, E. (2000) ‘Hostages Freed in Jungle Battle,’ The Guardian, September 11, available at www.guardian.co.uk/sierra/article/ 0,2763,366964,00.html, October 29, 2007. Mann, M. (2005) The Dark Side of Democracy: Explaining Ethnic Cleansing, Cambridge: Cambridge University Press. Mansson, K. (2005) ‘Use of Force and Civilian Protection: Peace Operations in the Congo,’ International Peacekeeping, 12(4): 503–519. Marshall, M.G. and Jaggers, K. (2003) Polity IV Project, available at www.cidcm. umd.edu/inscr/polity/. Marshall, M.G. and Jaggers, K. (2005) ‘Polity IV Project: Political Regime Characteristics and Transitions, 1800–2003,’ available at www.cidcm. umd.edu/polity/, accessed October 29, 2007. Marshall, M., Gurr, T.R., and Harff, B. (2002) ‘State Failure: Internal Wars and Failures of Governance,’ available at globalpolicy.gmu.edu/pitf/pitfdata.htm, accessed October 29, 2007. Martin-Brûlé, S. (2003) ‘Le rôle de la dissuasion dans les opération d’assistance humanitaire: le cas de la Somalie 1992–1993,’ Masters thesis, Université de Montréal. Massey, G., Hodson, R., and Seculiæ, D. (1999) ‘Ethnic Enclaves and Intolerance: The Case of Yugoslavia,’ Social Forces, 78: 669–691. Masters, D. (2004) ‘Support and Nonsupport for Nationalist Rebellion: A Prospect Theory Approach,’ Political Psychology. Mata López, J.M. (1993) El nacionalismo vasco radical. Discurso, organizaci\n y expresiones, Bilbao: Servicio Editorial, Universidad del Pais Vasco. Mathews, J.T. (1989) ‘Redefining Security,’ Foreign Affairs, 68(2): 162–177. Mees, L. (2001) ‘Between Votes and Bullets: The Basque Country,’ Ethnic and Racial Studies, 24. Mees, L. (2003) Nationalism, Violence and Democracy. The Basque Clash of Identities. New York: Palgrave Macmillan. Meho, L.I. and Tibbo, H.R. (2003) ‘Modeling the Information-seeking Behavior of Social Scientists: Ellis’s Study Revisited,’ Journal of the American Society for Information Science and Technology, 54(6): 570–587. Midlarsky, M. (2005) The Killing Trap: Genocides in the Twentieth Century, Cambridge: Cambridge University Press.

236 References Milanovic, B. (2005) Worlds Apart: Measuring International and Global Inequality, Princeton, NJ: Princeton University Press. Mills, G. (1999) ‘The South African National Defence Force: Between Downsizing and New Capabilities,’ Naval War College Review, Winter: 78–98. Minorities at Risk Project (2002) Codebook. Mission Report to the United Nations Security Council, Report of the Security Council Mission to Burundi on 10 and 11 February 1995. Mitchell, N.J. and McCormick, J.M. (1988) ‘Economic and Political Explanations of Human-rights Violations,’ World Politics, 40(4): 476–498. Moreno, L. (2004) ‘Divided Societies: Electoral Polarization and the Basque Country,’ in A. Guelke (ed.), Democracy and Ethnic Conflict, New York: Palgrave Macmillan. Morgan, P.C. (2005) ‘ “Taking the Long View on Deterrence’: A Round Table Review,’ Journal of Strategic Studies, 28(5) (October): 751–763. Moser, R.G. and Scheiner, E. (2004) ‘Mixed Electoral Systems and Electoral System Effects: Controlled Comparison and Cross-National Analysis,’ Electoral Studies, 23, 575–599. Mozaffar, S. (1998) ‘Electoral Systems and Conflict Management in Africa: A Twenty-eight State Comparison,’ in T.D. Sisk and A. Reynolds (eds), Elections and Conflict Management in Africa, Washington, DC: US Institute of Peace Press. Mueller, J. (2004) The Remnants of War, Ithaca, NY: Cornell University Press. Muggah, R. (2006) ‘Emerging from the Shadow of War: A Critical Perspective on DDR and Weapons Reduction in the Post-Conflict Period,’ Contemporary Security Policy, 27(1):. Murshed, S.M. (2002a) ‘Civil War, Conflict and Underdevelopment,’ Journal of Peace Research, 39(4): 387–393. Murshed, S.M. (2002b) ‘Perspectives on Two Phases of Globalization,’ in S.M. Murshed (ed.), Globalization, Marginalization and Development, London: Routledge. Murshed, S.M. (2004) ‘When Does Natural Resource Abundance Lead to a Resource Curse,’ IIED-EEP, Working Paper 04–01, available at www.iied.org, accessed October 29, 2007. Murshed, SM. and Gates, S. (2005) ‘Spatial-horizontal Inequality and the Maoist Conflict in Nepal,’ Review of Development Economics, 9(1): 121–134. Nafziger, E.W., Stewart, F., and Väyrynen, R. (eds) (2000) War, Hunger, and Displacement: Volume 1 – Analysis, Oxford: Oxford University Press. Nagengast, C. (1994) ‘Violence, Terror, and the Crisis of the State,’ Annual Review of Anthropology, 23: 109–136. National Memorial Institute for the Prevention of Terrorism (2004) ‘MIPT Terrorism Knowledge Base,’ available at www.tkb.org/Home.jsp, accessed October 29, 2007. Ndikumana, E. (2004) ‘Difficile partage du pouvoir entre Hutus et Tutsis au Burundi’ [Difficult sharing of power between the Hutus and the Tutsis in Burundi], Agence France-Presse, July 22. Nordlinger, E.A. (1972) ‘Conflict Regulation in Divided Societies,’ Occasional Papers in International Affairs, No. 29, Cambridge, MA: Harvard University. North, D. (1990) Institutions, Institutional Change, and Economic Performance, Cambridge and New York: Cambridge University Press. Núñez, X.M. (2001) ‘What is Spanish Nationalism Today? From Legitimacy Crisis

References 237 to Unfulfilled Renovation (1975–2000),’ Ethnic and Racial Studies, 24: 719–752. Nyanda, S. (2000) ‘Seeking Affordable Defence in the 21st Century,’ Paper presented at the Triennial International Seminar of the Royal United Services Institute of Australia. Official site of the Forces Nouvelles of the Cote d’Ivoire, available at www.nouvelleci.org/secret001/secretaire-general/pgsg_lesg.html, accessed October 29, 2007. Okorie, K. (2006) ‘Supreme Court’s invalidation of Electoral Act good for Nigeria’s democracy,’ February 9, available at www.usafricaonline.com/ kokorie.supcourtelect.html, accessed October 29, 2007. O’Leary, B. (2001) ‘Comparative Political Science and the British-Irish Agreement,’ in J. McGarry (ed.), Northern Ireland and the Divided World, Oxford: Oxford University Press. Olonisakin, F. (2003) ‘Liberia,’ in J. Boulden (ed.), Dealing With Conflict in Africa: The United Nations and Regional Organizations, Basingstoke: Palgrave Macmillan. Olson, M. (1965) The Logic of Collective Action, Cambridge, MA: Harvard University Press. Olson, M. (1996) ‘Big Bills Left on the Sidewalk: Why Some Nations are Rich, and Others Poor,’ Journal of Economic Perspectives, 10: 3–24. O’Neill, B. (2001) ‘Risk Aversion in International Relations Theory,’ International Studies Quarterly, 45 (December): 617–640. Oppel, R.A. (2006) ‘Sectarian Rifts Foretell Pitfalls of Iraqi Troops’ Taking Control,’ New York Times, November 12, A1. Orogun, P. (2004) ‘ “Blood Diamonds” and Africa’s Armed Conflicts in the PostCold War Era,’ World Affairs, 166 (Winter). Oudraat, C. (2000) ‘Making Economic Sanctions Work,’ Survival, 42: 105–127. Ould-Abdallah, A. (2000) Burundi on the Brink 1993–95, Washington, DC: US Institute of Peace Press. Paris, R. (2004) At War’s End: Building Peace After Civil Conflict, Cambridge: Cambridge University Press. Paye-Layleh, J. (2004) ‘Liberia’s Three Ex-warring Factions Officially Disband After Disarmament,’ available at www.boston.com/dailynews/309/world.Liberia, accessed November 4, 2005. Peceny, M., Beer, C.C., and Sanchez-Terry, S. (2002) ‘Dictatorial Peace?,’ American Political Science Review, 96(1): 15–26. Penniman, H.R. and Mujal-Léon, E.M. (eds) (1985) Spain at the Polls, 1977, 1979, 1982. A Study of National Elections, Durham, NC: Duke University Press. Persson, T. and Tabellini, G. (2003) The Economic Effects of Constitutions, Cambridge, MA: MIT Press. Petersen, R.D. (2002) Understanding Ethnic Violence. Fear, Hatred, and Resentment in Twentieth-Century Eastern Europe, New York: Cambridge University Press. Poe, S. and Tate, C.N. (1994) ‘Repression of Human-Rights to Personal Integrity in the 1980s – a Global Analysis,’ American Political Science Review, 88(4): 853–872. Posen, B. (1993) ‘The Security Dilemma and Ethnic Conflict,’ Survival, 35: 27–47. Posner, D.N. (2005) Institutions and Ethnic Politics in Africa, Cambridge: Cambridge University Press.

238 References Power, S. (2002) ‘A Problem from Hell’: America and the Age of Genocide, New York: Basic Books. Prunier, G. (1995) The Rwanda Crisis: History of a Genocide, New York: Columbia University Press. Przeworski, A. (1991) Democracy and the Market, Cambridge: Cambridge University Press. Pugh, M. and Cooper, N., with Goodhand, J. (2004) War Economies in a Regional Context: Challenges of Transformation. Boulder, CO: International Peace Academy Press. Quist-Arcton, O. (2003) ‘Still No Sign of Peacekeepers for Liberia, Annan Presses Security Council to Act,’ All Africa website, available at www.allafrica.com/ stories/200307300818.html, accessed October 29, 2007. Rabushka, A. and Shepsle, K.A. (1972) Politicians in Plural Societies: A Theory of Democratic Instability, Columbus, OH: Charles E. Merrill. Radio Netherlands (2004) ‘Peace Radio: Sierra Leone,’ available at www.radionetherlands.nl/features/media/dossiers/sierra-leone-p.html, accessed October 29, 2007. Raento, P. (1997) ‘Political Mobilisation and Place-specifity: Radical Nationalist Street Campaigning in the Spanish Basque Country,’ Space and Polity, 1: 191–204. Raghavan, S. (2006a) ‘Distrust Breaks the Bonds of a Baghdad Neighborhood. In Mixed Area, Violence Defies Peace Efforts,’ Washington Post, September 27, A1. Raghavan, S. (2006b) ‘Militias Splintering into Radicalized Cells. New Groups Appear More Ruthless in Use of Bombings and Death Squads,’ Washington Post, October 19, A1. Raghavan, S. (2006c) ‘In Iraq, Reprisals Embolden Militias,’ Washington Post, November 25, A1. Regan, P. (1996) ‘Conditions for Successful Third Party Intervention in Intrastate Conflicts,’ Journal of Conflict Resolution, 40: 336–359. Regan, P. (1998) ‘Choosing to Intervene, Outside Interventions in Internal Conflicts,’ Journal of Politics, 60: 754–779. Regan, P.M. (2000) Civil Wars and Foreign Powers: Outside Intervention and Intrastate Conflict, Ann Arbor, MI: University of Michigan Press. Regan, P. and Stam, A. (2000) ‘In the Nick of Time: Conflict Management, Mediation Timing, and the Duration of Interstate Disputes,’ International Studies Quarterly, 44: 239–260. Reilly, B. (2003) ‘Political Engineering of Parties and Party Systems,’ Paper presented at the 2003 APSA Annual Meeting, Philadelphia, August 28–31. Reilly, B. (2005) ‘Does the Choice of Electoral System Promote Democracy? The Gap Between Theory and Practice,’ in P.G. Roeder and D. Rothchild (eds), Sustainable Peace: Power and Democracy After Civil Wars, Ithaca, NY: Cornell University Press. Reilly, W.M. (2004) ‘$500M Pledged at UN for Liberia Recovery,’ The Washington Times, February 6, available at www.washtimes.com/upi-breaking/20040206–055947–2668r.htm. Reiter, Dan (2001) ‘The Bargaining Model of War,’ Paper presented at the Political Economy of Conflict Conference, Yale University, March 23–24. Report of the Secretary General to the United Nations Security Council (2001) ‘Interim Report of the Secretary-General to the Security Council on the Situation in Burundi,’ S/2001/1076, November 14.

References 239 Report of the Secretary General to the United Nations Security Council, ‘Report of the Secretary-General to the Security Council on the situation in Burundi,’ S/2003/1146. Report of the Secretary General to the United Nations Security Council, ‘First report of the Secretary-General on the United Nations Operation in Côte d’Ivoire,’ S/2004/443. Report of the Secretary General to the United Nations Security Council, ‘Report of the Secretary-General on the Sudan,’ S/2004/453. Resina, J. (2002) ‘Post-National Spain? Post-Spanish Spain?,’ Nations and Nationalism, 8: 377–396. Reyntjens, F. (2004) ‘Rwanda, Ten Years On: From Genocide to Dictatorship,’ African Affairs, 103(411): 177–210. Ricks, T. (2006) Fiasco: The American Adventure in Iraq, New York: Penguin. Robert, J. (1976) Perception and Misperception in International Politics, Princeton, NJ: Princeton University Press. Roberts, A. (1995–96) ‘From San Francisco to Sarajevo: The UN and the Use of Force,’ Survival, 37(4) (Winter): 7–28. Roeder, P.G. (1991) ‘Soviet Federalism and Ethnic Mobilization,’ World Politics, 43(2): 196–232. Roeder, P.G. (2005) ‘Power Dividing as an Alternative to Ethnic Power Sharing,’ in P.G. Roeder and D. Rothchild (eds), Sustainable Peace: Power and Democracy After Civil Wars, Ithaca, NY: Cornell University Press. Roeder, P.G. and Rothchild, D. (eds) (2005) Sustainable Peace: Power and Democracy after Civil Wars. Ithaca, NY: Cornell University Press. Rose, W. (2000) ‘Security Dilemma and Ethnic Conflict: Some New Hypotheses,’ Security Studies, 9(4): 1–51. Ross, M.L. (2004) ‘How Do Natural Resources Influence Civil War? Evidence from Thirteen Cases,’ International Organization, 58 (Winter): 35–68. Rothchild, D. and Emmanuel, N. (2004) ‘The Impact of Economic Assistance in Africa’s Peace Processes,’ Presented at the 47th Annual Meeting of the African Studies Association, New Orleans, November 11–14. Rothschild, J. (1981) Ethnopolitics: A Conceptual Framework. New York: Columbia University Press. Rotberg, R. (ed.) (2003) When States Fail: Causes and Consequences, Princeton, NJ: Princeton University Press. le Roux, L. and Boshoff, H. (2005) ‘The State of the Military,’ in J. Daniel, R. Southall, and J. Lutchman (eds), State of the Nation: South Africa 2004–5, Cape Town, South Africa: Human Sciences Research Council Press. Rummel, R.J. (1995) ‘Democracy, Power, Genocide, and Mass Murder,’ Journal of Conflict Resolution, 39(1): 3–26. Rummel, R.J. (1997) Power Kills, New Brunswick, NJ: Transaction Publishers. Rummel, R.J. (2000) Death by Government: Genocide and Mass Murder Since 1900, New Brunswick, NJ: Transaction Publishers. Rummel, R.J. (2004) Death by Government, New Brunswick, NJ: Transaction Publishers. Russett, B. and Oneal, J.R. (2001) Triangulating Peace: Democracy, Interdependence, and International Organizations, New York: W. W. Norton & Company. Rwanda Arusha Accords (1993) available at www.incore.ulst.ac.uk/services/ cds/agreements/pdf/rwan1.pdf, accessed August 8, 2005.

240 References Saideman, S.M. (1998) ‘Is Pandora’s Box Half-Empty or Half-Full?,’ in D.A. Lake and D. Rothchild (eds), Ethnic Conflict: Fear, Diffusion, Escalation, Princeton, NJ: Princeton University Press. Saideman, S.M. (2001a) ‘Taking Sides: Dyadic Analyses of Involvement in Other Countries’ Ethnic Conflicts,’ Working Paper, Department of Political Science, Texas Tech University, Lubbock, TX. Saideman, S.M. (2001b) The Ties That Divide: Ethnic Politics, Foreign Policy, and International Conflict, New York: Columbia University Press. Saideman, S.M. (2002) ‘Discrimination in International Relations: Analyzing External Support for Ethnic Groups,’ Journal of Peace Research, 39: 27–50. Saideman, S.M. and Ayres, R.W. (2000) ‘Determining the Sources of Irredentism: Logit Analyses of Minorities at Risk Data,’ Journal of Politics, 62 (November): 1126–1144. Saideman, S.M., Lanoue, D.J., Campenni, M., and Stanton, S. (2002) ‘Democratization, Political Institutions, and Ethnic Conflict: A Pooled Time-Series Analysis, 1985–1998,’ Comparative Political Studies, 35(1) (February): 103–129. Saideman, S.M., Lanoue, D.J., Cox, A., and Weeraratne, S. (2004) ‘Fixed and Endogenous Elections and Ethnic Conflict: Similar or Different Causal Relationships?,’ Annual Meeting of the International Studies Association, February 17–20, Montreal, QC. Salehyan, I. (2007) ‘Transnational Rebels: Neighboring States as Sanctuary for Rebel Groups,’ World Politics, 59(2): 217–242. Sambanis, N. (2001) ‘A Note on the Death Threshold in Coding Civil War Events,’ Conflict Processes Newsletter, June, available at http: Mailer.fsu.edu/~whmoore/ cps/newsletter/june2001/. Sambanis, N. (2002) ‘A Review of Recent Advances and Future Directions in the Literature on Civil War,’ Defense and Peace Economics, 14(3): 215–243. Sambanis, N. (2004) ‘What is a Civil War? Conceptual and Empirical Complexities of an Operational Definition,’ Journal of Conflict Resolution 48(6): 814–858. Sambanis, N. and Zinn, A. (2002) ‘The Escalation of Self-Determination Movements: From Protest to Violence,’ Presented at the Annual Meeting of the American Political Science Association. Sarailh de Ihartza, F. (Frederico Krutwig) (1963) Vasconia, Buenos Aires: Ediciones Norbait. Sarkees, M.R., Wayman, F., and Singer, J.D. (2003) ‘Inter-state, Intra-state, and Extra-state Wars: A Comprehensive Look at Their Distribution Over Time, 1816–1997,’ International Studies Quarterly, 47(1): 49–70. Schelling, T. (1980) The Strategy of Conflict, Cambridge, MA: Harvard University Press. Schaub, G. Jr. (2004) ‘Deterrence, Compellence, and Prospect Theory,’ Political Psychology, 25(3): 389–411. Sidanius, J. and Pratto, F. (1999) Social Dominance: An Intergroup Theory of Social Hierarchy and Oppression, Cambridge: Cambridge University Press. Silverstein, J. (1980) Burmese Politics: The Dilemma of National Unity, New Brunswick, NJ: Rutgers University Press. Singer, J.D. (1976) ‘The Correlates of War Project: Continuity, Diversity, and Convergence,’ in Francis Hoole and Dina Zinnes (eds), Quantitative International Politics: An Appraisal, New York: Praeger.

References 241 Sisk, T.D. (1996) Power Sharing and International Mediation in Ethnic Conflicts, Washington, DC: US Institute of Peace Press. Skaperdas, S. (2002) ‘Warlord Competition,’ Journal of Peace Research, 39(4), 435–446. Smith, R. (2000) ‘The UK in Sierra Leone: What next?,’ BBC News, Monday, September 11, available at news.bbc.co.uk/1/hi/world/africa/919942.stm. Snyder, J. (2000) From Voting to Violence. Democratization and Nationalist Conflict, New York: W.W. Norton. Snyder, J. (2001) Myths of Empire: Domestic Politics and International Ambition, Ithaca, NY: Cornell University Press. Snyder, J. and Jervis, R. (1999) ‘Civil War and the Security Dilemma,’ in J. Snyder and B. Walter (eds), Civil Wars, Insecurity, and Intervention, New York: Columbia University Press: 15–37. Southall, R. (1992) ‘The Integration of Armies and Civil–Military Relations in Post-Apartheid South Africa,’ South African Defence Review, no. 5. Spear, J. (2006) ‘From Political Economies of War to Political Economies of Peace: The Contribution of Disarmament, Demobilization and Reintegration After Wars of Predation,’ Contemporary Security Policy, 27(1): 168–189. Srivastava, V.K. and Giles, D.E.A. (1987) Seemingly Unrelated Regression Equations Models, New York: Dekker. Stanton, G. (2002) ‘Genocides, Politicides and Other Mass Murder Since 1945, with Stages in 2002,’ Genocide Watch, available at www.genocidewatch.org/ genocidetable.htm. Stedman, S.J. (1997) ‘Spoiler Problem in Peace Processes,’ International Security, 22(4): 5–53. Stedman, S.J. (2003) ‘Peace Processes and the Challenges of Violence,’ in J. Darby and R. MacGinty (eds), Contemporary Peacemaking: Conflict, Violence and Peace Processes, Hampshire: Palgrave Macmillan. Stein, J. G. (1991) ‘Deterrence and Reassurance,’ in P.E. Tetlock et al. (eds), Behavior, Society and Nuclear War, New York: Oxford University Press. Steinberg, D. and Saideman, S.M. (forthcoming) ‘Laissez Fear? Assessing the Impact of State Intervention in the Economy on Ethnic Conflict,’ International Studies Quarterly. Steinberger, P.J. (2002) ‘Hobbesian Resistance,’ American Journal of Political Science, 46(4): 856–865. Stephens, D., (2005) ‘The Lawful Use of Force by Peacekeeping Forces: The Tactical Imperative,’ International Peacekeeping, 12(2) (Summer): 157–172. Stewart, F. (2000) ‘Crisis Prevention: Tackling Horizontal Inequalities,’ Oxford Development Studies, 28(3): 245–262. Strohlein, A. and Evans, G. (2007) ‘A Responsibility to Protect: The World’s View,’ available at www.opendemocracy.net, accessed April 4, 2007. Suhrke, A. and Noble, L.G. (eds) (1977) Ethnic Conflict and International Relations, New York: Praeger Publishers. Sullivan, J. (1988) ETA and Basque Nationalism. The Fight for Euskadi, 1890–1986, London: Routledge. Tavernise, S. (2006) ‘Cycle of Revenge Fuels a Pattern of Iraqi Killings,’ New York Times, November 20, A1. Thürer, D. (1999) ‘The “failed state” and international law,’ International Review of the Red Cross, 836: 731–761.

242 References Thomas, C. (2000) Global Governance, Development and Human Security: The Challenge of Poverty and Inequality, London: Pluto Press. Tierney, D. (2005) ‘Irrelevant or Malevolent? UN Arms Embargoes in Civil Wars,’ Review of International Studies, 31: 645–664. Tilly, C. (1985) ‘War Making and State Making as Organized Crime,’ in P.B. Evans, D. Rueschemeyer, and T. Skocpol (eds) Bringing the State Back In, Cambridge: Cambridge University Press. Toft, M.D. (2003) The Geography of Ethnic Violence: Identity, Interests, and the Indivisibility of Territory, Princeton, NJ: Princeton University Press. Totten, S. and Jacobs, S. (2002) Pioneers of Genocide Studies, New Brunswick, NJ: Transaction Publishers. Tversky, A. and Kahneman, D. (1981) ‘The Framing of Decisions and the Psychology of Choice,’ Science, 211: 453–458. United Nations Association of Great Britain and Northern Ireland (2002) ‘Elections in Sierra Leone – An Essential Step Towards Lasting Peace,’ available at www.una-uk.org/UN&C/sierraleonepeace.html. United Nations Department of Peacekeeping Operations (UNDPKO), available at www.un.org/Depts/dpko/dpko/home.shtml. United Nations Department of Peacekeeping Operations (UNDPKO) UNOMSIL mission website, available at www.un.org/Depts/DPKO/Missions/unomsil/ Unomsil.htm United Nations Department of Peacekeeping Operations (UNDPKO) (1999) Human Development Report – 1999, New York: United Nations Development Programme. United Nations Department of Peacekeeping Operations (UNDPKO) (2001) United Nations Assistance Mission for Rwanda – Background Summary. Information Technology Section: Department of Public Information. Available at www.un.org/Depts/dpko/dpko/co_mission/unamirS.htm (Accessed August 8, 2005). United Nations Department of Peacekeeping Operations (UNDPKO) (2003) UNOMIL mission, available at www.un.org/Depts/dpko/dpko/co_mission/ unomil.htm, accessed February 14, 2008. United Nations Department of Peacekeeping Operations (UNDPKO) (2004) Liberia-UNIL-Mandate, available at www.un.org/Depts/dpko/missions/unmil/ mandate.html, accessed February 14, 2008. United Nations Department of Peacekeeping Operations (UNDPKO) (2006) UN Missions Summary of Military and Police, www.un.org/Depts/dpko/dpko/contributors/2006/jan06_4.pdf, accessed February 14, 2008. United Nations Department of Political Affairs (UNDPA) ‘Democratic Republic of the Congo: Background,’ available at www.un.org/Depts/dpa/prev_dip/africa/ dem_rep_ congo/fr_dem_rep_congo_background_2.htm, accessed August 8, 2005. United Nations Security Council (UNSC) (2003) ‘Report of the Secretary-General to the Security Council on Liberia,’ September 11, available at www.womenwarpeace.org/liberia/docs/sgreptosc11sep03.pdf. United States Institute of Peace (USIP) ‘Peace Agreement Digital Collection,’ available at www.usip.org/library/pa.html, accessed August 8, 2005. United States Institute of Peace (USIP) (1996) Future US Engagement in Africa, Special Report 17, Section IV, July, available at www.usip.org/pubs/specialreports/early/USAfrica3.html, accessed August 8, 2005.

References 243 Uvin, P. (2002) ‘The Development/Peacebuilding Nexus: A Typology and History of Changing Paradigms,’ Journal of Peacebuilding and Development, 1(1): 5–24. Valentino, B.A. (2004) Final Solutions: Mass Killing and Genocide in the 20th Century, Ithaca, NY: Cornell University Press. Valentino, B., Huth, P., and Balch-Lindsay, D. (2004) ‘Draining the Sea: Mass Killing and Guerrilla Warfare,’ International Organization, 58(2): 375–407. Van den Berghe, P.L. (1990) State Violence and Ethnicity, Niwot, CO: University of Colorado Press. Vanhanen, T. (2000) ‘A New Dataset for Measuring Democracy, 1810–1998,’ Journal of Peace Research, 37(2): 251–265. Walt, S. (1991) ‘The Renaissance of Security Studies,’ International Studies Quarterly, 35(2): 211–239. Walter, B.F. (1997) ‘The Critical Barrier to War Settlement,’ International Organization, 51: 335–364. Walter, B.F. (1999) ‘Designing Transitions from Civil War: Demobilization, Democratization, and Commitments to Peace,’ International Security, 24: 127–155. Walter, B.F. (2002) Committing to Peace: The Successful Settlement of Civil Wars, Princeton, NJ: Princeton University Press. Waltz, K. (1979) Theory of International Politics, New York: Random House. Walzer, M. (1974) ‘Political Action: The Problem of Dirty Hands,’ in M. Cohen, T. Nagel and T. Scanlon (eds), War and Moral Responsibility, Princeton, NJ: Princeton University Press. Watson, C. (1999) ‘Imagining ETA,’ in W.A. Douglass, C. Urza, L. White, and J. Zulaika (eds), Basque Politics and Nationalism on the Eve of the New Millennium, Reno, NV: Basque Studies Program. Watts, R. (1999) Comparing Federal Systems, Montreal and Kingston: McGillQueen’s University Press. Weber, M. (1978) Economy and Society: An Outline of Interpretive Sociology, Berkeley, CA and London: University of California Press. Weingast, B.R. (1997) ‘The Political Foundations of Democracy and the Rule of Law,’ American Political Science Review, 91(2): 245–263. Weinstein, J.M. (2007) Inside Rebellion: The Politics of Insurgent Violence, Cambridge: Cambridge University Press. Weiss, T.G. (1994) ‘Whither the United Nations?,’ The Washington Quarterly, 1 (Winter): 109–128. Werth, N. (2003) ‘The Mechanism of a Mass Crime: The Great Terror in the Soviet Union, 1937–1938,’ in R. Gellately and B. Kiernan (eds), The Specter of Genocide: Mass Murder in Historical Perspective, Cambridge: Cambridge University Press. West Jr, F.J. (1985) The Village, Madison, WI: The University of Wisconsin Press. Wilkinson, S.I. (2004) Votes and Violence: Electoral Competition and Ethnic Riots in India, Cambridge: Cambridge University Press. Williams, R. (1993) ‘Military Control Over the Armed Forces During the Transition,’ South African Defence Review, No. 9. Williams, R. (2002) ‘Integration or Absorption? The Creation of the South African National Defence Force, 1993 to 1999,’ African Security Review, 11(2). Wolters, S. (2004) ‘Update on the DRC: Is the Transition in Trouble?,’ Reliefweb. Institute for Security Studies, July 20, available at www.reliefweb.int/w/rwb.nsf/ 0/534fe43f692d3cbcc1256ee400201ce7?OpenDocument, accessed August 8, 2005.

244 References Wood, E.J. (2000) Forging Democracy from Below: Insurgent Transitions in South Africa and El Salvador, Cambridge: Cambridge University Press. Wood, E.J. (2003) ‘Modelling Robust Settlements to Civil War: Indivisible Stakes and Distributional Compromises,’ available at www.prio.no, accessed October 29, 2007. Woodward, S.L. (2005) ‘Fragile States: Exploring the Concept,’ Paper presented to the ‘States and Security’ Learning Group at the Peace and Social Justice meeting of the Ford Foundation, Rio de Janeiro, Brazil, November 29, 2004. World Bank (2002) World Development Indicators, Washington, DC: World Bank. Young, W.L. (1982) Minorities and the Military, CT: Greenwood Press. Zagare, F.C. (2004) ‘Reconciling Rationality With Deterence: A Re-examination of the Logical Foundation of Deterrence Theory,’ Journal of Theoretical Politics, 16(2): 107–141. Zagare, F.C. and Kilgour, M.D. (2000) Perfect Deterrence, Cambridge: Cambridge University Press. Zahar, M.-J. (2005a) ‘Des dangers de l’imposition étrangère lors du nation-building: la question nationale au Liban et en Irak,’ Critique Internationale, 28(3): 153–168. Zahar, M.-J. (2005b) ‘Political Violence in Peace Processes: Voice, Exit and Loyalty in the Post-Accord Period,’ in John Darby (ed.), Violence and Reconstruction, Southbend: Notre Dame University Press. Zahar, M.-J. (2005c) ‘Power Sharing in Lebanon: Foreign Protectors, Domestic Peace, and Democratic Failure,’ in D. Rothchild and P. Roeder (eds), Sustainable Peace: Power and Democracy after Civil Wars, Ithaca, NY: Cornell University Press: 219–240. Zahar, M.-J. (2006) ‘Understanding the Violence of Insiders: Loyalty, Custodians of Peace, and the Sustainability of Conflict Settlement,’ in Edward Newman and Oliver Richmond (eds), Spoilers and Peace Processes: Conflict Settlement and Devious Objectives, Tokyo: United Nations University Press. Zakaria, F. (1997) ‘The Rise of Illiberal Democracy,’ Foreign Affairs, 76(6): 22–43. Zartman, I.W. (1989) Ripe for Resolution, New York: Oxford University Press. Zartman, I.W. (ed.) (1995) Collapsed States: The Disintegration and Restoration of Legitimate Authority, Boulder, CO: Lynne Rienner. Zellner, A. (1962) ‘An Efficient Method of Estimating Seemingly Unrelated Regressions and Tests for Aggregation Bias,’ Journal of the American Statistical Association, 57: 348–336. Zulaika, J. (1988) Basque Violence; Metaphor and Sacrament, Reno, NV: University of Nevada Press.

Index

Abidjan Accords 190 accountability groups 78, 81 Accra Comprehensive Peace Agreement (2003) 147 Acemoglu, D. 77 action–repression–action strategy 123 Addison, T. 54, 55, 67 Adebajo, A. 189, 190, 191, 194 Adidbe, C. 190, 191, 193 adjusted winner mechanism 65 Afghanistan 91, 206, 213, 218, 219–20 Africa Confidential 143, 153 Africa Research Bulletin 142–3 Africa, power sharing institutions: limits of reassurance 150–5; overview 138–9, 211–12; patterns of experience 142–50; short- and long-term implications 139–41; see also South Africa Africa: credible commitment 68; external support 174–80 African National Congress (ANC) 112–13, 114 Afrikaners 112–13 Ajuria-Enea Pact (1988), Spain 124 Akbaba, Yasemin 17, 161–80, 209, 214–15 Akosombo Agreement 194 Albania 15 Algeria 91 Anderson, B. 110 Anderson, M. 202 Angola 91 anocracies 9 apartheid regime, South Africa 111–16 Arendt, H. 40 Argentina 91 Argolid, Greece 24–8 Armed Forced of Liberia (AFL) 190–1 arms embargoes 185 Armstrong II, David A. 33–52, 206, 207–8, 218 Arusha Accords (1993) 38, 97, 111, 144–5, 147–8 Arusha Process and Reconciliation Agreement (2000) 146

asset inequality 62 assurance and the military 210–11 assuring as balancing act 9–15 autocracies 9, 35; and genocide 35, 36, 80–1, 90–1 Autonomous Communities, Spain 120–5, 127, 129, 130, 132–6 Aydin, Aysegul 12, 16, 69, 72–93, 156, 206–7, 208, 211 Ayres, R.W. 163 Azam, J.-P. 67 Azanian People’s Liberation Army (APLA) 113 Azarya, Victor 142 Aznar, J.M. 130, 131 Baldwin, R. 57 Banks’ Cross-National Time-Series (CNTS) Data Archive 42 bargaining 184 Basque Country–Spain Bilateral Commission 133 Basque nationalism: framing violence 122–8; plan Ibarretxe 132–4; and political institutions/nationalism 128–32; overview 16–17, 120–2, 213–14 Batasuna 120, 127, 131 BBC News 140, 143, 144 Bélanger, L. 175, 176 Belgium 104, 106–7 Berman, E.G. 191, 192 Besançon, M. 61, 62 Bett, R.K. 183 bicameral legislature 157 Binh Nghia, Vietnam 29–30 Boshoff, H. 115 Botha, P.W. 112 Bougarel, X. 22 Boyd, R. 72 Brams, S.J. 64, 65 Breslauer, G.W. 40 British colonial policy 105–6, 108 Brown, M.E. 165, 167

246 Index Brunné, J. 183 Bryant, Gyude 143 Bueno de Mesquita, B. 74, 85 Burma 91, 104, 107–8 Burton, M. 79 Burundi 38, 91, 104–6, 142–3, 146–7 Canada 104, 106–7, 135–6 Carment, D. 161, 168–9, 185, 214 Catalan Statute of Autonomy 134, 136 Cederman, L.-E. 84, 92 Chang, J. 72 China 58–9, 60, 91 Choi, G. 169 Cingranelli, D.L. 36, 42 civil conflict theories 4–7 civil war 22–31; in failed states 189–90; fatalities 36–43 civilian control of military 12 Clark, R.P. 122 Coalition pour la Défense de la République (CDR), Rwanda 145 Cochran, E. 116 coercive force 185–6 coercive power 39–40 Collier, P. 1, 6, 54, 55, 67, 84, 86, 92, 142, 154, 185, 215–16 colonialism 57, 62–3, 104-5, 108, 110 commitment see credible commitment communal polarization 151 Comprehensive Peace Agreement (2004), Sudan 144 conflict-ridden states 216–17 Connaughton, R. 197 Conneh, Damateh 153 Constitution: Nigeria 159; Spain 120–1, 129 constructive anger 165–6 contested territory 23–31 Convention of Government (1994), Burundi 146 convergence 56–7 Conversi, D. 126 Cooper, N. 185 Copts 165 Correlates of War project 43 Coser, L. 5 Côte d’Ivoire 142, 143–4, 148 Cotonou Agreement 191, 193–4 credible commitment: failed states 183–4; lack of 151–2; negotiated settlements 66–9; problem of 11 Cunningham, D.E. 213 Czechoslovakia 104, 106–7 Dallin, A. 40 Daragahi, B. 30 Darfur 215 Davenport, Christian 16, 33–52, 72, 76, 206, 207–8, 218

David, S.R. 2, 8 De Figueiredo, R.J.P. Jr. 5, 11, 20 De Jouvenal, B. 36 De Klerk, F.W. 113, 115–16 deaths 34–6; generation of 39–41; genocide/ politicide 33–6; overview 33–4, 207 decisionmaking, exclusion from 153–4 democracies 9, 75–6, 79 Democratic Republic of the Congo (DRC) 91, 140, 142, 148, 154 democratic transition, Spain 123–4 demographics 7–8; South Africa 111–16 Denselow, R. 197 destructive anger 165–6 deterrence: as balancing act 9-15; implications of 14–15; role of 182–202; theory 3, 10–11 Diamond, L. 81 Diarra, Seydou 143 discount rates, commitment 67–8 discrimination 61–2, 163–6 dissident capability 39–43 “divide-and-rule” policy 105–6, 110 divisible goods 63–6 domestic state threats 35–6 dominant parties, power sharing 139–41, 142, 152–3 Dougherty, C. 154 Douglass, W. 127 Downes, A.B. 179 Doyle, M. 195 Drake, P.W. 77 Drysdale, J. 183 Dukuly, A. 200 Durch, W.J. 189 Duvall, R. 40 Dwyer, L. 72 Eastern Europe 59 economic discrimination 163–5, 172–80 economic integration 56–7 economic issues, power sharing 154–5 economic mismanagement 62–3 economic openness 90, 92 economic policies 55–6 economic sanctions 184–5 ECOWAS 191, 196; Monitoring Group in Liberia (ECOMIL) 196, 199 Electoral Act, Nigeria 158 electoral competition 75 electoral system, integrative 158–9 elite politics 4–5 elite power sharing regimes 156 elites, South Africa 111–16 Emmanuel, N. 154–5 empirics 22–31; fit with theory 206 envy-free allocation 64–6 ETA (Euzkadi ta Askatasuna): framing violence 122–8; plan Ibarretxe 132–4; and

Index 247 political institutions/nationalism 128–32; overview 16–17, 120–2, 213–14 Ethiopia 91, 105, 107–9 ethnic affinity 168–70 ethnic composition, military 101–10 ethnic kin 173, 174–80 ethnic polarization 31, 81–2, 92 ethnic status reversal 98–101 ethnic structure, military 96–117 ethno-linguistic fractionalization (ELF) indices 85, 92–3 Euskal Herria 125 Euskal Herritarok (EH) 124, 126 Euskera 129 Eusko Alkartasuna (EA) 124, 125 Euskobarómetro 128 Evans, Gareth 215 executive constraints 77–9, 83–5, 88–9, 211–12 executive recruitment 79–80, 83–4, 89–90 exit strategies 8–9 external protection, appropriateness of 152–6 external support: and discrimination/ rebellion 163–6; and groups 214–15; hypotheses 167–70; overview 161–3; study 171–9 failed states 8; civil war within 189–90; credible commitment in 183–4 Failed States Index 182 fear and the military 98–101 Fearon, James D. 6, 10, 34, 39, 40, 43, 54, 64, 67, 151, 216, 217 feasible compromise equilibria 66 federalism, non-ethnic 159–60 Fellman, M. 22 Fiji 104–6 Flemings, Belgium 106–7 force, constraints on use 209–12 Fox, J. 165, 166, 169 Franco period, Spain 122–5 Freedman, L. 187, 188 Freud, S. 52 Gagnon, V.P. Jr. 5 Gates, Scott 12, 16, 61, 62, 69, 70, 72–93, 206–7, 208, 211 Gbagbo, Laurent 143–4 GDP 43–9, 107, 109, 110 Geddes, B. 84, 85, 88, 91 Gelfand, L.201 Gellner, Ernest 110 General Framework for Peace (Dayton Peace Agreement) 211 genocide: deaths from 33–52; discussion 92–3; empirical analysis 83–92; institutional configuration/political survival strategy 77–82; strategic versus structural explanations 75–7 Genocide Watch 42

George, A. 186 Germany, occupation of Greece 24 Ghana 104, 107–8 Gibney, M. 42 Gillespie, R. 125 Girardin, L. 84, 92 Gleditsch, K.S. 37, 173 Gleditsch, N.P. 36, 41–2, 47 global inequalities 56–61 GNP 104, 105–10 Gonzáles Catarain, Dolores 127 goods, divisible 63–6 government type 9 Grandin, G. 72 Granja Sainz de la, J.L. 124 great power intervention 192–202 greed 6, 54–5 Greek Civil War 23–8, 31, 206 grievance 54–5, 165–6 Grossman, H. 7, 61–2 group discrimination 164–6 groups and external support 214–15 guerrilla war 42–9 Gunther, R. 81 Gurr, T.R. 161, 163, 164, 165, 166, 167, 169, 170, 171, 173 Guyana 104–6 Habyarimana, Juvénal 144–5 Hafner-Burton, E. 36 Halliday, J. 72 Hamilton, C. 126, 156 “Hamlet Evaluation System” 24, 28–9 Hardin, R. 20 Harff, B. 36, 42, 75, 83, 85, 91, 92–3, 164, 166 Harman, D. 191, 197, 200 Hartzell, Caroline 141, 145, 152, 211, 212 Harvey, F. 185, 186 Hathaway, O. 36 Hegre, H. 54 Herbst, J. 209 Herri Batasuna (HB) 124, 129 Herz, J.H. 2, 7 Heston, A. 85 Higgs, J. 115 Hill, S. 186 Hirsh, J.L. 183, 185, 189, 191, 192 Hobbes, T. 72–3, 206–7 Hoddie, Matthew 145, 211, 212 Hoeffler, A. 1, 84 horizontal inequalities 55–6, 61–3 Horowitz, D.L. 5, 159 Howe, H. 113, 116 Human Rights Watch 33, 38, 199 human security analysis: dependent variable/estimator choice 83; discussion 92–3; independent variables 83–5; results 85–92

248 Index Human Security Report 54 human security, interstate/non-state threats 35 Huth, P.K. 185, 186 Hutu 38, 105, 111–12, 143, 144 Ibarretxe plan 132–4 inclusive decisionmaking 138 income per-capita 43–9, 56–61, 69–70 incumbent control 23–31 India 58–9, 60, 105, 107–9 indiscriminate targeting 12 indiscriminate violence 24–5, 27–8 indivisibilities 55–6, 63–70 Indonesia 74, 91, 105, 107–9 inequalities 55–63, 70 information: in peace operations 187; reliability of 151 insecurity, mechanisms of 208–9 institutional characteristics 76–7 institutional configuration 77–82 institutional solutions 111–16 institutions of commitment 12, 69 instrumental violence 30–1; ETA 121, 122–5 insurgency 6–7 insurgent control 23–31 Integrated Regional Information Networks (IRIN) 140, 144, 154, 196 Inter-Congolese Dialogue (2002) 154 intermixing 23–31 International Crisis Group (ICG) 154, 196, 201 international environment 215–16 International Monetary Fund (IMF) 62 international peacekeeping see peace operations International Relations 2, 3, 7–8, 163 interstate threats 35; fatalities from 36–43 Iraq 30, 66, 91, 104, 107–8, 206, 207, 213, 214–15, 219–20 Israel 104–6, 165 Ivorian Popular Front (FPI) 143 Izquierda Unida (IU), Spain 124 Jacobin vision, Spain 121, 130 Jacobs, S. 36, 45 Jaggers, K. 43, 83 James, Patrick 17, 161–80, 209, 214–15 Jenne, E.K. 215 Jervis, R. 2, 7, 14 Jones, B.D. 145 Jones, Bruce 199 Jordan 105, 107–9 judiciary, independence of 157–8 Kabila, Joseph 140 Kalyvas, S.N. 7, 12, 14, 16, 20–32, 74, 207–8 KAS alternative 123–4

Kasfir, N. 144 Kasmir, S. 126 Kenya 105, 107–9 Khadiagala, G.M. 145 Kheng, C.B. 22 Kilgour, M.D. 187 Kim, M. 7 Kippenberg, J. 38 Kocher, M.A. 31 Korey, W. 36 Krain, M. 2, 11 Krasner, S. 182 Krutwig, Frederico 122–3 Kuperman, Alan 112, 145 KwaZulu Police (KZP) 113 Lacina, B. 34, 36, 41–2, 47, 50 Laitin, D. 6, 34, 39, 40, 43, 216 Lake, D.A. 151 Lamb, G. 114, 115 language, Basque country 126, 129, 130 Lebanon 105, 107–9, 153 Lebow, R.N. 185, 186, 187 Lecours, André 11, 13–14, 15, 16–17, 120–36, 213 legislature, bicameral 157 LeMarchand, R. 111–12, 113, 115 lethality: comparative assessment 36–43; conclusions 49–52; on death and dying 34–6; findings 43–9; overview 33–4 Leviathan 72–3 Liberia 142, 143, 147, 153–4, 183, 184, 188–90; ECOMIL peace operations 196; ECOMOG peace operations 190–6, 197, 199–200; ECOMOG/UNOMIL peace operations 193–4; overview of civil war 189–90; UNMIL peace operations 200–1 Liberians United for Reconciliation and Democracy (LURD) 143, 153 Liberty 157 Lichbach, M.I. 49 Lijphart, A. 141, 150 Linas–Marcoussis Agreement (2003) 149 Lincoln, Abraham 22 Lischer, S.K. 202 Lizarra Pact (1998), Spain, 124–5 Lomé Peace Accord 197 Long, J.S. 43 lootable resources 6 McCormick, J.M. 36 McCubbins, M.D. 77 Machakos Protocol (2002) 149 majoritarian electoral process 158–9 Malaysia 22, 105, 107–9 Malik, S.P. 186 Mann, M. 112 Mansson, K. 183 Marshall, M. 43, 83

Index 249 Martin, P. 57 Martin-Brûlé, Sarah-Myriam 17, 182–202, 217 mass murderers: institutional configuration and political survival 77–82; overview 72–5; strategic versus structural explanations 75–7 Massey, G. 22 Mata López, J.M. 127 Mees, L. 123, 124, 125 Meho, L.I. 165 Midlarsky, M. 2, 72, 74 Milanovic, B. 58, 59–60 military: and assurance 210–11; civilian control over 12; South Africa 111–16 military capability 40–1, 43–9 military regimes, and genocide 80–1, 90 military structure: broader context of 101–10; cluster analysis 104–10; and fear and resentment 98–101; institutional solutions 111–16; overview 96–8 militias, disbanding of 153–4 Milosevic, Slobodan 5, 73 Minorities at Risk Project (MAR) 162–6, 169–74, 177–8 Mitchell, M.J. 36 Monitoring Group of the Economic Community of West African States (ECOMOG) 188–9, 190–3, 195–6, 197, 198, 199, 200 Moreno, L. 129 Morgan, P.C. 187 Moser, R.G. 157 Mouvement Pour la Liberation du Congo (MLC) 140, 154 Mouvement Révolutionnaire National pour le Développement, Rwanda (MRND) 144–5 Movement for Democracy and Elections in Liberia 153 Mozaffar, S. 158 Mueller, J. 22 Muggah, R. 211 Mujal-Léon, E.M. 129 multi-ethnicity, military 96–117 multiple-agent political systems 78 Murshed, S. Mansoob 16, 54–71, 141, 209, 211, 213, 215–16 Mutual Assured Destruction (MAD) 28 mutual deterrence 28 Nafziger, E.W. 63 Nairobi Declaration (2004) 144 Naivasha Protocol (2004) 149 nation states, inequalities within 61–3 National Assembly, South Africa 157 National Congress Party (NCP), Sudan 144 National Council of Provinces, South Africa 157

National Forces for Liberation (FNL), Burundi 142 National Liberation Movement, Spain 124, 127 National Memorial Institute for the Prevention of Terrorism 42 National Party, South Africa 113 National Patriotic Front for Liberia (NPFL) 190–1 National Union for Progress (UPRONA) Burundi 143 nationalism, Spain 128–34 nationalist elites 4–5 Ndikumana, E. 143 negotiated settlements, commitment problems 66–9 negotiations 65–6 New Forces, Côte d’Ivoire 144 Nigeria 105, 107–9, 158, 159, 192–3; peace operations in Liberia/Sierra Leone 190–1 non-coercive deterrence 187–8 non-ethnic federalism 159–60 non-state threats 35 North, D. 74 Nyanda, S. 115 O’Leary, B. 150 Oakley, R.B. 183 OECD 59, 60 Okorie, K. 158 Olonisakin, F. 194 Olson, M. 54, 61, 69 Oneal, J.R. 169 “Operation No Living Thing” 192 Operation Restore Hope 186 Oppel, R.A. 30 opponent type 10 Organization of African Unity (OAU) 191 organizational fragmentation 30–1 Orogun, P. 185 Otegi, Arnaldo 120 Oudraat, C. 185 Ould-Abdallah, A. 139 overt discrimination 166 Pakistan 91, 105, 107–9 Paris, R. 142 Partido nacionalista vasco (PNV), Spain 122, 123, 124, 125, 129, 131, 133, 134 Partido Popular (PP), Spain 121, 124, 130, 131, 132, 134, 135 Partido Socialista Obrero EspaZol (PSOE), Spain 121, 124, 132–5 peace, value of 67–8 peace agreements 63–6 peace operations: case study 188–90; failure of 193–5; overview 182–3, 217–18; partially successful 190–3, 195–8; successful 198–201; theoretical discussion 183–8

250 Index Peceny, M. 80, 81 Penniman, H.R. 129 personalist regimes 80–1 Persson, R. 74 Petersen, Roger D. 16, 22, 30, 96–117, 208, 210 Poe, S. 36 policy implications 218–20 political constraints 211–12 political discrimination 163–5, 172, 174–80 political institutions, Spain 128–32 political instrument, violence as 122–5 political interests 68 political openness 90, 92 political order: actions against 35; defense of 35–6 political participation 79–80, 83–4, 89 political power limits analysis 83–92 political re-alignment 124 political survival 77–82 Political Terror Scale (PTS) 42 politicide: deaths from 33–52; discussion 92–3; empirical analysis 83–92; institutional configuration/political survival strategy 77–82; strategic versus structural explanations 75–7 politics, violence as see violence as politics Polity 98, 174 Polity IV 43, 83–4, 88 population 43–9 Posen, B. 2, 7, 20, 22, 162, 163 power retention South Africa 111–16 power sharing institutions, Africa: limits of reassurance 150–5; overview 138–9, 211–12; patterns of experience 142–50; short- and long-term implications of 139–41 powers, separation of 156–7 preemptive violence 28 presidential elections 159 Pretoria Agreement (2002) 143, 148 Pretoria Protocol (2003) 146–7 prisoners’ dilemma 14 Pronk, Jan 144 proportional representation (PR) 158–9 proportionality, military 101–10 Przeworski, A. 150 public spending, discrimination in 61–2 public support, ETA 128–32 Pugh, M. 185, 190 purchasing power parity (PPP) 58 Quist-Arcton, O. 196 Rabushka, A. 82 racial differences 169–70, 173–80 radicalization, Basques 126–7 Radio Netherlands 197 Raento, P. 126–7 Raghavan, S. 22, 30, 31

Rassemblement Congolais Pour la Démocratie (RCD–Goma) 140, 154 reassurance, limits of 150–5 rebellion: and discrimination/rebellion 163–6; hypotheses 167–70; overview 161–3; study 171–9 recession 62–3 recruitment policy, military 101–10 Regan, P. 161, 165, 168, 175 regime type: and external support 174–80; and genocide 43–9, 84, 90–2 Reilly, B. 82, 158–9, 200 Reiter, D. 184 relational bargain, security as 212–14 religious differences 169–70, 173–80 Reno, Will 182 repression 42–9 research directions 218–20 resentment and the military 98–101 Resina, J. 130 revenge 30–1 revolution 42–9 Reyntjens, F. 38 Richards, D.L. 36, 42 Ricks, T. 12 Roberts, A. 183 Robinson, J.A. 77 Roeder, P.G. 138, 139, 156–7 Romania 15 Root, H. 74 Rose, W. 2 Ross, M.L. 6 Rotberg, R. 218 Rothchild, Donald 5, 11, 17, 67, 69, 138–60, 211–12, 215–16 Rowlands, D. 168 Ruberwa, Azarias 140 rulers as mass murderers: institutional configuration and political survival 77–82; overview 72–5; strategic versus structural explanations 75–7 Rummel, R.J. 2, 11, 36, 42, 72, 76 Russet, B. 169 Rwanda 8, 38–9, 74, 91, 97–8, 104–6, 111–12, 116–17, 142, 144–5, 147–8, 153 Rwandan Patriotic Army (RPA) 111–12 Rwandan Patriotic Front (RPF) 38, 111–12, 144–5 Saideman, Stephen M. 1–17, 20, 51, 73, 96, 140, 159, 162, 163, 164, 166, 168, 171, 173, 174, 175, 176, 177, 205–221 Salehyan, I. 214 Sambanis N. 39 Sams, K.E. 191, 192 sanctions 184–5 Santikarma, D. 72 Sarkees, M.R. 33 Schaub, G. Jr. 185

Index 251 Scheiner, E. 157 Schelling, T. 185 security: as balancing act 216–18; in civil conflict theories 4–7; connection with security 206–7; as relational bargain 212–16 security dilemmas 7–9; in action 163–6; challenging 205–8; theory of 21–2 Seemingly Unrelated Regression (SUR) 174–9 selective violence 24, 26–9 separatism 173 separatist minority groups 168–70, 174–80 shared control 23–31 shared norms 154 Shepsle, K.A. 82 Shi’a 30, 213, 214–15 Sierra Leone 183, 185, 188–90; ECOMOG peace operations 191–3; ECOMOG/UNOMSIL peace operations 194–5; overview of civil war 189–90; UNAMSIL peace operations 196–200 Singapore 105, 107–9 Singer, J.D. 35 single-party regimes 80–1, 90 Sinhalese 105–6, 107 Sisk, T.D. 141 Skaperdas, S. 67 Smoke, R. 186 Snyder, J. 2, 5, 9, 82, 89 social contract 61, 62–3, 70–1, 72–3 societal cleavages 60, 74–5, 82 Sogge, David 182 Somalia 91, 186 South Africa 97–8, 104–6, 111–17, 157, 159–60 South African Defence Force (SADF) 112–16, 210 South African National Defence Force (SANDF) 114–15 Spanish nationalism 128–32 Spear, J. 211 “spoiling”, South Africa 113 Sri Lanka 91, 104–6, 169 Stam, A. 38 Staniland, Paul 96–117, 208, 210 Stanton, G. 36, 42 state capacity 39–52; decline of 70–1 state capture 208–9 state–military relationships 96–101 state-sponsored violence: comparative assessment of lethality 36–43; conclusions 49–52; on death and dying 34–6; findings 43–9; overview 33–4 states: connection with security 206–7; and international environment 215–16 status, South Africa 97–8, 111–16 status quo 11, 15, 35 status reversals: ethnic groups 98–101; South Africa 111–16

Stedman, S.J. 212 Stein, J.G. 185, 186, 187 Steinberg, D. 177 Steinberger, P.J. 73 Stephens, D. 183 Stewart, F. 61 Stohl, M. 40 strategic explanations, mass murder 75–7 Strohlein, Andrew 215 structural explanations, mass murder 75–7 Sudan 91, 104–6, 142, 144, 149, 154, 215 Sudan People’s Liberation Movement (SPLM) 144 Sullivan, J. 123 Sunnis 30, 213 support, intensity of 171 Surinam 104–6 Switzerland 104, 106–7 symbolic violence, ETA 121, 126–8 Syria 91, 105, 107–9 Ta’if agreement 153 Tabellini, G. 74 Tamils 105–6, 169 Tate, C.N. 36 Tavernise, S. 30, 31 taxation, discrimination in 61–2 Taydas, Zeynep 17, 161–80, 209, 214–15 Taylor, Charles 189–90, 191, 194 terrorist incidents 42–9 Thailand 104, 107–8 theoretical discussion, peace operations 183–8 Tibbo, H.R. 165 Tierney, D. 185 time horizons, commitment to peace 68–9 Toft, M.D. 10 Toope, S.J. 183 Totten, S. 36, 45 transitional arrangements, power sharing 140 Tutsi 38, 105, 111–12, 143, 144–5 Uganda 91, 104, 107–8 UK, and UNAMSIL 198–200 unconsolidated democracies 79 Union pour la démocratie et le progress social, Liberia 154 United Nations (UN): Assistance Mission in Sierra Leone (UNAMSIL) 188–9, 196–7, 198–201, 202; Department of Peacekeeping Operations (UNDPKO) 57, 153, 189, 195, 196, 200; Mission in Liberia (UNMIL) 188–9, 198–201, 202; Observatory Mission in Liberia (UNOMIL) 188, 193–5; Observatory Mission in Sierra Leone (UNOMSIL) 188, 194–5; Operation in Somalia (UNOSOM) 186

252 Index United States Institute of Peace 154 US: Civil War 22; and UNMIL 200–1 Uvin, P. 216 Valentino, B.A. 12, 36, 75 Van den Burghe, P.L. 36 Vanhanen, T. 83–4 Vietnam War 23, 28–30, 31, 206 violence: dynamics of 22–31; level of 167–70, 171, 174–5; as symbol 126–8 violence as politics: ETA 122–8; ETA, nationalism and political institutions 128–32; overview 120–2; plan Ibarretxe 132–4 Walloons, Belgium 106–7 Walter, B.P. 8, 63, 140, 141, 180, 183, 184, 217 Waltz, K. 2, 7 Watson, C. 127 weak parties, power sharing 139–41, 142, 150, 152–3, 155–6 weak states 209

Weber, M. 182 Weingast, B.R. 5, 6, 11, 20, 79 Weiss, T.G. 183 West, F.J. Jr. 29–30 Westminster Parliament model 156–7 white minority status, South Africa 115 Wilkinson, S.I. 10, 15 Wood, E.J. 63, 64, 66, 209 Woodward, S.L. 182 World Bank 1, 6, 57, 62 Yaya, Thomas Ninely 153 Yugoslavia 5, 8, 22, 31, 91, 97–8, 105, 107–9 Zagare, F.C. 187 Zahar, Marie-Joëlle 1–17, 51, 73, 96, 140, 153, 162, 205–20, 219 Zakaria, F. 73, 79, 90 Zartman, I.W. 10, 217 Zinn, A. 39 Zulaika, J. 127